Datasets:

sulovexin

/

mysummary

like 0

journal.pcbi.1005117

Estimating Copy Number and Allelic Variation at the Immunoglobulin Heavy Chain Locus Using Short Reads

The variation between human genomes in gene copy number is understudied and poorly characterized ., One such region where this variation is known to exist is the immunoglobulin heavy variable ( IGHV ) locus ., It is a vital component of the adaptive immune system , containing the V genes that code for a component of the heavy chain of antibody molecules ., Like other multigene receptor families , the gene segments in this region have been accumulated over time through a process of gene duplication and diversification 1–3 ., As such , many of the genes in this locus are highly similar and there are repetitive DNA elements interspersed throughout the region ., IGHV haplotypes ( instances of the IGHV locus ) vary not only by single nucleotide polymorphisms , but also in the copy number and ordering of gene segments 4–12 ., All these characteristics make it difficult to study this region and , to date , only two reference sequences of the full IGHV locus exist 12 , 13 ., The human IGHV locus lies at the telomeric end of chromosome 14 and is approximately 1 Mb in length ., In this 1 Mb region , there are about 40 functional genes , each approximately 300 bp in length ., There are also approximately 80 non-functional pseudogenes in the region , so-called because they are either truncated or contain premature stop codons ., Known allelic variants of individual IGHV genes are currently curated in the International Immunogenetics Information System ( IMGT ) Repertoire database 14 ., Throughout this article , we suppress the standard prefix “IGHV” in gene names for ease of reading , e . g . , we use 6-1 instead of IGHV6-1 ., The nomenclature for IGHV genes is further detailed in Materials and Methods ., Given the role of the IGHV locus in the adaptive immune response , IGHV genotypes are obvious candidates as genetic determinants for susceptibility to infectious disease ., Several early targeted studies of the IGHV locus have implicated allelic variation and copy number in determining expressed antibodies repertoires and understanding disease susceptibility 5 , 10 , 11 , 15–18 ., Allele 3-23*03 , for example , has been shown to be more effective in binding Haemophilus influenzae type ( Hib ) polysaccharide than the most common allele , 3-23*01 19 ., Despite such findings , however , the IGHV locus is rarely included in genome-wide association studies , due in large part to the lack of standard format and tools to quantitatively characterize variation in the region ., Lack of tools for genotyping the IGHV locus also hampers the burgeoning field of antibody repertoire sequencing 20–24 , which is being used in numerous medical applications , including inferring the evolutionary path of broad and potent monoclonal antibodies against human immunodeficiency virus ( HIV ) 25–27 , detecting blood cancers 28 , 29 , assessing the impact of aging on the antibody response 30 , and measuring the adaptive immune response to vaccination 31 , 32 ., The first step in many of these studies is to align each read , sequenced from the antibody repertoire of an individual , to its germline gene ., The current practice is to use germline alleles in a public database of all known alleles ( such as the IMGT Repertoire database ) for alignment ., Aligning to all germline alleles is a severe limitation of the process because after undergoing somatic hypermutation , antibody sequences may be so different from the germline that the top-matching allele in the database no longer corresponds to the germline allele in the individual ., The increasing availability of whole-genome sequencing ( WGS ) provides a new opportunity to investigate genetic variation in the IGHV locus ., Specifically , the large sample sizes of these WGS datasets and the high-throughput manner in which the data can be analyzed could provide valuable information ., This approach can complement and guide genotyping efforts based on locus-specific assays 9–12 ., Yet there are currently no methods to interrogate the IGHV locus using WGS short reads ( though a tool that extracts genotypes from long contigs exists 33 ) ., Here , we address this pressing need for methods that quantitatively characterize the IGHV locus from WGS short-read data ., By leveraging the IMGT database of known alleles , we construct a pipeline that gives a systematic description of the IGHV locus from short-read data ., This description is in terms of a set of operationally defined gene clusters , so called because each cluster comprises IGHV alleles that are operationally indistinguishable ., Our method does not attempt to reconstruct the organization of the locus or sequence the intergenic regions , both of which are important and challenging tasks ., It does , however , allow the quantification of coarser measures of genetic variation ., With reads as short as 70 bp and with coverage of 30× , our pipeline accurately detects the presence of gene clusters from simulated reads of the two known IGHV reference sequences ( GRCh37 and GRCh38 ) ., With sufficiently long read lengths ( 250 bp ) , the pipeline also outputs accurate nucleotide sequences of gene segments present in single copy ., We then run the pipeline on an empirical dataset of whole-genome sequencing reads from a sixteen member family , obtaining for the first time distributions of copy number in this family ., Our copy number calls are consistent with the family pedigree and confirm known multigene variants of the IGHV locus ., Our results also suggest evidence of copy number variants that are mosaics of the existing reference haplotypes and variants that might be transitional between them ., The main difficulty in accurately genotyping the IGHV locus is the high level of similarity between alleles of different gene segments ., Fig 1A illustrates the level of nucleotide similarity between the IGHV segments in GRCh37 ., For example , the alleles of segments 3-30 and 3-33 in GRCh37 , circled in Fig 1A , differ in only 1 . 4% of their nucleotides ., Since some segments have alleles that differ by more than 1 . 4% in their base pairs ( Fig 1B ) , it becomes problematic to distinguish between alleles of the GRCh37 genes 3-30 and 3-33 based on nucleotide dissimilarity ., To be more concrete , if one had reads of length 100 bp from a haplotype containing both 3-30 and 3-33 segments , it would be algorithmically very difficult , if not impossible , to correctly map reads that are from regions common to 3-30 and 3-33 ., We note that this difficulty in distinguishing between alleles becomes even more pronounced when analyzing antibody repertoire sequencing data , where somatic hypermutation further confounds the matching of repertoire sequences to germline alleles 34 ., This problem also occurs with other gene segments: across all full-length functional IMGT alleles , there is a 10 . 6% overlap in the distribution of nucleotide differences between alleles with the same segment name and alleles with distinct segment names ( Fig 1B ) ., Reads from the alleles in this overlapping region cannot be operationally distinguished from each other , leading to unreliable and ambiguous read mapping results ., Thus , in the context of mapping short reads , it does not make sense to keep these alleles separate , so we pool them together into units we call “operationally defined gene clusters” , or gene clusters for short ., As we show in the next sections , this strategy allows us to extract useful information , such as copy number estimates , with less ambiguity ., To determine these gene clusters in a systematic manner , we perform hierarchical clustering within each family of full-length , functional IMGT alleles ( Materials and Methods ) ., By grouping the alleles together according to their sequence similarity , we reduce the overlap to a greater extent than grouping according to segment name alone ., ( Fig 1C ) ., We see that although we cannot eliminate the overlap completely , in most gene clusters , alleles are within 5% nucleotide differences of each other ., Some families have clearly defined gene clusters ., In family 1 , the gene clusters correspond to segment name , as long as duplicate segments 1-69D and 1-69 are merged ( Fig 1D ) ., In families 2 and 5 , which have three and two segments respectively , the alleles cluster by segment name ( S1 and S2 Figs ) ., In family 3 , six segments that have distinct names—namely , 3-30 , 3-30-3 , 3-30-5 , 3-33 , 3-53 , and 3-66—form two gene clusters {3-30 , 3-30-3 , 3-30-5 , 3-33} and {3-53 , 3-66} ( Fig 1E ) ., Families 6 and 7 each have only one functional gene segment and therefore do not require clustering ., Surprisingly , the same clustering algorithm that leads to clean gene clusters in the other families fails to identify clear-cut gene clusters in family 4 ( Fig 1F ) ., Not only are the boundaries between gene clusters fuzzy in this case , but alleles of the same segment cluster separately ., For example , 4-4*01 and 4-4*02 cluster separately from 4-4*07 and 4-4*08 ., The alleles in family 4 also seem to be more similar to each other than alleles in other families ., It is not clear why alleles in family 4 in particular should cluster poorly compared to those of other families ., Gene conversion events in IGHV family 4 and a more recent common ancestor than that of other IGHV families are both possible explanations that are consistent with the observed distance matrix ., A better clustering , based on a combination of mutational distance and indel distance , was ultimately used to define the gene clusters for family 4 ( S4 Fig ) ., With the caveat that family 4 gene clusters are more speculative , Table 1 summarizes the operationally defined gene clusters as determined by hierarchical clustering ., Only gene clusters which disagree with the IMGT V gene segment name are listed ., For the remainder of this manuscript , we use the term gene cluster to refer to our operationally defined clusters and IMGT V gene segment to refer to the standard IMGT nomenclature ., When an operationally defined gene cluster is the same as an IMGT V gene segment—e . g . , in the case of segments in gene families 5 , 6 , and 7—we use the terms interchangeably ., It may help the reader to keep in mind that with the exception of family 4 , the majority of gene clusters coincide either with the IMGT V gene segment names , or with IMGT V gene segments merged with their duplicates ( e . g . , 3-64 and 3-64D ) ., The operationally defined gene clusters ( Table 1 ) address the main difficulty in genotyping the IGHV locus and is the key idea behind our data pipeline ( Fig 2 ) ., Without this crucial step , it is difficult to determine IGHV alleles from read mapping alone ( S1 Table ) ., The input of the pipeline is a file of whole-genome sequencing reads from an individual ., The output is a genetic profile of the IGHV locus: for each gene cluster it reports a point estimate of copy number , the closest matching existing IMGT allele , and a nucleotide sequence of the contig assembled from reads mapping to the gene cluster ., Fig 3 shows the performance of our pipeline at three levels of genotypic resolution on simulated reads from the two complete IGHV haplotype sequences ( Materials and Methods ) ., At the coarsest scale , we ask whether the pipeline correctly identifies the presence or absence of each gene cluster ., We find the pipeline to be highly accurate , with precision of 100% for all coverage depth ( 30× , 40× , 50× ) and read length ( 70 bp , 100 bp , 250 bp ) combinations ., This means that all the gene clusters identified by our pipeline are present in the reference ., The recall , the fraction of gene clusters in the reference that are identified by our pipeline , is 100% for all but two of the coverage depth/read length combinations ( Fig 3A ) ., At the next level of resolution , we ask whether the pipeline can correctly determine the copy number of each gene cluster ., We use the read coverage depth of the assembled contig as our point estimate for copy number ., Fig 3B shows that contig coverage depth is indeed correlated with copy number , though there is variation above and below the true copy number and some gene clusters which are present in single copy have high coverage depth ., This is because pseudogenes in the IGHV locus , which are not included in our reference set , may share common subsequences with functional genes ., Reads from pseudogenes can therefore be erroneously mapped , artificially inflating the contig coverage depth ., This is particularly an issue with 70 bp length reads as these reads are more likely to completely fall within a conserved region ., This problem can be partly alleviated with paired-end reads , a strategy we use on the real dataset in the next section ., At the highest level of resolution , we compare the assembled contig obtained from the pipeline to the known nucleotide sequence for each gene cluster ., When a gene cluster is only present in single copy in the locus , and if the read lengths are 250 bp , the recall of the nucleotide sequence is 100% in all but one of the simulated datasets ( Fig 3C ) ., With shorter reads , the frequency of correctly calling alleles is lower ., As with copy number determination , this lower accuracy is likely due to erroneously mapped reads from pseudogenes and highly similar functional genes that interfere with the assembly algorithm ., For the same reason , when a gene cluster is present in more than one copy and as different alleles , the allele calls are also less accurate ., Note that higher coverage depth does not necessarily improve accuracy because the error arises not from sequencing error , which occurs in random locations and can be mitigated with higher coverage depth , but from erroneously mapped reads , which are systematically incorrect regardless of coverage ., We next apply the pipeline to the publicly available Platinum Genomes dataset 40 , a set of whole-genome sequencing reads of length 100 bp at roughly 30× coverage depth from a family of 16 individuals ( four grandparents , a mother , a father , and ten children , all of European ancestry ) ., Because these reads are paired , we perform an additional filtering step ( Materials and Methods ) to discard reads that are potentially from pseudogenes in order to improve our allele calls and decrease the false discovery of duplicated genes ., A summary of copy number and allelic variation in IGHV gene clusters in this dataset is shown in Fig 4 ( S3 Table lists all raw coverage depth values from the dataset ) ., For all the results that follow , the raw coverage depth of each gene cluster is scaled by the coverage depth of segment 3-74 in the same individual to eliminate variation due to differences in read coverage between individuals ( IMGT V gene segment 3-74 coincides with the gene cluster 3-74 ) ., We choose segment 3-74 because it has no documented examples of copy number variation and is located at the telemoric end of the chromosome ., Specifically we assume that 3-74 has two copies , one on each chromosome , and divide the coverage depth of all other gene clusters by half of the coverage depth of 3-74 ., A normalized coverage depth of 1 therefore corresponds to a single copy on one , but not both , of the chromosomes ., Note that the coverage depth tends to decrease towards the 6-1 end of the locus due to VDJ recombination , an issue we will return to in the Discussion ., With the approach introduced here , we can begin to obtain population-level statistics on the IGHV locus from WGS data and systematically quantify variation with respect to operationally defined gene clusters ., Given the small sample size of the Platinum Genomes data , we have focused here on quantifying variation in genes known to vary in copy number ., As larger whole-genome sequencing datasets become available , it will be possible to compare IGHV copy number profiles at the population scale ., These profiles can then be studied to find correlations between multiple gene segments and clusters and to discover new copy number variants ., Even with the coarse measure of presence/absence of gene clusters , we can begin to address basic open questions such as whether there is a minimal number of IGHV gene clusters required for a healthy immune system and whether there is a common core set of IGHV gene clusters that are shared by all individuals ., Our study makes clear that read depth information can be used to accurately determine the presence and absence of gene segments and clusters ., However , complications remain for ascertaining copy number and allelic content to high accuracy ., The first complication arises from the cell type on which whole-genome sequencing is commonly performed ., The Platinum Genomes data were generated from immortalized B lymphocytes ., The IGHV locus in these cell types have undergone VDJ recombination ., This rearrangement , which truncates the IGHV locus , confounds the correlation between read coverage depth and copy number of a gene cluster ., We can see this from the pipeline output , where coverage depth tends to decrease towards the centromeric end of the locus ., The extent of this decrease can be quite marked , for example in the case of NA12877 , or not noticeable at all , for example in NA12891 ( Fig 7; the distribution of read coverage depth of all the individuals is summarized in S5 Fig ) ., If one knew the number of B cell lineages used to prepare the library and the fraction of haplotypes that underwent rearrangement , it is possible to adjust the raw coverage values to reflect actual coverage values ( S1 Appendix ) ., However , in the case of the Platinum Genomes data , this information is unavailable ., As whole-genome sequencing becomes more widespread , we anticipate that datasets from other cell types will become available and this issue will be resolved ., The second complication is that the majority of whole-genome sequence reads are generated from diploid cells ., Because the majority of segments on both chromosomes are of different alleles , the single allele call generated by our pipeline may be composed of sequence from all the alleles present or represent just one of the alleles ., Allele calls can thus hide the heterozygous state of an individual ., S8 Fig gives examples of segments which are present as two alleles in the family and for which the allele calls are misleading ., This problem could be addressed with an assembler or method tailored to reconstruct the nucleotide sequence of alleles of short genomic regions ( popular assemblers are currently designed for whole-genome assembly ) ., Such a task is nontrivial , however , and beyond the scope of the current paper ., There has been some success in identifying unique alleles using an alternative data type: antibody repertoire sequencing data 8 , 16 , 46 ., However , such studies cannot directly quantify the copy number of an exactly duplicated gene because read abundances in these studies are not correlated with germline gene abundances due to differential gene usage in the development of an antibody repertoire ., Furthermore , the V gene segment can be truncated during the genomic rearrangement for producing the antibody coding sequence , so that full-length alleles may not always be obtained from antibody repertoire sequencing data ., We note that there are many existing methods for estimating copy number based on coverage depth using whole-genome sequencing 48–52 ., These methods , however , do not utilize the IMGT database of IGHV alleles nor do they specifically target the IGHV locus , a region with a higher amount of repetitions and duplications than most of the genome ., They therefore may be prone to biases introduced by targeting the entire genome , which has loci of varying characteristics , rather than targeting a particular region ., Additionally , some existing methods 53 intended for whole exome sequencing may be further biased when introduced to data from whole-genome sequencing ., True determination of IGHV haplotypes must ultimately come from sequencing the 1 Mb region in its entirety and in multiple individuals ., Indeed , because the GRCh37 reference is a chimera of three diploid haplotypes 13 , there is currently only one true reference haplotype for the IGHV locus ., However , the technology to accurately sequence structurally varying regions remains expensive and low-throughput ., We can instead take advantage of the increasing availability of whole-genome sequencing datasets and the extensive IMGT database to systematically describe this locus in a high-throughput manner albeit at lower genotypic resolution ., Using this strategy , we have found evidence of haplotypes that are mosaics of reference genome configurations or that are transitional between them ., The existence of these haplotypes further indicates that our approach of representing the locus in terms of a reference set of gene clusters is a less cumbersome means of cataloging the high copy number variation in this locus , compared to reconstructing full sequences of the IGHV locus with annotated breakpoints ., The fundamental strategy applied here is not specific to the IGHV locus ., Reads from whole-genome sequencing datasets can similarly be used to characterize other gene families and in other species , where the genes are of comparable length and similar level of diversity ., Some examples include T cell receptor genes and olfactory receptor genes ., The use of whole-genome sequencing data therefore need not be restricted to single nucleotide variants , but can also be applied to study regions exhibiting copy number variation ., IGHV genes are named according to their “family” and genomic location ., The families , numbered 1 to 7 , comprise genetically similar genes ., The segment 6-1 , for example , is in IGHV family 6 and is the first gene in the locus , counting from the centromeric end ., Gene names with a suffix “D” denote a duplicate gene , for example 1-69D , while an appended number , for example 1-69-2 , indicates that the gene was discovered subsequent to the original labeling and is located between 1-69 and 2-70 ., An allelic variant of an IGHV gene is denoted by a *01 , *02 , etc . , as in 1-69*01 , 1-69*02 ., Nucleotide sequences for IGHV gene alleles were downloaded from the IMGT database 37 ., Only full-length functional alleles were used for clustering ., Multiple sequence alignment was performed on each family of alleles using Fast Statistical Alignment with default parameterization ( FSA , 54 ) ., The aligned alleles were then clustered using the hclust function in R 55 ( method parameter set to “single” , although using the “complete” method gives the same result for all families with the exception of family 4 ) ., The clustering algorithm tries to organize the alleles so that alleles with higher nucleotide similarity are in the same cluster while those with lower nucleotide similarity are in different clusters ., The algorithm starts by first putting each allele in a separate cluster , then iteratively joining the two most similar clusters ., For example , the cladogram in S3 Fig shows how the clusters are formed for family 3 alleles ., For all the IGHV families except family 4 , gene clusters were determined using distance matrices calculated from Hamming distance based on FSA alignment , with gap differences treated in the same way as mutations ., Visual inspection of the alignment of family 4 suggested that indels may be important in partitioning the alleles ., Hence , a combination of an evolutionary distance “TN93” ( based on 56 ) and indel distance ( number of sites where there is an indel gap in one sequence and not the other ) was used to determine the gene clusters for family 4 ., R scripts are included as a supplementary file ( S1 File ) ., Our scripts and example datasets are available at: https://github . com/jyu429/IGHV-genotyping ., We assume the WGS data is in BAM or SAM format 57 , with reads already filtered to come from the IGHV locus ., For WGS reads aligned to GRCh37 , this is chr14:105 , 900 , 000-107 , 300 , 000 ., For reads aligned to GRCh38 , this is chr14:105 , 700 , 000-106 , 900 , 000 ( coordinates extend beyond the IGHV locus to be conservative ) ., Bowtie2 36 is used to map these reads to all functional , full-length IMGT alleles ( the same set used for hierarchical clustering ) ., The default Bowtie2 local alignment threshold led to too many multiple matches ., S9 Fig illustrates how we increased this threshold to be more restrictive ., Mapped reads are then pooled according to the gene clusters described in the Results section ., For example , all reads that map to the alleles of segments 3-30 , 3-30-3 , 3-30-5 , and 3-33 are pooled together ., SPAdes de novo assembler 38 is run on the pooled reads for each operational segment ., This assembler first performs error-correction on the reads and then attempts to piece together reads based on their overlap ., SPAdes has an option to report diploid contigs ( one for each chromosome ) , but running SPAdes with this option on the Platinum Genomes dataset did not produce more than one segment-length contig per gene cluster ., The assembled contigs are compared with the IMGT database using stand-alone IgBLAST 39 to determine the closest matching allele , the length of match , and the number of nucleotide mutations or indels that separate the contig from the closest-matching allele ., The read coverage depth of the contig as reported by SPAdes is also recorded for further analysis ., To test the capabilities and quality of our methods , ART 58 was used to generate simulated Illumina reads from GRCh37 and GRCh38 of lengths 70 , 100 , and 250 bp , each at coverage depths of 30× , 40× , and 50× ., Error profiles of simulated reads and adjustments to default ART parameters are illustrated in S10 and S11 Figs ., For the Platinum Genomes data , which comprises paired-end reads , we apply an additional filtering step to remove reads from pseudogenes that share a common subsequence with a functional gene ., One way to identify reads of a pseudogene is to compare its mapped position with the position of its mate ., If the mate read maps to a region that is substantially farther from the region the first read maps to ( we use a threshold of 1000 bp to be conservative ) then there is a chance it comes from a pseudogene and the original read is discarded ., S12 Fig demonstrates that this filtering step eliminates more than half the reads from pseudogenes ., Note that as a tradeoff , this filtering step will in some cases also incorrectly discard reads from duplicates that are located in a different region of the genome ., For segments where the starting position relative to the genome is undetermined , no filtering occurs ., In the case of the Platinum Genomes data , which is aligned to GRCH37 , this means that filtering is not applied to reads from segments 7-4-1 , 5-10-1 , 4-38-2 , 4-30-2 , and 1-69-2 ., For gene clusters that comprise more than one V gene segment , we use the position of the first segment in the cluster ( e . g . 3-53 for the gene cluster containing 3-53 and 3-66 ) as the mapped position of the first read ., Depending on how uniquely mappable the segments within a cluster are , this can also result in underestimates of gene cluster copy number ., Alleles of 7-4-1 have high nucleotide similarity to subsequences of pseudogenes 7-81 , 7-40 , and 7-34-1 ., The mate-pair filtering step above does not apply to 7-4-1 because the Platinum Genomes reads are aligned to GRCh37 , which does not contain 7-4-1 ., To filter out reads from these pseudogenes for 7-4-1 , we ran stand-alone IgBLAST on reads mapped to segment 7-4-1 ., The reads that had the highest match to a pseudogene were removed ., The remaining reads were then used as input for SPAdes de novo assembler .

Introduction, Results, Discussion, Materials and Methods

The study of genomic regions that contain gene copies and structural variation is a major challenge in modern genomics ., Unlike variation involving single nucleotide changes , data on the variation of copy number is difficult to collect and few tools exist for analyzing the variation between individuals ., The immunoglobulin heavy variable ( IGHV ) locus , which plays an integral role in the adaptive immune response , is an example of a complex genomic region that varies in gene copy number ., Lack of standard methods to genotype this region prevents it from being included in association studies and is holding back the growing field of antibody repertoire analysis ., Here we develop a method that takes short reads from high-throughput sequencing and outputs a genetic profile of the IGHV locus with the read coverage depth and a putative nucleotide sequence for each operationally defined gene cluster ., Our operationally defined gene clusters aim to address a major challenge in studying the IGHV locus: the high sequence similarity between gene segments in different genomic locations ., Tests on simulated data demonstrate that our approach can accurately determine the presence or absence of a gene cluster from reads as short as 70 bp ., More detailed resolution on the copy number of gene clusters can be obtained from read coverage depth using longer reads ( e . g . , ≥ 100 bp ) ., Detail at the nucleotide resolution of single copy genes ( genes present in one copy per haplotype ) can be determined with 250 bp reads ., For IGHV genes with more than one copy , accurate nucleotide-resolution reconstruction is currently beyond the means of our approach ., When applied to a family of European ancestry , our pipeline outputs genotypes that are consistent with the family pedigree , confirms existing multigene variants and suggests new copy number variants ., This study paves the way for analyzing population-level patterns of variation in IGHV gene clusters in larger diverse datasets and for quantitatively handling regions of copy number variation in other structurally varying and complex loci .

Regions of the human genome that vary in gene copy number are challenging to identify and analyze ., This is particularly true for the immunoglobulin heavy variable locus ( IGHV ) , which codes for a component of the antibody molecule ., Previous approaches to interrogate the IGHV locus using locus-specific assays have provided detailed information about genetic variation , but tend to be low-throughput ., Here , we introduce a method that leverages the increasing availability of large whole-genome sequencing datasets to genetically profile all functional IGHV genes in terms of a reference set of operationally defined gene clusters ., We demonstrate this approach both on simulated data and on reads from a sixteen-member family of European descent ., In the European family , not only did we find instances of known copy number variants , but also evidence of new variants ., As larger , more diverse , datasets become available , our approach will allow the investigation of inter-individual copy number variation in larger samples for this and similarly hypervariable regions .

sequencing techniques, pseudogenes, population genetics, genomic databases, platinum, genome analysis, gene types, molecular biology techniques, population biology, research and analysis methods, sequence analysis, genome complexity, sequence alignment, chromosome biology, biological databases, chemistry, molecular biology, genetic loci, nucleotide sequencing, haplotypes, cell biology, database and informatics methods, genetics, biology and life sciences, physical sciences, genomics, evolutionary biology, computational biology, chemical elements, chromosomes

journal.ppat.1003745

Identification of Novel Target Genes for Safer and More Specific Control of Root-Knot Nematodes from a Pan-Genome Mining

Plant-parasitic nematodes ( PPN ) cause significant damage to agriculture throughout the world ., A global survey in 1987 evaluated crop losses at $78–125 billion per year 1 ., More recent direct global estimates are not available , but when the increase in agricultural productivity is taken into account , the extrapolated 2001 loss for crops totaled $118 billion ( 11% of production ) 2 ., The current figure is thus probably much higher ., Measures such as growing resistant crop varieties and the use of nematicides are extensively employed to control PPN infections ., Billions of Euros have been spent annually on soil fumigants and other nematicides ., Current and previous chemical controls against nematodes are not only costly but they are highly toxic and hazardous , and involve application of environmentally unacceptable compounds ., Such toxicological problems and environmental damage caused by nematicides have led to banning of the most efficient chemicals that were commonly used so far ( EC directive 2007/619/EC ) ., In the absence of alternative control methods or development of specific and environmentally safe molecules , severe crop losses within major sectors of the agricultural industry are a distinct possibility ., Indeed , nematode problems recently re-emerged in some areas where the use of traditional nematicides had been abandoned for a short while 3 , 4 ., Therefore , novel control measures are urgently needed ., The identification of PPN-specific genes expressed during the interaction with the plant host is one of the most promising approaches for identification of new anti-parasitic strategies ., Infective PPN larvae in the soil are nearly microscopic worms , virtually invisible to the naked eye ., Although a few nematode species feed on above ground plant parts , such as leaves , stems , flowers , and seeds , the majority of these parasites feed on underground parts of plants , including roots , bulbs , and tubers ., Most PPN feed on root tissue and damage their host mainly by stunting the root system , resulting in reduced water uptake and by promoting microbial infections through wound sites or by serving as vectors for pathogenic viruses ., Some nematode species exhibit a hit-and-run strategy , remaining migratory during their plant root-associated life cycle ., An increase in complexity of host-parasite interactions is observed in sedentary parasite species with their enhanced capacity to manipulate host plant genes in their favor 5 ., These endoparasitic nematodes settle down after an initial migratory phase and assume a sedentary life style while transforming plant cells into complex feeding structures ., Nematodes of this category represent the most damaging species for crops ., Some of these nematodes have a relatively specialized host range ( e . g . cyst nematodes Heterodera and Globodera genus ) while others are able to reproduce on thousands of unrelated host plant species ( e . g . root-knot nematodes , Meloidogyne spp . ) ., Because root-knot nematodes represent the most economically-important PPN , they constitute the most explored group of species and can now be considered as one of the most advanced models for understanding mechanisms of plant parasitism in nematodes ., As with other PPN , they have a syringe-like stylet that is used to pierce and penetrate plant cell walls , to release esophageal secretions into the host tissue and to take up nutrients ., During their infective life-cycle root-knot nematode larvae penetrate plant root tissue and migrate along the vascular cylinder ., By injecting secretions into plant cells , they induce the formation of a feeding site indispensable for their development ., As a consequence of the formation of these feeding structures , root-knots or galls are observed as symptoms of the infestation ., Plant nutrient and water uptake are substantially reduced by the resulting damage to the root system , and infested plants are therefore weak and give low yields ., Once the feeding structure is established , female nematodes continue their development and eventually become pear-shaped and produce hundreds to thousands of eggs ., These eggs are then extruded as an egg-mass , protected within a gelatinous matrix , at the outer surface of the root ., Mining the genomes of root-knot nematodes 6 , 7 , 8 through an evolutionary and comparative genomics approach , we searched genes conserved in various plant-damaging species while otherwise absent from the genomes of non target species such as those of chordates , plants , annelids , insect pollinators and mollusks ., We identified a set of root-knot nematode genes absent from non-target species but present in several plant-damaging organisms ., Further bioinformatics pruning of this set of genes yielded new candidates that were silenced using RNA interference ( RNAi ) ., Upon silencing experiments , 75% of the candidates induced a significant and reproducible diminution of infestation and are thus particularly promising for the development of new and more specific control strategies ., Our main objective was to identify root-knot nematode ( RKN ) genes that could be used as targets for the development of new control means against these pests ., As we absolutely wanted to minimize the risk of collateral effects and preserve non-targeted species , we systematically discarded RKN genes that had putative homologs in non-target species ( Figure 1 and methods ) ., To select RKN proteins without predicted homologs in non-target species , we first performed an OrthoMCL 9 analysis comparing all predicted proteins in M . incognita and M . hapla ( 34 , 780 proteins ) with the whole proteomes of 23 other species ( Figure 2 ) ., This step was aimed at eliminating RKN proteins having evident orthologs in fully-sequenced non-target genomes and to substantially reduce the number of proteins that will be subsequently compared against the NCBIs nr library ., We selected , in priority , species whose whole genomes have been annotated to a quality level allowing a reliable prediction of the ensemble of protein-coding genes ., Our selection of species comprised 4 other nematodes , 5 insects , 9 vertebrates ( including mammals , ray-finned fishes , amphibian and sauropsida ) , 4 fungi and 1 plant ., Among selected species , we included two plant-pathogenic fungi ( Magnaporthe grisea and Fusarium graminearum ) , one nematode parasite of animals ( Brugia malayi ) and two insects that feed on living plant tissue ( Acyrthosiphon pisum and Bombyx mori ) ., The 18 other species were blacklisted and whenever a RKN protein had a predicted ortholog in these blacklisted species , the protein was discarded from the rest of the analysis ., According to OrthoMCL , a total of 15 , 181 RKN proteins had a predicted ortholog in at least one blacklisted species and were thus eliminated ., The rest of RKN proteins ( 19 , 599 ) had no predicted ortholog in any of the blacklisted species and passed this first filter ., Among these proteins , a total of 2 , 446 were redundant between M . incognita and M . hapla ., To avoid redundancy , and because subsequent biological assays will be performed in M . incognita , we kept as representative the M . incognita versions ., At the end of this first filtering step , a total of 17 , 153 Meloidogyne proteins were kept for further analysis ., Although , with a total of 25 species representing >500 , 000 proteins , the OrthoMCL analysis we performed is far from negligible , this only represents a limited sample of the whole sequence biodiversity available in public databases ., Thus , using a BLASTp 10 analysis , we compared the 17 , 153 RKN proteins that passed the OrthoMCL filter against the NCBIs nr library ., Applying a similar filter as was applied to the OrthoMCL results , we systematically eliminated RKN proteins having putative orthologs in non-target , blacklisted species ., Those that had no putative ortholog in any of the blacklisted species or returned no significant similarity at all in any other species , were kept for subsequent analysis ., Because there is no comprehensive database indicating the lifestyles of the plethora of species with a sequence in the nr library , we generated a list of blacklisted taxa ( methods ) ., In total , our blacklist included 170 , 258 species covering 4 whole clades ( annelida , chordata , mollusca and viridiplantae ) in addition to the 18 species already blacklisted in the OrthoMCL analysis ., Overall , a total of 10 , 105 RKN proteins did not return any significant BLASTp hit in nr using the thresholds we had set ( methods ) ., More than half of these proteins ( 5 , 536 ) also had no predicted ortholog in the OrthoMCL analysis and were thus considered as potentially orphan or restricted to RKN at this stage ., In contrast , 1 , 201 RKN proteins returned significant BLASTp hits in at least one blacklisted species and were discarded ., In total , 15 , 952 RKN sequences were kept and constituted our protein set 1 ., This set 1 represents RKN proteins predicted to be absent from blacklisted species and possibly present in other plant-damaging species ., We assessed whether part of the RKN proteins absent from non-target species were present in other plant-damaging species ., The rationale of this analysis is that the more a gene is shared between plant pests while absent from other species , the more it is likely to be involved in core interaction processes with the plant ., To assess conservation in plant-damaging species , we filtered the results of both the OrthoMCL and BLASTp analyses ., In the OrthoMCL analysis , two plant-pathogenic fungi were included as well as two insects that feed on plant ., A total of 4 , 398 RKN proteins had predicted orthologs in , and only in , these plant-damaging species ., Similarly to the list of blacklisted species for the BLASTp filtering , we built up a list of 28 , 054 potentially plant-damaging species in the NCBIs taxonomy ( methods ) ., We identified 1 , 252 RKN proteins that returned significant BLASTp hits with at least one plant-damaging species ., After removing redundancy between the OrthoMCL and BLASTp analyses , we obtained a non-redundant list of 5 , 297 RKN proteins absent from non-target species but present in at least two plant-damaging species ., To gain functional insight on the proteins that appeared restricted to RKN and other plant-damaging species , we searched and retrieved a series of functional annotations ., This included a search for signal peptides for secretion , a search for transmembrane regions , a search for known protein domains and associated functional annotations ., We also assessed whether corresponding genes had transcriptional support ., Root-knot nematodes and other plant parasites secrete , into plant tissue , proteins that support successful parasitism ., In nematodes , these proteins , called effectors are generally produced in esophageal gland cells and secreted via a syringe-like stylet in plant tissue ., Several RKN effectors have been characterized so far and shown to support parasitism by playing roles in different key processes such as degradation of the plant cell wall , suppression of plant defense , manipulation of plant cells to produce feeding structures or interaction with plant signaling pathways 11 , 12 , 13 ., Because these genes are directly involved in successful parasitism , they naturally constitute interesting targets to develop new control measures ., Provided that these proteins are specific to parasitic species they can lead to the development of more targeted and specific control measures ., In an a priori-based approach , we searched within protein set 1 , those presenting the same characteristics than known effectors ., Typically , effector proteins bear a signal peptide for secretion and no transmembrane region ., We also had identified previously , using the MERCI software 14 , a set of protein motifs that are frequent in known effector proteins but absent from housekeeping proteins in RKN ., Among protein set 1 , we found a total of 3 , 311 proteins that possessed a signal peptide , 2 , 453 that possessed an effector MERCI motif and 13 , 856 that had no predicted transmembrane region ., Overall , out of the 15 , 952 proteins present in set 1 , we found 993 proteins that cumulated all these 3 criteria and thus have the same characteristics than canonical effectors ., Because the M . incognita and M . hapla proteins have been deduced from the gene models predicted as part of automated genome annotations 6 , 8 , set 1 may contain a proportion of proteins deduced from wrongly or over-predicted genes ., To minimize the risk of functionally analyzing proteins representing false predictions , we required two additional criteria ., ( i ) The protein must be present in at least two different plant-damaging organisms ( including the two RKN species ) and ,, ( ii ) the corresponding gene must be supported by transcriptomic data from RKN ., We had previously assembled the ensemble of available M . incognita EST data , as described in 15 ., This represented a total of 63 , 816 ESTs assembled in 22 , 350 distinct unisequences ., Although substantial , this dataset can still be viewed as relatively limited ., To complete this relatively scarce transcriptomic dataset , we generated RNA-seq transcriptome sequencing for six different developmental life stages of M . incognita ( Table 1 and methods ) ., RNA-seq generated more than 190 million reads in total that were assembled in 137 , 733 contigs ( methods ) ., Combined with available ESTs , this dataset is likely to encompass a significant proportion of the diversity of transcripts in a RKN ., Out of the 15 , 952 proteins in set 1 , a total of 5 , 530 had a corresponding CDS sequence that received significant transcriptional support from RKN ESTs or RNA-seq data ( methods ) ., From the set of 109 putative transcription factors identified during the functional annotation , a total of 12 were supported by expression data and were present in at least two plant-damaging species ( Figure 1 ) ., From the set of 993 effector-like proteins , 232 were present in at least two plant-damaging species and were transcriptionally supported by alignments with Meloidogyne ESTs or RNA-seq data ( Figure 1 , Table S2 ) ., Among these 232 effector-like proteins , we found 42 previously reported RKN effectors , including SXP/RAL-2 like proteins 16 , Venom Allergen-like Proteins ( VAP ) 17 , Chorismate mutases 18 , Cathepsin L-like protease 1 ( MiCpl1 ) 19 as well as 32 plant cell wall-degrading enzymes , encompassing cellulases , xylanases , pectate lyases and expansin-like proteins 20 ., Finding previously known and characterized Meloidogyne effectors among our list of predicted effectors constituted an important validation of our approach ., Because the main aim of our genome mining approach was to find novel potential targets we were exclusively interested in the 190 remaining effector-like proteins ., Out of these 190 novel effector-like proteins , only 25 different Pfam domains were found in 46 proteins ., Because they all received transcriptional support from M . incognita and have a homolog in at least one additional plant-damaging species , we can rule out the hypothesis that they are the product of over-prediction due to gene calling software ., Having identified novel putative transcription factors and effector-like proteins , present in plant-damaging species but absent from blacklisted ones , we wanted to experimentally validate their potential as amenable targets for the development of new control methods ., Basically , we targeted selected genes one by one using small interfering RNAs ( siRNA ) on M . incognita infective J2 larvae , and infected host tomato plants with treated larvae ., Six weeks after inoculation , we compared the numbers of galls and egg masses in siRNA-treated and control nematodes , as described in the methods ., Starting from the 12 putative transcription factors and 190 novel effector-like RKN proteins , we further pruned the list according to the following criteria ., Because we perform biological assays on M . incognita , we first discarded proteins from M . hapla that had no ortholog in M . incognita ., To avoid potential compensation of the silencing effect by gene copies performing similar function , we also removed all proteins that were encoded by multigene families ., We ended up with a list comprising one putative transcription factor and 39 non-redundant effector-like proteins found in M . incognita , present in at least one other plant-damaging species , transcriptionally supported and without a homolog in a blacklisted species ( Figure 1 ) ., We examined the corresponding coding sequences for compatibility with the design of specifically-matching siRNAs and the design of quantitative PCR primers ( methods ) ., We were able to design specific siRNA as well as specific PCR primers for the putative transcription factor ( Minc07817 ) as well as for 15 out of the 39 genes encoding effector-like proteins ., These 16 protein-coding genes were all present both in the M . incognita and M . hapla genomes ., A total of 13 of the corresponding proteins do not have any predicted Pfam-A domain and , hence , no indication of the potential molecular function they may be involved in is available ., One of the proteins ( Minc03866 ) had a predicted C-type lectin domain and another ( Minc03313 ) had an Astacin ( peptidase family M12A ) domain ., Out of the 34 , 780 predicted proteins from the M . incognita and M . hapla whole proteomes , we have eliminated a total of 15 , 181 proteins because they had predicted orthologs in at least one of the 18 blacklisted species , based on OrthoMCL ., In comparison , our taxonomic BLASTp analysis against the NCBIs nr library allowed elimination of only 1 , 201 further RKN sequences ., This result suggests that our OrthoMCL filtering was able to eliminate most of the RKN proteins having potential orthologs in non-target species ., Despite our selection of 23 species compared to the RKN is far from representing a significant portion of the whole biodiversity available , it constituted a stringent filter , probably because representatives from various different lineages , ranging from fungi to vertebrates , were included ., This OrthoMCL filter also allowed us to dramatically reduce the number of proteins to be compared with the nr library in subsequent BLASTp comparison ., The 1 , 201 sequences eliminated at the taxonomic BLASTp step probably consisted of gene families not represented among the 23 compared species ., Besides allowing elimination of proteins having orthologs in non-target species , the OrthoMCL and BLASTp filters also allowed identification of RKN genes shared by several plant-damaging species ., A total of 5 , 297 non redundant RKN proteins were present in at least two plant-damaging species but absent in non-target species , according to OrthoMCL and BLASTp filters ., These proteins , apparently restricted to plant-damaging species , may be involved in core mechanisms common to several of these agricultural pests ., Another point of interest revealed by the OrthoMCL and BLAST analyses is the set of potential orphan genes in RKN ., A total of 5 , 536 non-redundant RKN proteins neither returned predicted orthologs in the OrthoMCL analysis nor had any significant BLASTp hits , in other species ., These apparently RKN-restricted proteins can represent true orphans but may also be the result of possible artifacts due to over-predictions made by gene calling software in RKN genomes ., However , 949 of the corresponding genes received transcriptional support from EST or RNA-seq data and are thus unlikely to be the results of over-predictions ., Similarly , 2 , 416 of these orphan genes are present both in the M . incognita and M . hapla genomes and it appears improbable that these genes have been over-predicted twice independently in two distinct genomes using distinct gene calling strategies ., These genes , apparently restricted to RKN and otherwise orphan , may be involved in processes specific to RKN such as the fine interactions between the nematode and the plant host ( e . g . induction of a feeding site in the plant ) or in the ontogeny of specialized organs ( e . g . gland cells or protrusible stylet ) ., Not only are those genes candidate targets for new treatments against RKN , but also fundamental genes to better understand adaptation to a plant-parasitic life ., Whether these genes are true orphans can be questioned when considering the relative scarcity of omics data available for plant-parasitic nematodes in general ., Our OrthoMCL analysis included only two proteomes of plant-parasitic nematode species ( M . incognita and M . hapla ) and to date , no whole proteome for a phytoparasitic nematode species is present in the NCBIs nr database ., Hence , these genes may have orthologs in other plant-parasitic nematode species ., Availability of further whole genomes , transcriptomes and deduced proteomes from additional phytoparasitic nematodes in the future will allow us to decipher whether some of these genes are shared with other plant-parasitic species and may , consequently , be involved in core processes linked to this lifestyle ., The series of filters we have set up in our bioinformatics pipeline resulted in a very stringent screening of the two whole RKN proteomes ., We have first eliminated all proteins that had potential orthologs in a series of blacklisted species that must be preserved if new nematode control means , targeting these genes , were developed ., We next ran two strategies in parallel to identify novel candidates in RKN proteomes that would be more clearly amenable for development of new control methods ., The first strategy was an ab initio data-driven one ., Because we noticed an over-abundance of putative transcription factors in the set of RKN proteins absent from blacklisted species , we focused on this category ., We identified 12 putative transcription factors absent from blacklisted species and supported by transcriptional evidence ., If these proteins actually function as transcription factors , they may be involved in regulation of genes involved in RKN-specific functions such as parasitism genes or modulate the expression of host plant genes ., One of those putative transcription factors was present as a single copy gene in M . incognita and was compatible with the design of specific siRNA and qPCR primers and thus amenable for biological assays ., The second strategy we used was an a priori based one ., Because effector proteins secreted by nematodes are known to be important in their plant-parasitic ability , we searched proteins that featured the same characteristics and identified a list of 232 putative effectors ., Validating our a priori strategy , we retrieved 42 proteins that were previously described as known effectors in the literature ., Obviously , not all effectors previously described so far were found ., This is mainly for the following reasons:, ( i ) several known effectors do not possess an N-terminal signal peptide and/or a MERCI effector-motif ( ii ) some nematode effectors have homologs in blacklisted species ., Because we were mainly interested in the discovery of novel potential targets , we focused our analysis on the 190 remaining novel effector-like proteins not present in blacklisted species ., A total of 39 corresponding genes were not redundant in M . incognita and present in at least one other plant-damaging species ., Out of these 39 genes , 15 were compatible with the design of siRNAs and qPCR primers and thus amenable for further biological assays ., During infestation tests on tomato plants , out of the 16 novel candidates identified ( 15 effector like and 1 putative transcription factor ) , 12 turned out to show significant and reproducible reduction in the number of egg masses or galls when treated with anti-candidate siRNAs ., Overall , our strategy was not to build a comprehensive list of candidate genes that might produce the most severe phenotypes on nematodes ., In contrast , the originality of our approach was to focus from the beginning on genes that were present in plant-damaging species but absent from non-target “blacklisted” species ., We thus produced a stringent and restrictive list of candidates that cumulated a series of characteristics that made them the most promising candidates for the development of safer and more-specific control methods ., Overall , after siRNA soaking , we measured significant and reproducible effect on infection on 12 targeted genes ., This effect was measured by a diminution in the number of galls or egg masses ., Reduction in the number of galls implies that fewer nematodes have managed to induce a feeding structure ., A reduction in the number of egg masses signifies that fewer female nematodes have managed to complete their development until the production of egg masses , a necessary step to propagate the infection at the next generation ., In 6 cases , we measured a reproducible and significant reduction in the number of galls ., Interestingly , 5 out of these 6 cases also led to significant and reproducible diminution of the number of egg masses ., This observation makes sense since reduction in the capacity of nematodes to form galls will have direct downstream impact on the number of egg masses produced ., Interestingly , targeting gene Minc01632 was responsible for both the most important reproducible diminution of the number of galls and of the number of egg masses ., The corresponding protein is 155 amino-acids long and has neither significant similarity in the NCBIs nr database nor predicted protein domain , as most of the 16 identified novel targets ., In contrast , observing significant reduction of the number of egg masses does not necessarily require upstream reduction of the number of galls ., Indeed , if the siRNA-targeted gene has functional consequences in processes that take place between the formation of galls and the production or extrusion of eggs we should observe a significant reduction in the number of egg masses but not in the number of galls ., This is indeed what we observed for 6 targeted genes ( Minc00801 , Minc03313 , Minc08335 , Minc12224 , Minc17713 and Minc07817 ) ., While reduction of the number of egg masses was significant and reproducible; reduction in number of galls was either not reproducible or did not reach the significance threshold ., Overall , we observed no correlation between reduction of infestation and a measurable effect on nematode motility or viability ., This indicated that the effect on infestation was globally not due to a toxicity of the siRNA treatment ., For instance , genes that showed among the most important reduction in the numbers of egg masses or galls ( e . g . Minc01632>40% reduction or Minc09526∼40% reduction ) did not show substantial diminution of viability or mobility 1 h or 16 h after soaking ., We can thus deduce that the reduced infestation observed is generally not a consequence of reduced motility but more likely results from modification in other processes important for parasitism ., Because the genes we have targeted are mostly specific to RKN and not shared by many species , we expected no systematic effect on viability or motility as opposed to evolutionarily conserved housekeeping genes 24 ., Treatments with siRNAs had reproducible significant effects on target transcript levels in 13 out of the 16 samples tested ( Table 2 ) ., Twenty-four hours after soaking , six genes showed a diminution of the transcript abundance while 7 yielded an increase of transcript level ., Because we suspected a possible bounce effect , we randomly picked 3 of these 7 genes and measured transcript abundance at an earlier time point ( 16, h ) ., One of the tested genes ( Minc03866 ) showed a significant and reproducible diminution of transcripts level at this time point ., It is possible that some of the six other genes that showed an increase of transcripts level at 24 h may also present an initial decrease at an earlier time point ., Such bounce phenomenon has already been described in plant-parasitic nematodes 21 , 22 , 23 ., Interestingly , the 13 siRNAs yielding effects on transcript level encompass 10 out of the 12 cases of reproducible significant reduction of infestation ., Furthermore , for 7 genes ( Minc00801 , Minc01632 , Minc02483 , Minc03866 , Minc08335 , Minc09526 , and Minc07817 ) , following the siRNA treatment , there is both a significant and reproducible diminution of the abundance of transcripts and of the infestation of nematodes ., Intriguingly , for two genes , there is significant and reproducible diminution of infestation but no significant effect on transcripts level ., Investigating earlier or later time points may reveal significant effects ., Alternatively , the corresponding mRNA may be sequestered away from the translation machinery without being itself degraded ., Such a mechanism of translation repression without mRNA degradation has already been documented in plants and animals 25 ., We performed in situ hybridization assays on the 12 genes that yielded significant reduction of infestation to try to gain information on their putative functions ., Because 11 of the 12 tested genes share characteristics with known RKN effectors , it could be expected that they show transcription localization patterns similar to the known effectors ., Canonical effectors are transcribed in secretory gland cells for injection by the nematode in plant tissue ., We found one gene expressed specifically in the subventral gland cell ( Minc03866 ) ., This gene could well encode an effector protein eventually secreted in plant tissue during infestation ., Interestingly , when targeted via siRNA , this gene returned one of the strongest effect on reduction of infestation ., Ubiquitous expression , which includes the secretory gland cells , was observed for 3 genes and these genes could be multi-functional , including possibly effectors depending on whether they are eventually secreted in planta or not ., A total of 5 genes returned no detectable signal and although they may function as effector , there is no further supporting data from in situ hybridization assays ., For the three other genes , expression localization does not support a possible secretion in planta , at least at the observed J2 stage ., One gene ( Minc02483 ) shows an expression localization specifically on nerve tissue surrounding a region called the metacorpus ., The metacorpus acts as a pump to inject secretion or to take up nutrients from the nematode syringe-like stylet ., It is possible that the gene expressed in the surrounding nerve cells may be involved in correct functioning of this pump ., siRNA targeted against this gene led to the second strongest reduction effect on the number of galls and egg masses ., The two other genes have an expression restricted to the intestinal tract and their targeting by siRNAs leads to significant and reproducible reduction of the number of egg masses ., Lacking any known protein domain , it would be too speculative to predict any function for the corresponding gene products ., Using soaking experiments with siRNAs targeting each of the 16 identified novel genes , we noticed a significant and reproducible diminution of infestation in 12 cases ., These results were obtained by inoculating infective J2 larvae after one hour soaking in a solution containing a siRNA concentration of 0 . 05 mg/ml ., Although siRNA delivery via soaking can be relatively efficient because of systemic propagation of the RNA interference , levels of inactivation can vary and duration of the effect is poorly known 26 ., Thus , it is possible that some of the genes we have identified would show significant reduction of infestation only when targeted at later stages of the nematode life cycle ., Unfortunately , J2 infective larvae is the only free-living stage that can be targeted with soaking approaches , the rest of RKN life cycle takes place within plant t

Introduction, Results, Discussion

Root-knot nematodes are globally the most aggressive and damaging plant-parasitic nematodes ., Chemical nematicides have so far constituted the most efficient control measures against these agricultural pests ., Because of their toxicity for the environment and danger for human health , these nematicides have now been banned from use ., Consequently , new and more specific control means , safe for the environment and human health , are urgently needed to avoid worldwide proliferation of these devastating plant-parasites ., Mining the genomes of root-knot nematodes through an evolutionary and comparative genomics approach , we identified and analyzed 15 , 952 nematode genes conserved in genomes of plant-damaging species but absent from non target genomes of chordates , plants , annelids , insect pollinators and mollusks ., Functional annotation of the corresponding proteins revealed a relative abundance of putative transcription factors in this parasite-specific set compared to whole proteomes of root-knot nematodes ., This may point to important and specific regulators of genes involved in parasitism ., Because these nematodes are known to secrete effector proteins in planta , essential for parasitism , we searched and identified 993 such effector-like proteins absent from non-target species ., Aiming at identifying novel targets for the development of future control methods , we biologically tested the effect of inactivation of the corresponding genes through RNA interference ., A total of 15 novel effector-like proteins and one putative transcription factor compatible with the design of siRNAs were present as non-redundant genes and had transcriptional support in the model root-knot nematode Meloidogyne incognita ., Infestation assays with siRNA-treated M . incognita on tomato plants showed significant and reproducible reduction of the infestation for 12 of the 16 tested genes compared to control nematodes ., These 12 novel genes , showing efficient reduction of parasitism when silenced , constitute promising targets for the development of more specific and safer control means .

Plant-parasitic nematodes are annually responsible for more than $100 billion crop yield loss worldwide and those considered as causing most of the damages are root-knot nematodes ., These nematodes used to be controlled by chemicals that are now banned from use because of their poor specificity and high toxicity for the environment and human health ., In the absence of sustainable alternative solutions , new control means , more specifically targeted against these nematodes and safe for the environment are needed ., We searched in root-knot nematode genomes , genes conserved in various plant-damaging species while otherwise absent from the genomes of non target species such as those of chordates , plants , annelids , insect pollinators and mollusks ., These genes are probably important for plant parasitism and their absence from non-target species make them interesting candidates for the development of more specific and safer control means ., Further bioinformatics pruning of this set of genes yielded 16 novel candidates that could be biologically tested ., Using RNA interference , we knocked down each of these 16 genes in a root-knot nematode and tested the effect on plant parasitism efficiency ., Out of the 16 tested genes , 12 showed a significant and reproducible diminution of infestation when silenced and are thus particularly promising .

journal.ppat.1000194

Recruitment of the Complete hTREX Complex Is Required for Kaposis Sarcoma–Associated Herpesvirus Intronless mRNA Nuclear Export and Virus Replication

The nuclear export of mRNA composes one part of a larger network of molecular events that begin with transcription of the mRNA in the nucleus and end with its translation and degradation in the cytoplasm ., During trafficking to the cytoplasm , a nascent mRNA undergoes numerous co-transcriptional processing steps , including 5′ capping , splicing to remove introns and 3′ polyadenylation 1–3 ., Of these events it has become clear that splicing is particularly important for mRNA nuclear export 4 ., The question of exactly which proteins regulate mRNA nuclear export has been the focus of several recent reviews 5–8 ., Two distinct multi-protein complexes are recruited to cellular mRNAs as a consequence of splicing , namely the human transcription/export complex ( hTREX ) and the exon-junction complex ( EJC ) ., The hTREX complex contains the proteins Aly ( a NXF/TAP-adapter ) , UAP56 ( a RNA-helicase ) and the hTHO-complex ( a stable complex composed of hHpr1 , hTho2 , fSAP79 , fSAP35 and fSAP24 ) 9 ., A second multi-protein complex , termed the exon-junction complex ( EJC ) is deposited 20–24 nucleotides upstream of the exon-exon boundary during splicing ., Until recently it was believed that Aly and UAP56 were components of the EJC 7 , 10–12 , however , new evidence suggests that Aly and UAP56 are associated exclusively with hTREX and not with the EJC ., Therefore , these results suggest that hTREX and EJC are distinct complexes , bind at separate locations on the spliced mRNA 13 and have separate functions , where hTREX directs nuclear export of mRNA and the EJC may instead monitor mRNA fidelity and function during translation 14–16 ., At present , it is not fully understood what regulates hTREX assembly on the mRNA but in addition to splicing the 5′ cap is also essential for its recruitment 9 , 13 ., Specifically , an interaction between Aly and the cap-binding complex protein , CBP80 appears to be critical for assembly ., Indeed , the 5′ cap has been shown to be required for mRNA export in Xenopus oocytes 13 ., In contrast to the EJC which binds near each exon-exon boundary , hTREX is recruited exclusively to the 5′ end of the first exon , presumably regulated in part by the reported interaction between CBP80 and Aly 13 ., It has been suggested that localising the export proteins at its 5′ end affords the mRNA polarity when exiting the nuclear pore ., Therefore , a current model for mRNA export favours a situation where hTREX is recruited to the 5′ cap of spliced mRNA and once bound Aly stimulates the recruitment of the export factor , TAP ., TAP then interacts with p15 and the nucleoporins , providing the connection between the ribonucleoprotein ( RNP ) and the nuclear pore 17 ., The functional roles , if any , played by UAP56 and hTHO-complex in this process remain poorly characterised ., Kaposis sarcoma-associated herpesvirus ( KSHV ) /Human herpesvirus 8 ( HHV8 ) is a γ-2 herpesvirus associated with a number of AIDS-related malignancies including Kaposis Sarcoma ( KS ) , primary effusion lymphoma ( PEL ) and multicentric Castlemans disease 18–21 ., In contrast to the majority of mammalian genes , a property shared amongst all herpesviruses is that a proportion of lytically expressed viral genes lack introns ., Although , KSHV expresses a higher proportion of spliced genes than other herpesviruses , it still encodes a significant proportion of lytically expressed late structural genes which lack introns ., KSHV replicates in the nucleus of the host mammalian cell , and therefore requires its intronless mRNAs to be exported out of the nucleus to allow viral mRNA translation in the cytoplasm ., This raises an intriguing question concerning the mechanism by which the viral intronless mRNAs are exported out of the nucleus in the absence of splicing ., To circumvent this problem , and to facilitate viral mRNA export , herpesviruses of all subfamilies encode a functionally conserved phosphoprotein which has an essential role in viral lytic replication 22 ., In KSHV this protein is encoded by the intron-containing open reading frame 57 ( ORF57 ) and has been the subject of several recent reviews 23–26 ., The ORF57 gene product interacts with Aly , binds viral mRNA , shuttles between the nucleus and the cytoplasm and promotes the nuclear export of viral mRNA transcripts 27–31 ., These properties are also conserved in ORF57 homologues such as ICP27 from Herpes simplex virus type-1 ( HSV-1 ) , SM protein from Epstein Barr virus ( EBV ) 32–35 and the Herpesvirus saimiri ( HVS ) ORF57 protein 27 , 31 , 36–38 ., Here we show that KSHV ORF57 interacts during viral replication with CBP80 and hTREX , but not the EJC ., We further show that ORF57 orchestrates the assembly of hTREX onto an intronless viral mRNA ., The ORF57-mediated recruitment of hTREX is achieved via a direct interaction between ORF57 and Aly ., Furthermore , in vitro data showed that UAP56 acts as a bridge between Aly and the hTHO-complex protein hHpr1 , thereby facilitating the formation of the complete hTREX complex ., When we prevented the recruitment of Aly onto intronless viral mRNA using an ORF57 Aly-binding mutant , this resulted in a failure of ORF57-mediated viral mRNA export and significantly reduced virus replication ., Strikingly , expression of a dominant negative Aly mutant that prevented the recruitment of UAP56 and hTHO-complex onto intronless viral mRNA resulted in a dramatic reduction in intronless viral mRNA export and infectious virus production ., We therefore propose that the entire hTREX complex must be recruited to intronless viral mRNA by ORF57 in order for efficient intronless mRNA nuclear export and KSHV replication to occur ., The hTREX complex contains several nuclear export proteins ., Given that KSHV ORF57s primary role is attributed to the nuclear export of intronless viral mRNA , we first assessed if ORF57 interacted with hTREX components using co-immunoprecipitation assays ., Moreover , as hTREX forms a complex with the 5′-cap protein CBP80 13 , we were interested if ORF57 also interacted with CBP80 ., 293T cells were transfected with pGFP or pORF57GFP and untreated or RNase treated total cell lysate was used in co-immunoprecipitation experiments with CBP80- , Aly- , UAP56- , fSAP79- and hHpr1- specific antibodies in addition to an unrelated antibody control ( a p53-specific antibody ) ., Each of the hTREX proteins and CBP80 co-precipitated with ORF57 , in an RNA-independent manner ( Fig . 1A ) ., Moreover , indirect immunofluorescence showed that a proportion of ORF57GFP co-localised with hTREX proteins ( Fig . S1 ) ., To assess whether ORF57 also interacts with the EJC , co-immunoprecipitation assays were repeated using an antibody specific for eIF4A3 , a core EJC component 39 and a hHpr1-specific antibody , serving as a positive control ., No interaction was observed with the EJC core component , eIF4A3 , in contrast , ORF57 was readily detectable in the hHpr1 immunoprecipitation ( Fig . 1B ) ., A control immunoprecipitation was performed to confirm that the eIF4A3 antibody precipitated EJC components ( Y14 ) in this assay ( data not shown ) ., In order to address potential overexpression artefacts and to assess whether ORF57 interacts with hTREX core components during lytic replication , KSHV-latently infected BCBL-1 cells were reactivated using the phorbol-ester , TPA , and lytic gene expression confirmed by detection of the ORF57 protein in TPA-treated cells by western blot analysis ( Fig ., 1C ( i ) ) ., Reactivated BCBL-1 cell lysate remained untreated or was treated with RNase and co-immunoprecipitations performed using an ORF57-specific antibody ., Western blot analysis using CBP80- and hHpr1- specific antibodies revealed that ORF57 interacts with CBP80 and hHpr1 during lytic replication , however ORF57 did not precipitate with either eIF4A3 ( the EJC core component ) or the cellular intronless mRNA-export protein , SRp20 ( Fig . 1Cii ) ., Moreover , to confirm that ORF57 failed to interact with additional components of the EJC , co-immunoprecipitations were repeated using reactivated BCBL-1 cell lysates and Y14- and Magoh-specific antibodies ., Results demonstrate that ORF57 did not precipitate with these additional EJC components ( Fig . 1Ciii ) ., A control immunoprecipitation was also performed to confirm that the Y14- and Magoh-specific antibodies precipitated eIF4A3 in this assay ( Fig . S2 ) ., Therefore , these data provide the first direct evidence of a viral protein associating with CBP80 and all the core components of the hTREX complex ., One possible explanation for how herpesvirus intronless mRNAs undergo nuclear export is that ORF57 mimics splicing by loading key mRNA export proteins , such as hTREX , onto the intronless viral mRNA ., In order to test if intronless KSHV transcripts were associated with hTREX proteins and if ORF57 was necessary for this interaction , RNA-immunoprecipitation ( RNA-IP ) assays were performed ., We chose to perform this assay using 2 intronless KSHV mRNAs , specifically ORF47 and gB ., RT-PCR and sequence analysis confirmed that both of these ORFs do not contain introns ( data not shown ) ., To perform the RNA-IPs , a vector expressing KSHV ORF47 ( a late structural intronless gene ) was transfected into 293T cells either alone or in the presence of pORF57GFP ., Total cell lysates were then used in immunoprecipitations performed with either CBP80- , Aly- , UAP56- or hHpr1-specific antibodies ., RNA-IPs performed on cell extracts transfected with ORF47 alone failed to show an interaction between Aly , UAP56 or hHpr1 and the viral ORF47 mRNA ( Fig . 2A ) ., In contrast , extracts from cells transfected with both pORF47 and pORF57GFP displayed a clear interaction between Aly , UAP56 and hHpr1 and the intronless viral ORF47 mRNA ( Fig . 2A ) ., CBP80 was found to bind to the intronless ORF47 viral mRNA independently of ORF57 ( Fig . 2A ) ., Moreover , this analysis was repeated with a second intronless KSHV mRNA , namely the late structural glycoprotein gB , and similar results were observed ( Fig . 2C ) ., These data show that ORF57 is required for the recruitment of core components of hTREX onto intronless viral mRNA ., To determine whether EJC components are recruited to intronless viral transcripts prior to export , RNA-IP assays were also performed using eIF4A3- , Y14- and Magoh-specific antibodies ., Results failed to show any interaction between the EJC core components and viral intronless ORF47 and gB mRNAs in the absence or presence of ORF57 ( Fig . 2B and 2C ) ., These results show that the EJC is not recruited to intronless viral transcripts by ORF57 and suggests that the EJC is not required for KSHV intronless viral mRNA nuclear export ., To determine whether the hTREX and EJC components were recruited to a spliced viral transcript , RNA-IPs were also performed using a vector expressing the genomic ( intron-containing ) KSHV ORF50 gene ., 293T cells were transfected with pORF50 in the absence or presence of ORF57 ., Total cell lysates were then used in immunoprecipitations performed with either CBP80- , Aly- , UAP56- , hHpr1- , eIF4A3- , Y14- or Magoh-specific antibodies ., Results demonstrated that CBP80 , hTREX and EJC components were recruited to the spliced ORF50 mRNA in an ORF57 independent manner ( Fig 2D ) ., This suggests that splicing of a viral transcript is sufficient to recruit the cellular proteins necessary for nuclear export ., In contrast , ORF57 is required for the recruitment of the hTREX proteins to an intronless viral transcript ., Currently , while it is known that hTREX recruitment to a mammalian mRNA is both 5′-cap- and splicing-dependent , the protein-protein interactions that govern assembly of the hTREX complex itself are not fully understood ., As ORF57 functions to recruit hTREX onto the intronless viral mRNA in a splicing independent manner we assessed whether this viral-system could be used to investigate hTREX assembly in more detail ., To this end , we sought to determine if any hTREX proteins directly interacted with ORF57 ., Radio-labelled ORF57 was generated by in vitro coupled transcription/translation ( ITT ) , RNase treated , and used in GST pull-down experiments using constructs expressing GST- , GST-Aly , GST-UAP56 and GST-hHpr1 fusion proteins ., Equal amounts of each expressed protein were used in each pulldown experiment ( Fig . 3A ) ., Analysis showed that ORF57 bound directly to GST-Aly but not to any other hTREX component ( Fig . 3B ) ., Due to the instability of GST-CBP80 , a reverse pulldown experiment was performed using GST-ORF57 ( Fig . 3C ) and radio-labelled ITT CBP80 , a GST-Aly pulldown with ITT CBP80 served as a positive control 13 ., Results also revealed a direct interaction between CBP80 and KSHV ORF57 ( Fig . 3D ) ., These data suggest that ORF57 only interacts directly with Aly and CBP80 , therefore the question remains how the complete hTREX complex associates with ORF57 ., It has previously been suggested that the hTREX complex is formed by UAP56 bridging the interaction between Aly and the hTHO-complex 9 ., Therefore , to further investigate ORF57-hTREX assembly , we assessed which hTREX components were required to reconstitute the ORF57-hHpr1 interaction ., GST pulldown experiments were performed using GST-hHpr1 and ITT ORF57 alone or combinations with ITT Aly or recombinant UAP56 ., When the GST-hHpr1 ITT ORF57 pulldown was repeated in the presence of both ITT Aly and purified UAP56 , analysis revealed a clear interaction between hHpr1 and ORF57 ( Fig . 3E ) , suggesting that ORF57 requires both Aly and UAP56 to recruit the hTHO-complex , thus facilitating formation of the ORF57-hTREX complex ., These findings provide the first direct evidence that UAP56 functions as a bridge between Aly and the hTHO-complex component hHpr1 to facilitate assembly of hTREX ., However , at present we cannot exclude the possibility that ORF57 interacts directly with other hTHO-complex components ., To assess whether hTREX is essential for viral mRNA nuclear export we produced an ORF57 mutant protein which was unable to interact with Aly and as such would be predicted to prevent the recruitment of the complete hTREX complex onto intronless viral mRNA ., A minimal region responsible for Aly-binding has been identified in ORF57 and spans 35aa between residues 181 and 215 28 ., Upon closer examination of this sequence , we identified a PxxP-polyproline motif ., To assess whether this motif was important for Aly-binding , both proline residues were substituted with alanine residues by site-directed mutagenesis to generate pORF57PmutGFP ., To determine if mutating the PxxP-motif in ORF57 led to a loss of Aly binding , GST-Aly pulldown assays were performed using ITT ORF57 or ITT ORF57Pmut ., Results demonstrated that the mutant ORF57 protein was unable to interact with GST-Aly , in contrast to the wild type protein ( Fig . 4A ) ., Moreover , similar results were observed using pull-down assays with pGFP- , pORF57GFP- or pORF57PmutGFP-transfected 293T cell lysates ( Fig . 4B ) ., These data demonstrate that the ORF57 PxxP-motif is required for the direct interaction with Aly ., To confirm that the mutagenesis of the PxxP motif had no effect on ORF57 protein stability or other reported functions , several independent experiments were performed to assess the ability of ORF57PmutGFP to localise to nuclear speckles , homodimerise , directly interact with ORF50 and bind viral intronless mRNA ( Fig . S3 ) , all of which are features of the wild type ORF57 protein ., In each case the ORF57PmutGFP phenotype was indistinguishable from that of wild type ORF57 ., Having established that ORF57PmutGFP is unable to interact with Aly and that the mutation does not affect other ORF57 functions , we then asked if , in the absence of Aly-binding , ORF57 was still able to complex with CBP80 and hTREX components ., 293T cells were transfected with pGFP , pORF57GFP or pORF57PmutGFP and total cell lysates were used in co-immunoprecipitation experiments , using CBP80- , Aly- , UAP56- , and hHpr1-specific antibodies ., In each case the hTREX antibody immunoprecipitated ORF57GFP but not ORF57PmutGFP , demonstrating that in the absence of the Aly-interaction ORF57 was unable to form a complex with hTREX ( Fig . 4C ) ., In addition , the ORF57PmutGFP exhibited a reduced but specific binding to CBP80 ( Fig . 4C ) ., This reduced binding may be due to the mutation of the PxxP-polyproline motif either affecting CBP80 binding directly or the loss of hTREX binding affects the stability of the CBP80-ORF57 complex ., To further investigate whether the mutation of the PxxP-polyproline motif affected direct binding to CBP80 , GST pulldown assays were performed using GST-ORF57 and GST-ORF57PmutGFP ., Equal amounts of each expressed protein was incubated with radio-labelled ITT CBP80 ., Results demonstrated that ORF57 and ORF57PmutGFP bound to CBP80 with similar affinity ( Fig . S4 ) ., This suggests that the reduced binding observed between ORF57PmutGFP and CBP80 may be due to the loss of hTREX , which is possibly required to stabilise the export competent vRNP ., To determine if ORF57PmutGFP was unable to recruit hTREX proteins to KSHV intronless mRNA transcripts in the absence of Aly binding , RNA-IP assays were performed using CBP80- , Aly- , UAP56- or hHpr1-specific antibodies ., These data demonstrate that in contrast to pORF57GFP , pORF57PmutGFP is unable to recruit hTREX components to intronless viral mRNA ( Fig . 4D ) ., This suggests that a direct interaction between Aly and ORF57 is required for hTREX recruitment onto intronless viral transcripts ., To test if a failure in ORF57-mediated recruitment of hTREX to the intronless ORF47 mRNA prevented nuclear export of intronless KSHV transcripts , two independent mRNA export assays were performed ., Firstly , northern blotting was used to detect if intronless ORF47 mRNA was present in the nuclear or cytoplasmic fraction of transfected cells ., Very little ORF47 mRNA was detected in the cytoplasmic RNA fraction of cells transfected with pORF47 alone ( 9 . 9±4 . 9% ) , whereas cells co-transfected with pORF47 and pORF57GFP displayed a clear shift in ORF47 mRNA from the nuclear to the cytoplasmic fraction ( 81 . 5±1 . 0% ) , indicative of ORF57-mediated viral mRNA nuclear export ., However , upon co-transfection with pORF47 and pORF57PmutGFP , the majority of ORF47 mRNA was no longer found in the cytoplasmic fraction ( 21 . 3±3 . 8% ) , instead it was retained in the nuclear pool at similar levels to those seen for the negative control , symptomatic of a failure in ORF57-mediated viral mRNA nuclear export ( Fig . 5A ) ., To confirm that the ORF57 mutant did not affect mRNA stability , total RNA levels were assessed by northern blot analysis ., No significant difference in ORF47 mRNA levels was observed between cells expressing wild type or mutant ORF57 proteins ( Fig . 5A , right panel ) ., However , a slight decrease in total mRNA levels is seen in the presence of both the ORF57 or ORF57PmutGFP compared to the GFP control ., At present , the reason for this is unknown , however , it could be due to the overexpression of the ORF57 protein ., To confirm the above result , a fluorescent in situ hybridisation assay was utilised ., 293T cells were transfected with pORF47 , in addition to either pGFP , pORF57GFP or pORF57PmutGFP ., 24 h post-transfection cells were fixed , permeabilised and incubated with a biotin-labelled oligonucleotide specific for the KSHV ORF47 mRNA ., After a 4 hr hybridisation cells were washed and ORF47 mRNA subcellular localisation was visualised using Cy5-streptavidin ., Cells transfected with pORF47 and GFP retained the ORF47 mRNA in the nucleus , whereas ORF47 mRNA was clearly visualised in the cytoplasm of cells transfected with pORF47 and pORF57GFP ., However , upon transfection with pORF57PmutGFP , ORF47 mRNA was only observed in the nucleus , symptomatic of a failure in ORF57-mediated viral mRNA nuclear export ( Fig . 5B ) ., Together , these two independent assays demonstrate that the ORF57-dependent recruitment of hTREX to intronless viral transcripts is essential for their efficient nuclear export ., We were also interested to determine whether the recruitment of the complete hTREX complex is required for virus replication and infectious virion production ., To this end , we utilised a 293T cell line harbouring a recombinant KSHV BAC36-GFP genome 40 ., This KSHV-latently infected cell line can be reactivated releasing infectious virus particles in the supernatant which can subsequently be harvested and used to infect 293T cells 41 ., The 293T-BAC36 cell line was transfected with pGFP , pORF57GFP or pORF57PmutGFP and concurrently reactivated using TPA and incubated for 72 hours ., The supernatants from each flask were then harvested and used to re-infect 293T cells and GFP positive cells were scored 48 h post-infection , as described above ., Results revealed similar levels of lytic replication and virus production from cells expressing pGFP or pORF57GFP ., However , virus production was significant reduced ( P\u200a=\u200a0 . 018 ) upon the expression of the ORF57PmutGFP ( Fig . S5 ) ., Therefore , these results demonstrate that the ORF57-dependent recruitment of the complete hTREX complex to intronless viral transcripts is essential for efficient virus lytic replication and infectious virion production ., The above data show that ORF57 binds viral intronless mRNA and directly interacts with Aly ., Given that Aly is able to recruit the export factor TAP directly , it was of interest to determine if UAP56 and the hTHO-complex are required for viral mRNA export ., In contrast to the cellular mRNA model , a major advantage of our viral system is that hTREX assembly on the viral mRNA is dependent upon an interaction with a virus-encoded protein , not splicing ., Specifically , ORF57 binds viral mRNA , directly interacts with and recruits Aly which in turn then interacts with and uses UAP56 to bridge an interaction with the hTHO-complex ., This ordered recruitment allows us to specifically disrupt the viral mRNA-ORF57-hTREX complex at different points and assess the functional significance on nuclear export ., Furthermore , rather than using an artificial in vitro assay to investigate the functional significance of hTREX , we assessed this in the context of the virus replication cycle using the 293T-BAC36 assay described above ., The trans-dominant mutant , pAlyΔC-myc , which has 20 residues deleted from the carboxy-terminus of Aly , is unable to interact with UAP56 42 ., We were interested in establishing if this mutant could be used to disrupt the assembly of UAP56 and hTHO-complex on an intronless viral mRNA and as such provide insights into whether these proteins are essential for nuclear export ., However , prior to its use in the replication assay it was essential to confirm that AlyΔC-myc is still recruited by ORF57 to intronless viral mRNA and is able to interact with TAP ., To this end , ORF57 , UAP56 and TAP were expressed as GST fusion proteins and incubated with either pmyc , pAly-myc or pAlyΔC-myc transfected cell lysates and pulldown analysis performed ., Western blotting using a myc-specific antibody demonstrated that Aly-myc interacted with ORF57 , TAP and UAP56 ., In contrast , AlyΔC-myc is unable to associate with UAP56 but retains the ability to interact with both ORF57 and TAP ( Fig . 6A ) ., These results suggest that AlyΔC-myc is an ideal mutant to inhibit the recruitment of UAP56 and hTHO-complex on the viral intronless mRNA ., However , one caveat to this system is that expression of pAlyΔC-myc may also act in a dominant negative capacity to inhibit spliced mRNA nuclear export 42 ., Therefore it was important to allow expression of the spliced ORF57 protein prior to accumulation of pAlyΔC-myc ., To this end , transient transfection of pAlyΔC-myc was performed concurrent with reactivation of the KSHV lytic replication cycle , and ORF57 protein levels assessed 24 h later ., Results show that comparable amounts of ORF57 were expressed in untransfected , pmyc , pAly-myc and pAlyΔC-myc transfected cell lysates ( Fig . 6B ) ., To test if AlyΔC-myc inhibited the recruitment of UAP56 and the hTHO-complex onto KSHV intronless RNAs , RNA-IPs were performed on reactivated KSHV-infected 293T cells transfected with pmyc , pAly-myc or pAlyΔC-myc ., We obtained similar results for pmyc and pAly-myc to those shown in Fig . 2A , where recruitment of hTREX components onto the viral RNA was readily detected 48 h-post reactivation ., However , RNA-IPs using cell extracts transfected with AlyΔC-myc showed a dramatic decrease in the recruitment of UAP56 and hHpr1 to viral mRNA ( Fig . 6C ) ., RNA-IPs performed using a TAP-specific antibody showed that TAP is recruited to the intronless viral mRNA , irrespective of Aly status ., Critically , RNA-IPs using an ORF57-specific antibody produced ORF47 RT-PCR products of a similar intensity , suggesting that ORF57 was not limiting in this assay ( Fig . 6C ) ., It should also be noted that we observed a decrease in TAP recruitment to the viral mRNA in the presence of both pAly-myc and pAlyΔC-myc , compared to pmyc control ., At present , we are unsure why TAP recruitment is reduced , however , no difference in mRNA nuclear export is observed between pmyc and pAly-myc transfected cells , suggesting that this reduction in TAP recruitment does not impede the nuclear export of intronless viral mRNAs ., To assess if the AlyΔC-myc mutant affected intronless viral mRNA export during replication , northern blot analysis was performed as described above ., Results demonstrated that ORF47 mRNA nuclear export is impaired in reactivated cells that expressed AlyΔC-myc , but not in cells expressing myc or Aly-myc ( Fig . 6D ) ., Moreover , to determine if expression of AlyΔC-myc had any effect on virus replication , the KSHV-latently infected 293T BAC36-GFP cell line was transfected with pmyc , pAly-myc or pAlyΔC-myc and concurrently reactivated using TPA and incubated for 72 hours ., The supernatants from each flask were then harvested and used to re-infect 293T cells ., The level of virus replication was determined by scoring the percentage of GFP positive cells 48 h post-infection , as previously described 41 ., Similar levels of lytic replication and virus production were observed from pmyc and pAly-myc pre-transfected cells ., Strikingly , virus production from pAlyΔC-myc pre-transfected cells was reduced by approximately 10 fold ( Fig . 6E ) ., These data demonstrate that ORF57-mediated recruitment of Aly and TAP to an intronless viral mRNA is insufficient for its nuclear export and that a lack of UAP56 and hTHO-complex on an intronless viral mRNA has a profound effect on intronless nuclear export and KSHV lytic replication ., Co-immunoprecipitation data show that ORF57 readily associates with components of hTREX , however , no such interaction was observed between ORF57 and the EJC proteins; eIF4A3 , Y14 and Magoh ., This result suggests that the EJC is not recruited to intronless viral transcripts and this was confirmed using RNA-IP assays ., In contrast , hTREX proteins readily precipitated with intronless viral mRNA , in the presence of ORF57 , which presumably functions as a linker between hTREX and the viral mRNA ., These findings suggest that the essential export adapter complex for intronless KSHV nuclear export is hTREX and not the EJC ., It should be noted that these findings are in contrast to previous observations using a homologue of KSHV ORF57 from the prototype γ-2 herpesvirus , Herpesvirus saimiri 29 ., One possible explanation for these contrasting data is that co-immunoprecipitations from Williams et al . were performed by over-expressing myc-tagged EJC components , whereas this in study , endogenous EJC proteins was precipitated using an eIF4A3- , Y14 and Magoh-specific antibodies ., To test this , we have performed co-immunopreciptations with EJC specific-antibodies using HVS-infected cell lysates ., No interactions were observed between HVS ORF57 and the endogenous EJC proteins ( Fig . S6 ) , suggesting the previously observed interactions may have been due to the overexpression of the EJC components ., In addition to splicing dependency , the cap-binding complex protein , CBP80 , is required to recruit hTREX to human pre-mRNA , via a direct interaction with Aly ., Interestingly , we detected a direct interaction between ORF57 and CBP80 , implying that the 5′ cap may also function in intronless KSHV mRNA export ., However , upon disrupting the ORF57 and Aly interaction ( via mutation of the PxxP motif ) , we also observed a reduction of the ORF57-CBP80 interaction ., Analysis suggests that this reduction maybe due to the loss of hTREX affecting the stability of the export competent viral RNP ., This suggests that although ORF57 interacts directly with Aly and CBP80 , these interactions may not overlap and more detailed analysis of the interacting domains for both proteins is required ., It is also worth noting however , that in the absence of ORF57 , CBP80 did not recruit Aly to the intronless viral transcripts , suggesting that ORF57 is essential for the loading of hTREX on viral mRNA ., The lack of EJC recruitment to intronless viral mRNA may have ramifications beyond those of nuclear export , for example , the EJC has been suggested to function in translational efficiency 16 , 43 ., Intriguingly , the herpes simplex virus type-1 ( HSV-1 ) ORF57 homologue , ICP27 , has been implicated in increased translation efficiency 44 , 45 , we are currently investigating whether ORF57 increases translation of KSHV transcripts during virus replication ., The current model for hTREX assembly on a spliced mRNA describes UAP56 and Aly associating with the mRNA in a 5′ cap- and splicing-dependent manner ., Moreover , as shown in Fig . 7A , it has been suggested that UAP56 may bridge an interaction between Aly and the hTHO-complex 9 , 42 ., In contrast , during KSHV replication hTREX appears to be tethered to an intronless KSHV mRNA via an exclusive interaction with ORF57 ., Taking advantage of this , we used the ORF57-hTREX complex to gain insight into how individual components of hTREX interact with one another ., Our data show that ORF57 interacts exclusively with Aly , which then binds directly to UAP56 and this in turn functions as a bridge to recruit hHpr1 and presumably the complete hTHO-complex ( Fig . 7B ) ., This order of hTREX assembly is in broad agreement with the model proposed by Cheng et al who showed using RNase H digestion analysis that Aly was the most 5′ of the hTREX components , with UAP56 and hTHO-complex binding further downstream ., Interestingly , the direct interaction observed between ORF57 and CBP80 suggest that ORF57 may recruit hTREX to the 5′ end of the intronless mRNA , perhaps to provide directionality to nuclear export as is the proposed case for spliced human mRNA 13 ., The functional significance of hTREX recruitment to intronless viral mRNA is substantiated using an ORF57 point mutant and a dominant-negative Aly mutant ., Specifically , we were able to disrupt the direct interaction between ORF57 and Aly by mutating two proline residues within a region of ORF57s Aly-binding domain 28 ., This ORF57Pmut was still able to recognise and bind intronless viral mRNA , however , it lacked the ability to recruit hTREX to these transcripts ., A failure to recruit hTREX rendered the ORF57Pmut non-functional as a viral mRNA export protein and provides direct evidence that the hTREX complex is essential for the efficient export of intronless viral mRNA and virus replication ., The export adapter Aly is able to interact directly with the export factor complex TAP/p15 46 , therefore , we were interested in assessing whether Aly-TAP/p15 recruitment produced an export-competent intronless viral mRNP or if UAP56 and the hTHO-complex were also required for nuclear export ., This is of particular importance as a number of ORF57 homologues , such as Herpes si

Introduction, Results, Discussion, Materials and Methods

A cellular pre-mRNA undergoes various post-transcriptional processing events , including capping , splicing and polyadenylation prior to nuclear export ., Splicing is particularly important for mRNA nuclear export as two distinct multi-protein complexes , known as human TREX ( hTREX ) and the exon-junction complex ( EJC ) , are recruited to the mRNA in a splicing-dependent manner ., In contrast , a number of Kaposis sarcoma–associated herpesvirus ( KSHV ) lytic mRNAs lack introns and are exported by the virus-encoded ORF57 protein ., Herein we show that ORF57 binds to intronless viral mRNAs and functions to recruit the complete hTREX complex , but not the EJC , in order assemble an export component viral ribonucleoprotein particle ( vRNP ) ., The formation of this vRNP is mediated by a direct interaction between ORF57 and the hTREX export adapter protein , Aly ., Aly in turn interacts directly with the DEAD-box protein UAP56 , which functions as a bridge to recruit the remaining hTREX proteins to the complex ., Moreover , we show that a point mutation in ORF57 which disrupts the ORF57-Aly interaction leads to a failure in the ORF57-mediated recruitment of the entire hTREX complex to the intronless viral mRNA and inhibits the mRNAs subsequent nuclear export and virus replication ., Furthermore , we have utilised a trans-dominant Aly mutant to prevent the assembly of the complete ORF57-hTREX complex; this results in a vRNP consisting of viral mRNA bound to ORF57 , Aly and the nuclear export factor , TAP ., Strikingly , although both the export adapter Aly and the export factor TAP were present on the viral mRNP , a dramatic decrease in intronless viral mRNA export and virus replication was observed in the absence of the remaining hTREX components ( UAP56 and hTHO-complex ) ., Together , these data provide the first direct evidence that the complete hTREX complex is essential for the export of KSHV intronless mRNAs and infectious virus production .

Following gene expression in the nucleus , newly transcribed messenger RNA ( mRNA ) is exported to the cytoplasm , where it is translated into protein ., In mammals the vast majority of mRNAs contain introns that must be removed by the spliceosome prior to nuclear export ., In addition to excising introns , splicing is also essential for the recruitment of a several protein complexes to mRNA , one example being the human transcription/export complex , which is required for mRNA export ., Herpesviruses , such as Kaposis sarcoma–associated herpesvirus , replicate by hijacking components of the host cells biological machinery , including those proteins necessary for mRNA export ., An intriguing caveat in herpesvirology is that herpesviruses , such as Kaposis sarcoma–associated herpesvirus , produce some mRNAs that lack introns and do not undergo splicing ., How then are these intronless mRNAs exported to the cytoplasm ?, The answer lies in a virus protein called ORF57 that is able to bind to the intronless mRNA and then export them to the cytoplasm ., ORF57 achieves this function by mimicking splicing and recruiting the human transcription/export complex to the intronless viral mRNA , thus facilitating its export into the cytoplasm .

molecular biology/mrna transport and localization, virology/viral replication and gene regulation, virology

journal.pgen.1002333

A Genome-Wide Screen for Interactions Reveals a New Locus on 4p15 Modifying the Effect of Waist-to-Hip Ratio on Total Cholesterol

Serum lipids are important determinants of cardiovascular disease and related morbidity 1 ., The heritability of circulating lipid levels is estimated to be 40%–60% and recent genome-wide association ( GWA ) studies implicated a total of 95 loci associated with serum high-density lipoprotein cholesterol ( HDL-C ) , low-density lipoprotein cholesterol ( LDL-C ) , total cholesterol ( TC ) , and triglyceride ( TG ) levels 2 ., Currently identified common variants explain 10%–12% of the total variation in lipid levels , corresponding to ∼25% of the trait heritability 2 ., Epidemiological risk factors , such as alcohol consumption , smoking , physical activity , diet and body composition are known to affect lipid levels 3–5 ., These risk factors also show moderate to high heritabilities , and over 120 loci with genome-wide significant association have been identified ( http://www . genome . gov/26525384 ) ., To better understand the biological processes modifying lipid levels , several twin studies 6–8 and candidate gene studies 9–14 have tested for interactions between genes and epidemiological risk factors ., Interactions between genes and modifiable risk factors might help us develop new lifestyle interventions targeted to susceptible individuals based on their genetic information ., The effects of genetic loci and risk factors have been studied widely separately , but to date no GWA studies for interactions on lipids have been reported ., We conducted a genome-wide screen for interactions between 2 . 5 million genetic markers and sex , lifestyle factors ( smoking and alcohol consumption ) , and body composition ( BMI and WHR ) in association to serum lipid levels ( TC , TG , HDL-C , and LDL-C ) in 18 population-based cohorts ( max N\u200a=\u200a32 , 225; Table S1A , Text S1 ) ., We defined interaction as a departure from a linear statistical model allowing for the additive main effects of both the SNP and the epidemiological risk factor ., 18 SNPs with suggestive interactions for at least one of the trait – epidemiological factor combinations ( P-value for the interaction <10−6 ) in stage 1 analyses were taken forward to stage 2 analysis in eight additional cohorts ( max N\u200a=\u200a14 , 889; Table S1B , Text S1 ) ., In inverse variance meta-analyses combining the results from stage 1 and stage 2 ( Table S2 ) , the interaction between rs6448771 in chromosome 4p15 and WHR on TC ( Figure 1 ) was statistically genome-wide significant ( stage 1 and 2 combined P\u200a=\u200a9 . 08×10−9 ) ., This interaction was tested in stage 3 in two further cohorts ( N\u200a=\u200a7 , 813; Table S1C , Text S1 ) , which showed an effect to the same direction ., After combining results from all three stages ( total N\u200a=\u200a43 , 903 ) , the P-value for interaction was 4 . 79×10−9 ., The association between WHR and TC was strongest for individuals carrying two G alleles of rs6448771 , for whom a one standard deviation ( sd ) difference in WHR corresponds to 0 . 19 sd difference ( confidence interval 0 . 13–0 . 25 ) in TC concentration , while for individuals homozygous for the A allele the difference was 0 . 12 sd ( confidence interval 0 . 09–0 . 16 ) ( Table S3A , Figure S1 ) ., The effect corresponds to 0 . 5% and 0 . 2% of the total variance explained in a cohort of young individuals ( YFS , mean age\u200a=\u200a37 . 6 ) and an old cohort ( HBCS , mean age\u200a=\u200a61 . 49 ) , respectively ., Additionally , when looking at the effect of the SNP on TC in WHR tertiles , the estimates differed in a way that the estimated SNP effect is higher for the individuals with higher WHR ( Table S3B ) ., The SNP did not have a direct effect on either TC or WHR ( P\u200a=\u200a0 . 46 and P\u200a=\u200a0 . 51 , respectively , Figure 1 ) ., The SNP rs6448771 is located 249 kb downstream of the protocadherin 7 ( PCDH7 ) gene ., Since the polymorphisms associated with complex phenotypes often influence gene expression , we examined whether individuals carrying different genotypes of rs6448771 have variation in their transcript profiles ., As WHR reflects adipose tissue function , we selected 54 individuals from Finnish dyslipidemic families with available fat biopsies and GWA data ., We used linear regression to find genes that were differentially expressed in adipose tissue depending on the rs6448771 genotype ., We found two potential candidate genes with nominally significant cis-eQTL effects , PCDH7 ( P\u200a=\u200a0 . 027 , distance from the rs6448771 250 kb ) and CCKAR ( P\u200a=\u200a0 . 017 , distance from the SNP 4 . 9 Mb ) ., The region with CCKAR has previously been linked with obesity 15 ., Additionally , using Ingenuity software ( IPA ) , we conducted a pathway analysis for genes with eQTL P-value<0 . 01 ( both trans- and cis-eQTLs ) ., Among other diverse IPA-defined biological functions , there was an eQTL association enrichment among genes belonging to the ‘degradation of phosphatidylcholine’ ( 3 genes out of 6 , P\u200a=\u200a6 . 64×10−5 , Benjamini-Hochberg corrected P\u200a=\u200a0 . 0138 ) and ‘degradation of phosphatidic acid’ ( 4 genes out of 8 , P\u200a=\u200a4 . 71×10−4 , B-H corrected P\u200a=\u200a0 . 0349 ) functions , which are members of broader defined IPA categories “Lipid Metabolism” and “Carbohydrate Metabolism” ., These pathways were up-regulated in individuals carrying the G allele of rs6448771 , possibly indicating a role for rs6448771 in lipid and carbohydrate metabolism ., The associated SNP also shows evidence for interactions with WHR on LDL-C ( effect estimate for the interaction\u200a=\u200a0 . 03 , P\u200a=\u200a0 . 0016 ) and HDL-C ( effect estimate\u200a=\u200a0 . 02 , P\u200a=\u200a0 . 029 ) in our stage 1 meta-analysis and after adjusting for TC no residual interaction effect on LDL-C and a little on HDL-C remains ( P\u200a=\u200a0 . 834 and P\u200a=\u200a0 . 131 respectively ) when testing in data subset ., Therefore we tested the SNP – WHR interaction also on a range of lipoprotein subclasses measured using NMR metabonomics platform 16 available in two cohorts ( NFBC1966 , N\u200a=\u200a4624 mean age\u200a=\u200a31 . 0; YFS , N\u200a=\u200a1889 , mean age\u200a=\u200a37 . 6 ) ., The results show that the SNP has a positive interaction effect on large HDL particle concentration ( combined effect for the interaction\u200a=\u200a0 . 538 , P\u200a=\u200a0 . 0186 ) and a negative effect on large very-low-density lipoprotein ( VLDL ) particles ( combined effect\u200a=\u200a−0 . 466 , P\u200a=\u200a0 . 0291 ) and total triglycerides ( combined effect\u200a=\u200a−0 . 454 , P\u200a=\u200a0 . 0343 ) ( Figure 2 ) ., Our genome-wide scan for interactions between SNP markers and traditional epidemiological risk factors in population-based random samples found a genome-wide significant locus , rs6448771 , modifying the relationship between WHR and TC ., The effect of WHR is estimated to be 64% stronger for individuals carrying two copies of the G allele than for individuals carrying two A alleles ., The interaction explains around half a percent of the TC variance that is in par with the main effects of the strongest previously identified TC SNPs individually ., This SNP also shows similar interaction effects on a cascade of more detailed lipid fractions suggesting broad involvement in lipid metabolism , which was also suggested by our eQTL association enrichment analysis with adipose tissue expression data ., The eQTL analysis pointed towards two potential candidate genes in the region ., The first one of these was protocadherin 7 ( PCDH7 ) gene , which produces a protein that is thought to function in cell-cell recognition and adhesion ., The other candidate gene , cholecystokinin A receptor ( CCKAR ) regulates satiety and release of beta-endorphin and dopamine in the central and peripheral nervous system ., It has been previously shown that rats with no expressed CCKARs developed obesity , hyperglycemia and type 2 diabetes 17 ., To test whether our eQTL finding was adipose tissue specific , we ran the eQTL analysis for PCDH7 and CCKAR in another dataset with genome wide expression data from blood leukocytes ( N\u200a=\u200a518 ) available ., CCKAR could not be tested due to its negligible expression in blood leukocytes , and no association was found for the PCDH7 ( P-value\u200a=\u200a0 . 284 ) gene most likely indicating an adipose tissue specific eQTL for PCDH7 as a function of rs6448771 ., One interesting aspect of this study , given our large sample size , is that only one signal achieved genome-wide significance , where previously published lipid GWA studies have found close to a hundred ., Although power to detect interaction is typically lower than for main effects , especially for rare exposures and SNPs , several of the exposures considered here ( such as WHR , BMI , and gender ) were common and available for a large proportion of the study sample ., This suggests that the contribution of two-way G×E interactions to lipid levels , at least for the risk factors we examined , is rather small , or that our current measures of risk factors may not be robust enough for identifying interactions ., More specific measures of both phenotypes and interacting risk factors would give better statistical power in future screens of G×E interactions ., Our findings allow us to draw several conclusions ., First , to our knowledge , this is the first time an interaction between a genetic loci and a risk factor has been identified in a genome-wide scan using a stringent statistical threshold for genome-wide significance ., Second , in our samples , rs6448771 modified the relationship between WHR and TC , but was not associated with either WHR or TC alone ., This observation suggests that genome-wide screens for interactions may be complementary to the current large-scale GWAS efforts for finding main effects ., Third , in addition to careful harmonization of both risk factor data and phenotypes , large sample sizes are needed to identify interactions ., In our study , 43 , 903 samples were combined to robustly identify the interaction ., Our data , however , suggest that the contribution of G×E interaction using current phenotypes appears limited ., Finally , from clinical point of view , the interaction may open up possibilities for targeted intervention strategies for people characterized by specific genomic profiles but more refined measures of both body-fat distribution and metabolic measures are needed to understand how their joint dynamics are modified by the newly found locus ., 18 studies , with a combined sample size of over 30 , 000 individuals , participated in the discovery phase of this analysis; 8 studies were available for replication with over 14 , 000 individuals ., In the discovery stage , only population-based cohorts not ascertained on the basis of phenotype , with a wide variety of well-defined epidemiological measures available , were included ., In the replication datasets , the NTR cohort was selected on the basis of low risk for depression and the Genmets samples were selected for metabolic syndrome ., In further replication of rs6448771 , the EPIC cases were ascertained by BMI ., Descriptive statistics for these populations are detailed in Table S1A ( discovery ) , S1B ( replication ) and S1C ( further replication ) ., Brief descriptions of the cohorts are provided in the Text S1 section “Short descriptions of the cohorts” ., Individuals were excluded from analysis if they were not of European descent or were receiving lipid-lowering medication at the time of sampling ., TC , HDL-C , and TG concentrations were measured from serum or plasma extracted from whole blood , typically using standard enzymatic methods ., LDL-C was either directly measured or estimated using the Friedewald Equation ( LDL-C\u200a=\u200aTC – HDL-C – 0 . 45×TG for individuals with TG≤4 . 52 mmol/l , samples with TG level higher than 4 . 52 were discarded in the calculation of LDL-C ) 18 ., Covariates and epidemiological risk factors were ascertained at the same time that blood was drawn for lipid measurements ., BMI was defined as weight in kilograms divided by the square of height in meters ., Waist circumference was measured at the mid-point between the lower border of the ribs and the iliac crest; hip circumference was measured at the widest point over the buttocks ., Waist-to-hip ratio was defined as the ratio of waist and hip circumferences ., Alcohol consumption and smoking habits were determined via interviews and/or questionnaires ., Both behaviors were coded as dichotomous ( abbreviations: ALC for drinker/abstainer and SMO for current smoker/current non-smoker ) and semi-quantitative traits ., Semi-quantitative alcohol usage ( ALCq ) was based on daily consumption in grams ( 0: 0 g/day; 1: >0 and ≤10 g/day; 2: >10 and ≤20 g/day; 3: >20 and ≤40 g/day; 4: >40 g/day ) ., Semi-quantitative smoking ( SMOq ) was assessed based on the number of cigarettes per day ( 0: 0 cigarettes/day; 1: >0 and ≤10 cigarettes/day; 2: >10 and ≤20 cigarettes/day; 3: >20 and ≤30 cigarettes/day; 4: >30 cigarettes/day ) ., Affymetrix , Illumina or Perlegen arrays were used for genotyping in the discovery cohorts ., Each study filtered both individuals and SNPs to ensure robustness for genetic analysis ., After quality control , these data were used to impute genotypes for approximately 2 . 5 million autosomal SNPs based on the LD patterns observed in the HapMap 2 CEU samples ., Imputed genotypes were coded as dosages , fractional values between 0 and 2 reflecting the estimated number of copies of a given allele for a given SNP for each individual ., Cohort specific details concerning quality control filters , imputation reference sets and imputation software are described in Table S4 ., Replication cohorts utilized genome-wide imputed data , as described above , where available ., Details on the genotyping methods implemented in the replication samples are described in Table S4 ., Proton NMR spectroscopy was used to measure lipid , lipoprotein subclass and particle concentrations in native serum samples ., NMR methods have been previously described in detail 16 , 19 ., Serum concentrations of total triglycerides ( TG ) , total cholesterol ( TC ) together with LDL-C and HDL-C were determined ., In addition , total lipid and particle concentrations in 14 lipoprotein subclasses were measured ., The measurements of these subclasses have been validated against high-performance liquid chromatography 20 ., The subclasses were as follows: chylomicrons and largest VLDL particles ( particle diameters from approx 75 nm upwards ) , five different VLDL subclasses: very large VLDL ( average particle diameter 64 . 0 nm ) , large VLDL ( 53 . 6 nm ) , medium-size VLDL ( 44 . 5 nm ) , small VLDL ( 36 . 8 nm ) , and very small VLDL ( 31 . 3 nm ) ; intermediate-density lipoprotein ( IDL ) ( 28 . 6 nm ) ; three LDL subclasses: large LDL ( 25 . 5 nm ) , medium-size LDL ( 23 . 0 nm ) , and small LDL ( 18 . 7 nm ) ; and four HDL subclasses: very large HDL ( 14 . 3 nm ) , large HDL ( 12 . 1 nm ) , medium size HDL ( 10 . 9 nm ) , and small HDL ( 8 . 7 nm ) ., Triglyceride concentrations were natural log transformed prior to analysis ., BMI and WHR were transformed to normality using inverse-normal transformation of ranks ., For analyses where sex was the epidemiological variable of interest , the phenotypes were defined as the rank-inverse normal transformed residuals resulting from the regression of the lipid measurement on age and age2 ., For the other analyses , the phenotypes were defined as the inverse normal transformed residuals resulting from the regression of the lipid measurement on age , age2 , and sex ., Associations between the transformed residuals and epidemiological risk factors/SNPs were tested using linear regression models under the assumption of an additive ( allelic trend ) model of genotypic effect ., The models regressed phenotypes on epidemiological factor , SNP , and epidemiological factor×SNP termsand tested if the effect for E×SNP was 0 using 1 df Wald tests ., In family-based cohorts , linear mixed modeling was implemented to control for relatedness among samples 21 ., Analysis software used by the individual cohorts is described in Table S1A and S1B ., The interaction terms from the regression analyses were meta-analyzed using inverse variance weighted fixed-effects models 22 ., Prior to meta-analysis , genomic control correction factors ( λGC ) 23 , calculated from all imputed SNPs , were applied on a per-study basis to correct for residual bias possibly caused by population sub-structure ., Meta-analyses were performed by two independent analysts using METAL ( http://www . sph . umich . edu/csg/abecasis/Metal/index . html ) and the R 24 package MetABEL ( part of the GenABEL suite , http://www . genabel . org/ ) ., All results were concordant , reflecting a robust analysis ., Results were selected for in silico replication if the meta-analysis P-value was less than 10−6 ., Results passing the threshold of suggestive genome-wide association ( P-value ≤5×10−7 ) were selected for further replication by direct genotyping ., The commonly accepted genome wide level of significance ( 5×10−8 ) reflects the estimated testing burden of one million independent SNPs in samples of European ancestry 25 ., To address the multiple testing arising from testing interactions with multiple risk factors , we set the genome wide significance threshold to 5×10−8/3\u200a=\u200a1 . 67×10−8 corresponding to three principal components explaining 97 . 8% of the total variation of the risk factors ( Table S5 ) .

Introduction, Results, Discussion, Materials and Methods

Recent genome-wide association ( GWA ) studies described 95 loci controlling serum lipid levels ., These common variants explain ∼25% of the heritability of the phenotypes ., To date , no unbiased screen for gene–environment interactions for circulating lipids has been reported ., We screened for variants that modify the relationship between known epidemiological risk factors and circulating lipid levels in a meta-analysis of genome-wide association ( GWA ) data from 18 population-based cohorts with European ancestry ( maximum N\u200a=\u200a32 , 225 ) ., We collected 8 further cohorts ( N\u200a=\u200a17 , 102 ) for replication , and rs6448771 on 4p15 demonstrated genome-wide significant interaction with waist-to-hip-ratio ( WHR ) on total cholesterol ( TC ) with a combined P-value of 4 . 79×10−9 ., There were two potential candidate genes in the region , PCDH7 and CCKAR , with differential expression levels for rs6448771 genotypes in adipose tissue ., The effect of WHR on TC was strongest for individuals carrying two copies of G allele , for whom a one standard deviation ( sd ) difference in WHR corresponds to 0 . 19 sd difference in TC concentration , while for A allele homozygous the difference was 0 . 12 sd ., Our findings may open up possibilities for targeted intervention strategies for people characterized by specific genomic profiles ., However , more refined measures of both body-fat distribution and metabolic measures are needed to understand how their joint dynamics are modified by the newly found locus .

Circulating serum lipids contribute greatly to the global health by affecting the risk for cardiovascular diseases ., Serum lipid levels are partly inherited , and already 95 loci affecting high- and low-density lipoprotein cholesterol , total cholesterol , and triglycerides have been found ., Serum lipids are also known to be affected by multiple epidemiological risk factors like body composition , lifestyle , and sex ., It has been hypothesized that there are loci modifying the effects between risk factors and serum lipids , but to date only candidate gene studies for interactions have been reported ., We conducted a genome-wide screen with meta-analysis approach to identify loci having interactions with epidemiological risk factors on serum lipids with over 30 , 000 population-based samples ., When combining results from our initial datasets and 8 additional replication cohorts ( maximum N\u200a=\u200a17 , 102 ) , we found a genome-wide significant locus in chromosome 4p15 with a joint P-value of 4 . 79×10−9 modifying the effect of waist-to-hip ratio on total cholesterol ., In the area surrounding this genetic variant , there were two genes having association between the genotypes and the gene expression in adipose tissue , and we also found enrichment of association in genes belonging to lipid metabolism related functions .

genome-wide association studies, genetics, biology, human genetics, genetics and genomics

journal.pgen.1003306

Mechanisms Involved in the Functional Divergence of Duplicated GroEL Chaperonins in Myxococcus xanthus DK1622

Chaperonins are essential cellular components that are responsible for protein folding , assembly and transport 1–6 ., Chaperonins are also a major group of heat shock proteins that are over-expressed at high temperatures and have fundamental roles in growth and survival at non-permissive temperatures 6–8 ., GroEL is a type I chaperonin , and in Escherichia coli , the GroEL chaperonin is required in vivo for the proper folding , at all temperatures , of approximately 300 newly translated polypeptides ( accounting for approximately 10% of the total ) that participate in various physiological processes 9 ., Because of its importance in many cellular processes , the groEL gene is ubiquitously distributed in bacteria ., Most bacterial species , such as E . coli , possess a single groEL gene , whereas other species ( nearly 30% of bacteria with sequenced genomes ) have evolved multiple groEL copies 1 ., The paralogous GroEL proteins are highly similar in sequence and , most likely , in structure ., However , some differences exist between duplicated groEL genes , and these duplicated GroEL proteins have evolved to play divergent roles in many different cellular processes in different bacterial species 10–15 ., Although the mechanisms of functional divergence are important for our understanding of the complexity of evolution , these mechanisms have not been characterized to date ., Myxobacteria are δ-proteobacteria with unique and complex multicellular behaviors , such as movement in swarms on solid surfaces , cooperative feeding on macromolecules or other microbial cells and the development of multicellular fruiting bodies containing numerous myxospores against adversity conditions 16 , 17 ., Myxococcus xanthus DK1622 is a model myxobacterium with a large genome ( 9 . 14 Mbp ) that includes many duplicated crucial genes 18 ., It has been suggested that such duplication is responsible for the complex social behavior of these cells , although this hypothesis has not been experimentally validated ., There are two copies of the groEL gene in the genome of M . xanthus DK1622 ., Previous studies indicate that either of the two paralogous groEL genes can be deleted in strain DK1622 without affecting cellular viability , although the deletion does result in distinct defects in the cellular heat-shock response , predation and development 15 ., In this study , we investigated the effects of the substrate selectivity of the GroEL proteins and the expression levels of the duplicated groEL genes on the functional divergence of heat-shock responses , development and predation ., We performed a comparative proteomics analysis of the substrate specificity of the two GroELs , and the relationships between the structural differences and substrate specificity were investigated using bioinformatics , molecular swapping and site-directed mutagenesis ., Otani et al . found that , although GroEL1 and GroEL2 are among the major proteins induced by heat shock , the density of GroEL2 spots in two-dimensional electrophoretic gels is much lower than that of GroEL1 19 ., It was also noted that the expression levels of groEL1 and groEL2 were not equal in the wild-type strain DK1622 in CTT growth medium and TPM development medium and that the two groEL genes played distinct roles in heat-shock responses , development and predation 15 ., To assess the changes in groEL1 or groEL2 expression in groEL-deletion mutants and whether these changes contribute to functional divergence , we inserted groEL1 or groEL2 , each with its own promoter , into the genome of groEL1- or groEL2-deletion mutants at the attB integration site using pSWU30 , producing four groEL-complemented strains ( Table S1 ) ., The groEL expression levels and cell viability were compared between these mutants and the wild-type strain DK1622 following heat shock at 42°C for 30 min ., Quantitative PCR assays indicated that the expression of the groEL genes was regulated in a complex manner in different mutants upon heat shock ( Figure 1 ) ., In the wild-type strain DK1622 , the groEL2 expression level was only one-quarter of that of groEL1 after heat shock ., The deletion of groEL1 ( strain YL0301 ) led to an increase in the groEL2 expression level ( approximately twofold ) ., The expression of groEL1 inserted in YL0301 ( strain YL0901 ) was approximately half that in DK1622 under the heat shock conditions , but the presence of exogenous groEL1 had no obvious effect on the expression level of groEL2 ( P>0 . 05 ) ., Thus , the total expression of all the groEL genes in YL0901 was similar to that in DK1622 ., In YL0902 , which contained an additional groEL2 gene , the total expression of groEL2 also doubled , reaching a level equal to four times that of groEL2 in DK1622 ., In YL0302 , the deletion of the groEL2 gene led to reduced expression of the groEL1 gene under the heat shock conditions ( approximately 60% of that in DK1622 ) ., Transformation of the YL0302 mutant ( strain YL0906 ) with another groEL1 gene increased the total groEL expression level to that of DK1622 ( P>0 . 05 ) ., The total expression level of groEL1 and groEL2 in the groEL2-complemented YL0302 mutant ( strain YL0907 ) also reached the level in DK1622 ( P>0 . 05 ) ., Interestingly , although YL0902 had two groEL2 genes and no groEL1 , the survival rates of both YL0901 and YL0902 were similarly increased after heat shock at 42°C for 30 min , paralleling the increase in the total groEL expression level in these two mutants ( Figure 1 ) ., The survival rates of the YL0906 and YL0907 mutants after the heat shock treatment also corresponded to an increase in the total groEL expression level ., These results suggest that the lethal nature of the heat shock in the groEL1 deletion mutant ( YL0301 ) and the increased sensitivity of the groEL2 deletion mutant ( YL0302 ) to high temperatures 15 result from a significant decrease in the total expression of GroEL , leading to an insufficient level of GroEL proteins to facilitate the refolding of denatured proteins ., This is consistent with a model where there is a threshold level of GroEL beneath which cells cannot survive ., When the total expression of groEL is higher than the threshold , there is a positive correlation between the survival rate and groEL expression in M . xanthus DK1622 cells after heat shock ( r\u200a=\u200a0 . 98 , P<0 . 01 ) ( Figure S1 ) , a result which supports the hypothesis that , after duplication , both groEL1 and groEL2 retain fundamental functions by balancing their expression dosage 20 , 21 ., Because the deletion of groEL1 and the deletion of groEL2 result in deficiencies in development and predation , respectively 15 , we performed a development assay on DK1622 , YL0301 , YL0901 and YL0902 and predation assays on DK1622 , YL0302 , YL0906 and YL0907 ., Figure 2A shows the expression levels of groEL1 and groEL2 in different strains after 12 , 36 and 60 h of incubation on TPM development medium ., When groEL1 was inserted into the genome of the groEL1-deletion mutant ( strain YL0901 ) , the developmental defect was mostly reversed , with sporulation reaching 70%–80% of that of DK1622 ., However , although YL0902 ( containing two copies of groEL2 ) had a total groEL expression level similar to that of YL0901 at different developmental stages , YL0902 displayed a development defect similar to that of YL0301 , and the sporulation ability of YL0902 was only approximately 20% of that of the wild-type strain DK1622 ( Figure 2B and 2C ) ., The insertion of groEL1 into YL0302 ( strain YL0906 ) did not improve the predation feeding ability of cells on an E . coli mat , and the swarming time of YL0906 to the edge of the E . coli colony was 60–65 h , which is similar to that of YL0302 ( P>0 . 05 ) ., When groEL2 was inserted into YL0302 ( strain YL0907 ) , the swarming time to the E . coli colony edge decreased to 40 h ( Figure 3 ) ., Because the presence of living E . coli cells in the E . coli predation experiments might affect the qPCR assay , we instead performed the analysis using a liquid feeding assay with casein as the only nutrient 15 ., The total expression level of groEL1 and groEL2 was also similar in the YL0906 and YL0907 mutants , suggesting that the changes in expression are not the major contributors to functional divergence ( Figure S2 ) ., The above results indicate that although the distinct heat-shock responses of the groEL1 and groEL2 mutants were determined by the total groEL expression level , the divergent functions of groEL1 and groEL2 in development and predation are the result of the substrate specificity of the corresponding GroEL chaperonins ., To explore the evolutionary relationships of groEL , we compared the M . xanthus GroEL sequences with those of ten genome-sequenced bacteria , including three Myxobacteria , three Actinobacteria , three Cyanobacteria , and E . coli ( Figure 4 and Table S2 ) ., With the exception of E . coli , these species possess duplicated groEL genes ., The maximum likelihood tree showed that the GroELs from Myxobacteria , Actinobacteria and Cyanobacteria clustered separately ( Figure 4A ) , suggesting that the groEL gene duplication originated from three independent evolutionary events in these three taxa ., We further calculated the Ka/Ks values of these orthologous groEL genes ( Figure 4B and Table S3 ) ., The average Ka/Ks values for Actinobacteria groEL2 and Cyanobacteria groEL1 were less than 0 . 1 , suggesting that they are highly evolutionarily conserved ., It has been reported that groEL2 in the three Actinobacteria species 12 , 22–24 and groEL1 in the three Cyanobacteria species 25–27 are housekeeping genes , which is consistent with their Ka/Ks values ., In M . xanthus , the Ka/Ks value for groEL1 was significantly lower than that for groEL2 ( P<0 . 05 ) but significantly higher than that for the housekeeping groEL genes in Actinobacteria or Cyanobacteria ( P<0 . 01 ) ., These results suggest that both of groEL1 and groEL2 in M . xanthus are suffered weak selection pressure , consistent to the finding that the deletion of either gene alone does not affect cell viability 15 ., Based on their structural characteristics and sequence conservation , the GroEL protein sequences have been divided into five regions ,, i . e ., , an N-terminal equatorial region , an N-intermediate region , an apical region , a C-intermediate region and a C-terminal equatorial region ( Figure S3 ) 28 ., The two intermediate regions have the highest level of conservation between M . xanthus DK1622 GroEL1 and GroEL2 ,, i . e ., , 97 . 7% and 97 . 2% identities for the N- and C-intermediate regions , respectively ., The identities for the other three regions are 81% for the N-terminal equatorial region , 75 . 4% for the apical region , and 62 . 6% for the C-terminal equatorial region ., Further Ka/Ks analysis showed similar sequence characteristics for the five GroEL1 and GroEL2 regions in the four Myxobacterial species referred to above ( Figure 4C ) ., For example , the Ka/Ks values of the N- and C-intermediate regions were very low ( <0 . 05 ) , suggesting that these two regions are highly conserved; in contrast , the other three regions had higher Ka/Ks values ( >0 . 3 ) , suggesting these regions are most likely involved in the functional divergence of GroEL1 and GroEL2 ., A sequence alignment showed that the high Ka/Ks values of the C-terminal equatorial regions were largely due to the variability of the C-terminal tail sequences ., For example , the C-terminal tail of GroEL1 in M . xanthus was composed of six repeated GGM motifs , similar to that of E . coli GroEL , whereas the C-terminal tail of GroEL2 is greatly different ., It was also noted that there are substantial differences in the C-terminal sequences between the duplicated GroELs ( Figure 4A ) 1 ., To clarify the relationships between the structural and functional divergence , we designed a series of single region-swaps between the groEL1 and groEL2 genes to determine the roles of the regions and their contributions to functional divergence ., The swapped regions included the N-terminal equatorial , apical , and C-terminal equatorial regions between GroEL1 and GroEL2; the two highly similar intermediate regions were not included ., The groEL2 hybrids containing the N-terminal equatorial , apical or C-terminal equatorial region of groEL1 were inserted into the genome of the groEL1-deletion mutant YL0301 using pSWU30 , producing the region-swapped strains YL0903 , YL0904 and YL0905 , respectively ., Similarly , the groEL1 hybrids with a swapped N-terminal equatorial , apical or C-terminal equatorial region of groEL2 were inserted into the genome of the groEL2-deletion mutant YL0302 to produce the mutant strains YL0908 , YL0909 and YL0910 , respectively ( Figure 5A ) ., Because the groEL2 mutant displays defective cellular predation and the groEL1 mutant displays deficient development and sporulation 15 , region swapping was performed in YL0301 using the groEL1 chimeras and YL0302 using the groEL2 chimeras ., Detailed descriptions of these mutants are listed in Tables S1 and S4 and Figure S4 ., The development and predation phenotypes of the region-swapped mutants were assayed using the intact groEL1- and groEL2-complemented mutants as controls ., The results showed that the developmental defect of the groEL1-deletion mutant was not reversed by GroEL2-equatorial-NGroEL1 ( YL0903 ) ., The sporulation ability of YL0903 was approximately 20% of that of DK1622 , which was the same as that of YL0301 ., However , the fruiting bodies of YL0904 ( YL0301 complemented with GroEL2-apicalGroEL1 ) were more similar to the fruiting bodies of the wild-type strain DK1622 than to the fruiting bodies of YL0301 , and the sporulation ability also increased to 55%–65% of that of DK1622 ., The sporulation of the strain complemented with GroEL2-equatorial-CGroEL1 ( YL0905 ) was 30%–40% of that of DK1622 ( Figure 5B ) ., In the predation experiments , the single swapped region in YL0908 did not noticeably improve the predation defects , which were similar to those of YL0906 ( P>0 . 05 ) ., However , the YL0909 strain significantly recovered its predation ability , which was similar to that of YL0907 ., These two mutants spread to the edge of the E . coli colonies within 40–45, h . The predation defect was also improved to some extent in YL0910 , which required 55–60 h to reach the edge of the E . coli mat ( Figure 5C ) ., Accordingly , the apical region and the C-terminal equatorial region determine the substrate preference , thus causing the functional divergence of the duplicated chaperonins with respect to development and predation; conversely , the N-terminal equatorial region has almost no effect ., In addition , we deleted the repeated GGM region ( GGMGGMGGMGGMGGMGM ) from GroEL1 in M . xanthus DK1622 , producing the YL1001 mutant ( Figure 5A ) ., Compared with the wild-type DK1622 , the mutant was markedly defective in development , and the sporulation ability of YL1001 was only 32 . 6% of that of DK1622 ( Figure 5D ) ., Furthermore , we swapped three of the six GGM repeats with YGGDDMDY in DK1622 , the corresponding sequence in GroEL2 , producing the YL1002 mutant ., Similar to YL1001 , YL1002 was also defective in development and exhibited a decreased sporulation ability ( 38 . 1% of that of DK1622 ) ( Figure 5D ) ., These results demonstrate that the GGM repeated region is necessary for GroEL1 to perform its normal functions in development ., To identify the proteins that interact with GroEL1 and GroEL2 in M . xanthus DK1622 , immunoprecipitation assays were performed using the groEL1- and groEL2-deletion mutants ( strains YL0301 and YL0302 ) , and the bound proteins were subjected to mass spectral identification ., Most of the non-specific substrates identified using two negative controls ( see Methods ) were ribosomal proteins ( Table S5 ) ., This result was consistent with the results for E . coli 9 ., After removing the non-specifically bound proteins , 151 and 114 proteins were found to be bound by GroEL1 and GroEL2 , respectively ., Of the bound proteins , 68 were bound to both GroEL1 and GroEL2 ( GroEL1/2 ) , whereas 83 and 46 proteins bound exclusively to GroEL1 and GroEL2 , respectively ( Table S5 ) ., Of the functionally annotated GroEL1/2 substrates ( 58/68 , 85 . 3% ) , many had functions or predicted functions related to fundamental physiological cellular processes; examples of such substrates are succinyl coenzyme A synthetase and isocitrate dehydrogenase , two key enzymes of the citric acid cycle 29 , 30 ., This result is consistent with the fact that either the groEL1 or groEL2 gene could be deleted without affecting cellular growth but that the double deletion of groEL1 and groEL2 resulted in inviable cells 15 ., However , except for PilA , no proteins involved in M . xanthus social behavior were found to bind to both GroEL1 and GroEL2 ., In contrast , of those annotated proteins that were exclusively bound by GroEL1 or GroEL2 ( accounting for 75 . 9% and 76 . 1% of bound proteins , respectively ) , a considerable number are involved in the social behaviors of M . xanthus DK1622 ( Table S5 ) ., For example , the frz signal transduction system is well known to play important roles in development process of M . xanthus DK1622 31 ., The frizzy aggregation protein FrzCD 31 is in the substrate list of GroEL1 ., Besides , Flp pilus assembly protein CpaB 18 , sensor histidine kinase/response regulator CheA4 32 and Type IV pilus secretin PilQ 33 were found to be specific substrates of GroEL1 , whereas type IV pilus assembly ATPase PilB 34 , gliding motility protein MglA 35 , type IV pilus biogenesis protein PilM 36 and several proteins related to the biosynthesis of secondary metabolites were found to be exclusively bound by GroEL2 ., These results are consistent with the hypothesis that GroEL is an essential component and that the duplicated groEL genes evolved to participate in various complex physiological processes in Myxococcus cells ., The structural characteristics of the substrate proteins were further analyzed by comparing their secondary structures with the known protein domain classification database CATH 37 ., After excluding the proteins that had low E-values ( >0 . 001 ) , we obtained 36 reliable secondary structures for the 68 identified GroEL1/2 substrates , 38 for the 83 GroEL1-specific substrates and 34 for the 46 GroEL2-specific substrates ., It is known that proteins with β-sheets exposed to the hydrophobic surface and packed with the hydrophobic surfaces of α-helices ( called the αβ domain ) have high-affinity interactions with the apical region of GroEL and are normally present as substrates of GroELs 38 ., As expected , most GroEL1/2 substrates ( 34 of 36 ) contain at least one αβ domain ., It is interesting that , although 31 of the 34 ( 91 . 18% ) GroEL2-specific substrates possess at least one αβ domain , only 27 of the 38 ( 71 . 05% ) GroEL1-specific substrates contain an αβ domain ( Figure 6A , Table S5 ) ., Another interesting difference is the difference in the molecular sizes of the GroEL1 and GroEL2 substrates ., According to the current model of GroEL 28 , 39–45 , substrate selection is heavily dependent upon the adaptation of a substrate molecule to the cavity volume of the GroEL chaperonin , which may change in response to GroEL sequence changes ., Previous studies have shown that GroEL strongly prefers to act on proteins with a molecular weight ranging from 20 kDa to 60 kDa 46 ., The average molecular weight of GroEL1-specific substrates was significantly smaller than that of GroEL2-specific substrates ( Figure 6B; P<0 . 05 ) ., For example , while 51 . 8% ( 43 of 83 ) of the GroEL1-specific substrate proteins were less than 40 kDa , the molecular weights of only 21 . 7% ( 10 of 46 ) of the GroEL2-specific proteins were less than 40 kDa ., A third important characteristic is the pI value of the substrate; there was no significant difference between the GroEL1 and GroEL2 substrates with respect to pI ( Figure 6B; P>0 . 05 ) ., Duplication is a major source of new genes and is equally important in Bacteria , Archaea and Eukarya 47 , 48 ., The duplication of the GroEL chaperonin gene has occurred in many different bacterial cells as part of the evolution of complexity 49 ., M . xanthus DK1622 is well known for its complex multicellular behaviors 16 , 17 , and this strain possesses a large genome ( 9 . 14 Mb ) in which there are many duplicated genes , including two copies of groEL 18 ., In addition to participating in fundamental processes involved in cellular growth , the two duplicated groEL genes have been demonstrated to play distinct roles in heat-shock responses , development and predation in DK1622 15 ., The results described in this report show that the divergent functions of GroEL1 and GroEL2 in various physiological processes result from different mechanisms ., The groEL expression level is the key reason for the difference in the heat-shock response after the deletion of groEL1 or groEL2 , suggesting that the duplicated groEL genes in Myxococcus have similar functions in cell survival ., These functions are most likely similar to their fundamental function in cell growth at normal temperatures ., Either of the two groEL genes can be deleted without significantly affecting cell growth , but at least one groEL gene is required for cell survival 15 ., In contrast , the functional divergence of the duplicated GroELs with respect to their roles in development and predation processes reflects in their substrate specificity , which has been suggested to evolutionarily relate to the unusual social behavior of Myxococcus ., The co-substrates of GroEL1 and GroEL2 have consistently been shown to be essential cellular components , but the duplicated GroEL chaperonins have also evolved their own substrate preferences related to late-appearing cellular processes , such as social behaviors and PKS/NRPS biosynthesis ., The evolutionary models for the functional divergence of duplicated genes are likely to include neofunctionalization , subfunctionalization , or a combination of thereof 20 , 47 , 50 ., Thus , the functional divergence of the duplicated groELs in M . xanthus is likely a combination of neofunctionalization and subfunctionalization , i . e . , the subneofunctionalization model 50 ., Although extensive studies have demonstrated that the duplicated groEL genes play distinct roles in different cellular physiological processes 10–15 , an understanding of the mechanisms involved in their functional divergence will provide insight into bacterial evolution ., The GroEL proteins have been divided into five regions based on structural characteristics and sequence conservation 28 ., Previous studies have shown that GroEL chimeras bearing equatorial or apical regions exchanged between M . tuberculosis and E . coli retained the normal chaperonin functions of GroEL 51 ., Bioinformatics analyses indicate that the duplicated GroELs from different bacteria share similar characteristics: the N- and C-intermediate regions are highly conserved , suggesting that these regions have essential functions in maintaining the functional structure of GroELs , and the apical , N-terminal and C-terminal regions are much more flexible , suggesting their possible roles in functional divergence ., The region-swapping experiments indicated that the functional divergence of the duplicated GroELs in M . xanthus was caused by the apical and C-terminal regions ., The GGM repeat at the tail of the GroEL1 C-terminal region , which is similar to that of E . coli GroEL , is important for the distinct functions of GroEL1 in development and sporulation ., These results are consistent with the positions of these two regions in the GroEL oligomeric complex , i . e . , the apical region is at the opening through which substrates enter the central cavity , and the C-terminal equatorial region is at the bottom of the cavity 43 ., However , it remains unclear whether region swapping has effects on in vivo chaperonin functions ., To address the question , we assayed the survival rates of the mutants YL0903 , YL0904 and YL0905 in response to heat shock and found that all the mutants rescued the lethality of heat-shock observed for YL0301 ., However , the survival rates of YL0903 , YL0904 , and YL0905 were low compared with the strains complemented with an intact groEL gene ( Figure S5 ) ., This result suggests that the region-swapped GroELs function in M . xanthus cells but that these functions were affected , at least at non-permissive temperatures ., It is still unclear whether the chimeras interact with intact GroELs and whether the in vivo functions of the chimeras result from mixed GroEL complexes ., Furthermore , although the protein substrates of the M . xanthus GroELs were consistent with those of the single E . coli GroEL with regard to their secondary structural features 9 , the substrate spectra varied significantly ., This variation is most likely due to the low level of sequence similarity between the E . coli GroEL and the M . xanthus GroELs ( E . coli GroEL is 67 . 3% and 65 . 2% similar to M . xanthus GroEL1 and GroEL2 , respectively ) and to the difference in the protein substrates between these two bacteria ., Therefore , there are many questions related to the GroEL chaperonins and their functional divergence that need to be addressed ., The strains and plasmids used in this study are listed in Table S1 ., For the growth assays , the M . xanthus strains were cultivated in the Casitone-based nutrient-rich CTT medium 52 ., The E . coli strains were routinely grown on Luria-Bertani ( LB ) agar or in LB broth ., E . coli was grown at 37°C , whereas the Myxococcus strains were incubated at 30°C ., When required , 40 µg/ml of kanamycin ( Km ) and 10 µg/ml of tetracycline ( Tet ) ( Sigma ) were added to the medium ., The groEL expression levels during heat shock and liquid predation were analyzed using quantitative real-time PCR ., M . xanthus DK1622 and other mutants were harvested after 18 h and exposed to 42°C for 1 h ., The RNA was extracted immediately using a total RNA extraction kit following the manufacturers instructions ( Promega ) ., Contaminating DNA was removed with a DNAfree kit ( Ambion ) ., The purified RNA was transcribed to yield cDNA , which was stored at −70°C ., The quantitative real-time PCR was performed using a Bio-Rad sequence detection system with 250 nM primers , 10 µl of SYBR Green PCR Master Mix ( Bio-Rad ) , 7 µl of RNase-free water , and 2 µl of cDNA template ., The PCR was performed for 3 min at 95°C , followed by 40 cycles of 30 s at 95°C , 30 s at 59°C , and 15 s at 72°C ., The 16S rRNA was used as a normalization signal ., Calibration curves ( groEL1 , groEL2 , and 16S RNA ) were generated using 10-fold dilutions of M . xanthus DK1622 genomic DNA ., The following pairs of forward and reverse primer pairs were used: groEL1 , 5′-CACCGAGACGGAGATGAAGG-3′ and 5′-TGAGGCAGCGGATGTAGGC-3′; groEL2 , 5′-ATCCGCACGCAGATTGAC-3′ and 5′-GCfCTTCTTCTCCTTCATCTCC-3′; and 16S rRNA , 5′-CGCCGTAAACGATGAGAA-3′ and 5′-TTGCGTCGAATTAAACCAC-3′ ., The groEL expression levels during predation were analyzed using quantitative real-time PCR ., The strains were cultured for 50 h in medium containing casein as a substrate instead of hydrolyzed proteins , and the RNA was extracted immediately ., The method and the primers used were the same as those described above ., The groEL expression level during different developmental stages was analyzed by measuring the β-galactosidase activity , as described by Li et al . 15 , 53 , with minor modifications ., The cells were broken using a Mini-Beadbeater ( BioSpec ) at a speed of 2500 rpm ., The β-galactosidase activity was determined using o-nitrophenyl-β-galactopyranoside ( Sigma ) , and the samples were analyzed at 420 nm ., The total protein concentration was determined using the bicinchoninic acid protein assay ( Pierce ) ., The specific activity was calculated as follows: specific activity\u200a=\u200a213×A420/ ( sample volume×protein concentration×reaction time ) 15 , 54 ., M . xanthus cells were harvested at mid-logarithmic phase and suspended to a final density of 5×109 cells/ml in TPM buffer ., Aliquots ( 10 µl ) were spotted onto TPM agar , and the plates were cultivated at 30°C and observed every 24 h to monitor the formation of fruiting bodies ., The sporulation rate was measured after 5 days as previously described ., The assays were performed at least three times 15 ., The predation assays were performed according to the method used in a previous study 15 ., E . coli and M . xanthus cultures were harvested at mid-logarithmic phase and washed three times with 10 mM MOPS buffer ( pH 7 . 6 ) ., The final cell densities of the cultures were 5×109 cells/ml for M . xanthus and 1×1011 cells/ml for E . coli ., Then , 50 µl of E . coli was pipetted onto a plate to form a 1-cm-diameter colony , and 2 µl of M . xanthus was added to the center of the E . coli colony , with an inoculation diameter of 0 . 15 cm ., The assay was repeated at least three times ., The plates were incubated at 30°C for 6 days , during which time the size of the M . xanthus growth area was recorded every 12 h ., The predation ability of M . xanthus was reported as the time required for M . xanthus to spread to the edge of the E . coli colonies ., M . xanthus cultures were harvested as described above ., The cells were heat shocked for 30 min at 42°C , serially diluted and plated on CTT agar ., After 6 days incubation , the CFUs were calculated 15 ., The groEL gene sequences from ten genome-sequenced bacterial strains were retrieved from the NCBI database ( Table S2 ) , and the amino acid sequences were aligned using the protein sequence alignment program in CLUSTALW 55 ., A maximum likelihood tree was constructed using MEGA5 56 ., The Ka/Ks values among orthologous groEL genes or among paralogous groEL genes were calculated using KaKs_Calculator 1 . 2 57 with the NG , MLWL and MLPB models 58 , 59 ., The region-swapping assay was conducted according to a previously published method 51 ., The regions responsible for the developmental defects of YL0301 and the predation defects of YL0302 were investigated by incorporating single groEL regions into YL0301 or YL0302 ., The complementation mutants were constructed with the site-specific integration plasmid pSWU30 ., The apical region of groEL1 was inserted into YL0301 to obtain YL0904 ( YL0301::pSWU- groEL2-apicalgroEL1 ) ., Briefly ( Figure S4 ) , 0 . 5 kb of the upstream sequence and the N-terminal region ( bp 1–597 ) of groEL2 and the C-terminal ( 597 to the end ) of groEL1 were amplified by PCR ., The two fragments were spliced by fusion PCR , digested with XbaI and BamHI , and ligated into pSWU30 digested with XbaI and BamHI ., The plasmid was transferred to E . coli λ-pir cells , and the plasmid DNA was extracted from the Tet-resistant transformants using the eZNA Plasmid Mini Kit I ( Omega Bio-Tek ) according to the manufacturers instructions ., The correct plasmid was used as the template in the second round of fusion PCR ., The plasmid containing the correct sequence was transferred by electroporation into YL0301 , and individual Tet-resistant colonies were screened ., The mutant phenotypes were observed to determine the effects of the apical region on development ., The same method was used to replace other regions ., The primers used are listed in Table S4 ., The GGM region deletion mutants were constructed using the positive-negative KG cassettes described by Ueki et al . Briefly , the upstream sequence ( before the GGM sequence ) and the downstream sequence ( after the GGM sequence ) were amplified by PCR ., The two fragments were fused to the XbaI restriction site to construct homologous fragments with in-frame deletions ., These homologous fragments were ligated into SmaI-digested pBJ113 ., The resulting plasmid containing the correct sequence was transferred by elec

Introduction, Results, Discussion, Methods

The gene encoding the GroEL chaperonin is duplicated in nearly 30% of bacterial genomes; and although duplicated groEL genes have been comprehensively determined to have distinct physiological functions in different species , the mechanisms involved have not been characterized to date ., Myxococcus xanthus DK1622 has two copies of the groEL gene , each of which can be deleted without affecting cell viability; however , the deletion of either gene does result in distinct defects in the cellular heat-shock response , predation , and development ., In this study , we show that , from the expression levels of different groELs , the distinct functions of groEL1 and groEL2 in predation and development are probably the result of the substrate selectivity of the paralogous GroEL chaperonins , whereas the lethal effect of heat shock due to the deletion of groEL1 is caused by a decrease in the total groEL expression level ., Following a bioinformatics analysis of the composition characteristics of GroELs from different bacteria , we performed region-swapping assays in M . xanthus , demonstrating that the differences in the apical and the C-terminal equatorial regions determine the substrate specificity of the two GroELs ., Site-directed mutagenesis experiments indicated that the GGM repeat sequence at the C-terminus of GroEL1 plays an important role in functional divergence ., Divergent functions of duplicated GroELs , which have similar patterns of variation in different bacterial species , have thus evolved mainly via alteration of the apical and the C-terminal equatorial regions ., We identified the specific substrates of strain DK1622s GroEL1 and GroEL2 using immunoprecipitation and mass spectrometry techniques ., Although 68 proteins bound to both GroEL1 and GroEL2 , 83 and 46 proteins bound exclusively to GroEL1 or GroEL2 , respectively ., The GroEL-specific substrates exhibited distinct molecular sizes and secondary structures , providing an encouraging indication for GroEL evolution for functional divergence .

GroEL is a type I chaperonin , involved in protein folding , assembly , and transport ., It is a major group of heat-shock proteins that are over-expressed at high temperatures and has fundamental roles in growth and survival at non-permissive temperatures ., Because of its importance in many cellular processes , the groEL gene is ubiquitously distributed in bacteria ., Most bacterial species possess a single groEL gene , while others ( close to 30% of sequenced bacterial genomes ) have two or more groEL copies ., Many studies have described the functional divergence of duplicated groEL genes in different bacterial species , but the involved mechanisms have not yet been characterized ., Myxobacteria are characterized by their unique multicellular behaviors ., Myxococcus xanthus DK1622 , the model strain of myxobacteria , possesses a large genome ( 9 . 14 Mb ) , containing many gene duplications , including two copies of the groEL gene ., Gene duplications and their functional divergence are suggested for complex cellular behaviors , which , however , have not yet been testified ., In this paper , using combined proteomic and genetic approaches , we explored how the duplicated groEL genes of M . xanthus DK1622 evolved to fit the functional divergence for social behaviors .

genetics, biology, evolutionary biology, genomic evolution, genetics and genomics

journal.pcbi.1003002

Folding Pathways of a Knotted Protein with a Realistic Atomistic Force Field

Natively-knotted proteins are increasingly studied as a new paradigm of “multiscale” folding coordination , which leads to establishing the native knot in the native position starting from the unknotted newly-translated state 1–4 ., Intuitively , the pathways associated to this process appear so improbable and prone to misfolding that it was long held that naturally occurring proteins would be protected against the occurrence of knots ., This a priori expectation , which has a sound statistical basis 5 , 6 , was so strong radicated that only several years after the publication of the human carbonic anhydrase II structure 7 it was realized that it actually accommodated a knot 8 ., Since then , hundreds of instances of naturally-occurring knotted proteins have been found and they now account for about 2% of the protein data bank ( PDB ) entries 6 ., The salient aspects of the folding phenomenology of several knotted proteins have been recently probed by various experiments ( for recent reviews see refs . 1–4 ) ., These studies have demonstrated that newly translated , unknotted proteins , can fold into the native knotted structure without the assistance of chaperones 9 , 10 , though the latter can significantly speed up the process 10 ., The details of the concerted backbone movements that lead to the self-tying of the protein in the native knot remain , however , beyond reach of current experimental techniques ., In this regard , numerical investigations can aptly complement experimental ones , by providing valuable insight into the repertoire of viable modes of knot formation , the stage at which the knot is formed , and the possible role of non-native interactions 11–14 ., To ease the major computational burden imposed by simulating the slow process of spontaneous folding/knotting of these molecules , the above-mentioned studies were performed using -type native-centric force fields , in either coarse-grained ( CG ) or atomistic protein representations ., The latter approach allowed for establishing the noteworthy result that by promoting native interactions alone it is possible to fold a natively-knotted protein 11 , 12 ., Non-native interactions have , however , been argued to be important for enhancing the efficiency of the process , by significantly increasing the accessibility of knotted configurations in the early folding stages 13 , 14 ., A natural test case for numerical studies of spontaneous knotting in polypeptide chains is represented by protein MJ0366 , which is the shortest known knotted protein ., The folding process of this 82-amino acid long protein appears to be governed by such a delicate interplay of amino acid stereochemical interactions that folding simulations employing different levels of spatial resolution have been shown to yield different knotting mechanisms ., In particular , the seminal study of Noel and co-workers 12 , where the folding of MJ0366 was characterized using pure native-centric force-fields , has shown that in coarse-grained folding simulations , the knot could form at either terminus , while only the C-terminal is involved in knotting when the full atomistic detail is used ., The observed sensitivity of the MJ0366 folding process on structural details poses a further fundamental question: to what extent is the knotting mechanism sensitive to details of the force field used in folding simulations ?, ., Towards this goal , we here analyze an ensemble of about 30 successful atomistic folding trajectories for protein MJ0366 , obtained by using a realistic force field , namely AMBER99ffSB 15 with implicit solvent ., To the best of our knowledge this represent the first instance where a realistic force-field is employed to follow the folding of initially unfolded , and unknotted conformations into a knotted native state ., To collect this sizeable number of productive trajectories in an affordable amount of computational time , we have used an advanced simulation technique known as the “dominant reaction pathway” ( DRP ) scheme ., In other protein contexts , this method was shown to yield results consistent with standard extensive MD folding simulations , performed with the same atomistic force field 16 ., We find that self-knotting of MJ0366 typically occurs at a late folding stage , when about of the native contacts are established and almost invariably involves a single dominant knotting mechanism ., The latter consisting of the threading of the C-terminus through an open region created by an already formed -sheet ., Based on various model calculations it is argued that the observed difference in knotting modes is strongly influenced by non-native interactions ., The selected 31 trajectories were analyzed by monitoring the evolution of several geometrical and topological parameters during the folding process ., As a first step we identified the folding stage at which the backbone self-ties into knot ., Accordingly , for each trajectory we calculated the percentage of native contacts ( overlap ) that are formed when the first knotting event occurs ., The distribution of these overlaps for the considered trajectories is shown in Fig ., 2 . The distribution is peaked at about 90% overlap ., This indicates that the knot is typically formed at a rather late stage of the folding process ., Next , to characterize the diversity of the folding pathways and the implications for the knotting mechanism , we computed the average “path similarity parameter” , ., As explained in the Materials and Methods section , this quantity measures the consistency of the temporal succession in which the native contacts are formed in two given pathways ., The parameter takes on values ranging from 0 , for no similarity , to 1 when all native contacts form with exactly the same succession in the two trajectories ., We emphasize that depends only on the time order of native contact formation events ( and not their exact timing ) ., To have a robust indication of the degree of heterogeneity of the selected trajectories , we computed the distribution of over all possible pairs of trajectories , see Fig ., 3 . As a term of reference , the same Figure shows the distribution computed over previously-studied folding trajectories of the unknotted WW domain FIP35 16 ., It is seen that the distribution of MJ0366 is narrower and shifted towards significantly higher values of than for the unknotted protein ., Indeed the former has a peak at while the latter has it at ., This relatively low value of and the distribution broadness is typical of folding processes that proceed by multiple pathways 16 , 24 , as FIP35 is known to do ., The different characteristics of the distribution for MJ0366 therefore strongly suggest the existence of one dominant folding pathway for MJ0366 ., We accordingly sought to analyze in detail the folding process to verify that knotting occurs via one dominant mechanism and characterize it ., In this regard a valuable term of reference is given by the earlier study of Noel et al . 12 where the folding thermodynamics of MJ0366 was systematically characterised with both atomistic and coarse-grained native-centric models ., When the atomistic model was employed , it was seen that knotting preferentially occurred via slipknotting ., Specifically , in most of the productive trajectories obtained at the folding temperature of the structure-based model , the C-terminal attained a hairpin-bent conformation and established the knot by threading the open region involving residues 17–54 ., The slipknotting mechanism was found to occur more frequently than that of other knotting modes , such as the threading of the open region by a non-bent C-terminus , or knot formation at the N terminus ., Interestingly , the coarse-grained native-centric model was more prone to unproductive kinetic traps and displayed significant heterogeneity for knotting mechanisms too ., These aspects indicated that the realistic treatment of protein structural detailed significantly helped reduce the impact of unproductive routes in the folding process 12 ., Here , by addressing the same protein folding process with a realistic , non native-centric force-field , it is possible to examine to what extent various aspects of the knotting process are sensitive to the treatment of inter-atomic interactions ., As a first step of the analysis , we profiled the folding trajectories along two relevant order parameters: the root mean square distance ( RMSD ) to the native structure and the RMSD to the native -sheet ., The first collective variable monitors the overall progress towards the native geometry ., The second one , instead , carries information about one of the expected entropic bottlenecks of the folding process , namely the formation of the native antiparallel -sheet which involves amino acid pairs with a sequence separation as large as 38 ., Since in the native MJ0366 structure the C-terminal helix protrudes through the region intervening between the two paired -strands , monitoring the formation of the -sheet is relevant to understand whether the sheet is formed before or after the knot ., The results shown in the left panel of Fig . 4 indicate that the -sheet is fully formed rather early , when the total RMSD to native of the chain is about 15 Å ., At this stage the fraction of formed native contacts is about 40–50 ., The self-tying of the molecule into a trefoil knot typically occurs after the formation of the -sheet ., This is evident from the placement of the diamond symbols in Fig . 4 which mark the first occurrence of knots for each of the 31 trajectories ., It is seen that all first-knotting events occur when the -sheet is fully formed , with only two exceptions that will be discussed later ., The detailed inspection of the trajectories indicates that the knotting process almost invariably occurs through the so-called “threading” mechanism , where the -terminal -helix ( residues 74–87 ) directly enters , without bending , the open region between amino acids 17–54 involving helices and and the intervening loop , see the sketch in the left panel of Fig . 5 ., In this case , the threaded region and the -sheet ( respectively shown in blue and red in Fig . 1 ) establish a tertiary contact before the terminal helix penetrates into the open region in between the helices and ( see left panel in Fig . 5 ) ., This mechanism accounts for as many as 26 of the 31 rMD trajectories ., In three other cases , the folding was found to occur through the so-called “slipknot” mechanism 12 ( i . e . where the open region is entered by the backward-bent C-terminus ) ., In all three instances the terminus entered the loose – region producing a shallow slipknotted trefoil , as shown in the central panel of Fig . 5 ., Finally , in two further cases we observed another knotting mechanism which involves a concerted backbone movement that had not been previously reported for MJ0366 ., Specifically , in two trajectories when the -sheet and the terminal -helix are already formed and juxtaposed in an unknotted configuration the loop performs a “mousetrap-like” movement establishing the native knotted topology ., This movement , which bears some analogies with the suggested knotting mechanism for an unrelated protein with a non-trefoil topology 25 , is schematically represented in the right panel Fig . 5 ., The mousetrap knotting events correspond to the two outlying diamonds reported in Fig . 4 , with collective coordinates ( 6 Å , 8 Å ) and ( 12 Å , 10 Å ) ., Videos obtained from the atomistic DRP trajectories which illustrate the three observed knotting mechanisms are included in the on-line SI ., It is important to notice that the trajectories associated to the various knotting modes do not present significant quantitative differences regarding the overall solvent accessibility of polar and non-polar residues during the folding process ., This point is illustrated in Fig . 6 where the number of exposed hydrophobic and hydrophilic residues are profiled versus the RMSD to the native state ., The consistency of the various profiles provides a quantitative basis for expecting that the relative weight of the knotting mechanisms should not depend critically on the specific model adopted to describe the solvent-induced interactions ., To understand how the interplay of amino acid interaction captured by the realistic force field favours knotting by threading , we have carried out a comparative analysis of the reaction mechanism in successful and unsuccessful folding trajectories ., Specifically , the productive , successful set consisted of the 26 trajectories displaying the dominant ( threading ) knotting mechanism ., The non-productive one included an equal number of trajectories that reached an unknotted configuration and nevertheless had a good native similarity ( namely an RMSD to the crystal structure less than 5 Å ) ., The projection of the unsuccessful trajectories along the two collective coordinates considered before is shown in Fig . 4B ., The qualitative difference with respect to the analogous plot for the successful ones ( panel A ) is striking ., In particular , it is seen that in successful trajectories the formation of the sheet involving strands and occurs rather early on and prior to the establishment of the overall tertiary organization of MJ0366 ., In fact , the total RMSD to native decreases appreciably only after the -sheet is established ., By converse , for unsuccessful trajectories , this hierarchy of contacts formation is not observed , and the -sheet formation proceeds in parallel with the acquiring of the overall native structure ., One therefore concludes that the early formation of the -sheet provides the most appropriate conditions for knotting by leaving the region delimited by the sheet accessible to threading events ., This conclusion is supported by the detailed inspection of the unsuccessful trajectories , which are exemplified in the sequence of snapshots shown in Fig ., 7 . As it is visible in this figure , the C-terminal helix threads the correct region between strands and prior to the formation of the -sheet ., When the latter is finally establishes , the -terminal segment remains trapped on the wrong side of the loop bridging and and , for steric reasons cannot go past it and attain the native knotted topology ., The relevance of this mechanism for misfolding is highlighted by the fact that all unsuccessful trajectories displayed a late formation of the -sheet ., We emphasize again that , according to our simulations , the correct knotting of the chain is not promoted by the formation specific contacts which fail to form in misfolding events ., Rather , for the chain to acquire the native topology , it is essential that the native contacts form in the correct order ., The fact that the observed dominant knotting mode differs from the one reported previously using pure native-centric force fields suggests that non-native interactions could be relevant for favouring the correct succession of contacts leading to self-knotting ( or avoiding unproductive ones ) ., This possibility is particularly interesting in connection with the ongoing discussion about the role that non-native interactions can have in aiding the knotting process even during the early folding 13 , 14 ., To investigate this aspect we generated several folding trajectories for MJ0366 using simplified models where the effect of non-native interactions could be easily turned on or off ., Specifically , we considered two different coarse-grained models: one with only native-centric interactions and the other additionally incorporating non-native interactions ., The latter included quasi-chemical and screened electrostatic pairwise interactions , as in the recent study of the early folding stages of a trefoil-knotted carbamoyltransferases 14 ., The folding process presents major differences in the two models ., First , they differ significantly in terms of knotting probability ., Specifically , for each model we considered an extensive set of 10 , 000 uncorrelated configurations , equilibrated at the nominal temperature of 300 K . In the native-only case , 12% of the sampled configurations were knotted , while this number had a sixfold increased , to 75% , in presence of non-native interactions ., This result aptly complements the atomistic DRP simulations , for highlighting the role of non-native interactions in aiding the formation of the native knotted topology of MJ0366 ., Secondly , productive trajectories follow different dominant mechanisms in the two models ., In fact , when the pure native-centric model is used , 8 out of the 10 trajectories involved the slipknotting mechanism , while the threading one was observed in all trajectories ( 10 out of 10 ) with the additional non-native interactions ., The latter result , which is in full accord with the atomistic DRP simulations , reinforces the concept that non-native interaction can promote the correct order of contact formation required for self-knotting ., This point is further supported by the inspection of the density plots in Fig ., 8 . In fact , non-native interactions are more clearly associated to the early formation of the -sheet than for the native-only case ., Furthermore , the path outlined in panel B bears more analogies than the one in panel a with the density plot of Fig . 4A , which captured the successful folding events obtained from atomistic DRP simulations ., Indeed , in the simplified model , the early formation of the sheet is promoted by the fact that the non-native quasi-chemical interaction generates an overall attractive interaction between the residues in and those in ., Consistently with the misfolding events discussed previously , one can therefore argue that the weaker drive of the native-centric model to form early on the -sheet , is also responsible for its lower knotting propensity ., Based on these results , we can argue that mutations in the sheet regions with residues characterized by a weaker effective attraction , would delay the formation of the -hairpin in the folding process and would make the chain more prone to reach the unknotted mis-folded state ., This prediction may be verified experimentally ., In conclusion , the DRP simulations presented here provided the first systematic attempt to characterize the folding process of a natively-knotted protein , MJ0366 , using a realistic atomistic force field ., MJ0366 knotting is observed to occur via threading at the C-terminal ., The comparison of productive and unproductive trajectories ( which respectively end up in natively-knotted and unknotted states ) further indicates that knotting is aided by the early formation of the native -sheet ., By comparing the MJ0366 knotting propensity and mechanisms in simplified folding models it is argued that non-native interactions are important for aiding knotting by promoting the correct order of contact formation ., While there is no a priori reason to expect that non-native interactions are crucial for guiding the folding process of knotted proteins in general , it is interesting to notice that their important role has been previously suggested for another trefoil-knotted protein carbamoyltransferases 13 , 14 ., In our view , it would be most interesting to further examine this effect , in future studies on MJ0366 or other proteins either through experiments ( e . g . involving mutagenesis ) or with more extensive simulations , possibly involving explicit solvent treatment or unbiased dynamics ., In order to generate an ensemble of trial trajectories connecting a given initial configuration to the native state we used the following variant of the rMD algorithm ., At each integration step , we evaluated a collective coordinate ( CC ) which measures the distance of between the instantaneous contact map and the native contact map: ( 1 ) with with a distance cutoff of Å ., In this equation , is the 3N-dimensional vector in configuration space , and and are the instantaneous and native contact map , respectively ., The entries of the contact map C are chosen to interpolate smoothly between 0 and 1 , depending on the relative distance of the atoms and : ( 2 ) where r0\u200a=\u200a7 . 5 Å is a fixed reference distance ., In the rMD algorithm , no bias is applied to the chain when it spontaneously diffuses towards the bottom of the folding funnel , i . e . any time the value of the CC at time is smaller than the minimum value so far ., On the other hand , fluctuations which would drive the contact map further from the native one ( hence increasing ) are hindered by introducing a biasing force , defined by the time-dependent potential ( 3 ) In these equations , kcal/mol is the so-called ratchet constant and is the minimum value assumed by the collective variable along the rMD trajectory , up to time ., In the original formulation of the rMD algorithm 19 , the variable is updated only when the system visits a configuration with ., With this choice , monotonically decreases during the course of the simulation ., In this work , we choose to significantly weaken the effect of the bias by allowing the system to backtrack along the direction defined by the CC ., This is done by occasionally updating also when it increases , according to a Metropolis accept/reject criterium ., Namely , is updated to if , where is a random number sampled from a uniform distribution and is an artificial “inverse thermal energy” ., This modification of the original rMD algorithm is required to escape from kinetic traps ., Without it the folding efficiency to the correct topologically non-trivial native state is strongly suppressed ., Each trial trajectory consisted of steps of rMD with a nominal integration time step of fs ., The DRP algorithm is used to identify the most probable path in each set of trial rMD trajectories sharing the same boundary conditions ., This is done by evaluating the relative probability for each path to be realized in the unbiased over-damped Langevin dynamics: ( 4 ) In this equation , the index runs over the different time-step in the trajectory , the index runs over the atoms in the protein , is the Boltzmanns constant and is the diffusion coefficient of the -th atom ., In both rMD and standard high-temperature MD simulations we used the AMBER ff99SB force field 15 in implicit solvent , within the Generalized Born formalism implemented in GROMACS 4 . 5 . 2 26 ., In such an approach , the Born radii are calculated according to the Onufriev-Bashford-Case algorithm 27 ., The hydrophobic tendency of non-polar residues is taken into account through an interaction term proportional to the solvent-accessible-surface-area ( SASA ) ., The solvent-exposed surface of the different atoms is calculated from the Born-radii , according to the approximation developed by Schaefer , Bartels and Karplus in 28 ., The CG folding simulations were based on the model developed in Ref . s 29 , 30 ., In this approach , amino acids are represented by spherical beads centered at the positions ., The non-bonded part of the potential energy contains both native and non-native interactions ., The former are the same used in the G-type model of Ref ., 31 , while the latter consist of a quasi-chemical potential , which accounts for the statistical propensity of different amino-acids to form contact and of a Debye-screened electrostatic term ., A detailed description of the force field of this model can be found in Ref ., 14 ., In our previous work , we have shown that such non-native interactions are able to strongly promote the knot formation in natively knotted polypeptides 14 ., Folding simulations for protein MJ0366 in this CG model were performed using a MC algorithm described in detail in Ref ., 14 ., This type of crankshaft-based MC algorithm is commonly employed in polymer physics 32 to study dynamic properties , since it is was shown that they mimic the intrinsic dynamics of a polymer in solution 33 at a much lower computational cost than standard MD simulations 34 ., The folding dynamics of CG model with native and non-native interactions was simulated by generating 200 MC trajectories , while the dynamics of the model with only native interactions was investigated by generating 500 MC trajectories ., For both CG models , trajectories consist of 1 . 5108 attempted MC moves , corresponding to 1 . 5104 saved frames ., MC moves that we have employed were the local crank-shaft and Cartesian moves , whose boldness was chosen such that the acceptance rate was nearly constant and approximately equal to 50 ., In both cases , we have collected a total of 10 folding transitions , leading to native configurations with the correct knotted topology ., In order to compute the frequency of knotted configurations at thermal equilibrium we performed MC simulations which combine local moves and global pivot moves ., The conformations visited during the MC dynamics were analysed to establish their global and local knotted state ., The global topological state was established and assigned by computing the Alexander determinants after suitable closure of the whole protein chain into a ring ., For each configuration , this entailed 100 alternative closures where each terminus is prolonged far out of the protein along a stochastically chosen direction , and the end of the prolonged segments are closed by an arc “at infinity” ( i . e . not intersecting the protein ) ., As in ref ., 14 , to avoid considering back-turning closures , stochastic exit directions are picked uniformly among those which form an angle of more than 90° with the oriented segment going from each terminus to the Cα at a sequence distance of 10 ., If the majority of the 100 stochastic closures return non-trivial Alexander determinants , then the whole conformation can be considered as globally knotted ., Because protein knotting can occur through slipknot formation 35 , the global topology investigation was complemented by a local one ., In fact , a slipknot can be detecting by identifying a non-trivially knotted portion of a chain that has a different global topology , in our case the unknotted one ., To this purpose , we repeated the above-mentioned statistical closure scheme for all possible subportions of length 20 , 30 , 40 , … of the protein chain so to identify the smallest knotted , or pseudo-knotted , chain portion 36 , 37 ., To quantitatively measure the folding pathways diversity we implemented the analysis described in Camilloni et al . 19 , that will be shortly summarized in the following ., A folding mechanism is here considered to be a specific sequence of native contacts formation ., Hence , for each path we measured the time of formation of each native contact , as the frame of the trajectory where the contact is first formed ., Given as the time of formation of the native contact in the trajectory , we computed for each path the matrix defined as: ( 5 ) containing all the information regarding the folding mechanism as defined above ., For each pair of pathways it is possible to compute the similarity defined as ( 6 ) being the total number of native contacts ., The similarity ranges from for completely different mechanisms , to for completely identical mechanisms ., Finally , we consider the distribution ( 7 ) of the similarity parameter , evaluated from all pairs of the folding pathways .

Introduction, Results/Discussion, Materials and Methods

We report on atomistic simulation of the folding of a natively-knotted protein , MJ0366 , based on a realistic force field ., To the best of our knowledge this is the first reported effort where a realistic force field is used to investigate the folding pathways of a protein with complex native topology ., By using the dominant-reaction pathway scheme we collected about 30 successful folding trajectories for the 82-amino acid long trefoil-knotted protein ., Despite the dissimilarity of their initial unfolded configuration , these trajectories reach the natively-knotted state through a remarkably similar succession of steps ., In particular it is found that knotting occurs essentially through a threading mechanism , involving the passage of the C-terminal through an open region created by the formation of the native -sheet at an earlier stage ., The dominance of the knotting by threading mechanism is not observed in MJ0366 folding simulations using simplified , native-centric models ., This points to a previously underappreciated role of concerted amino acid interactions , including non-native ones , in aiding the appropriate order of contact formation to achieve knotting .

It has been recently observed that the native structure of proteins can contain knots ., These are formed during the folding process and are tightened in a specific ( i . e . native ) location , along the poly-peptide chain ., The existence of knots hence implies a high degree coordination of local and global conformational changes , during the folding reaction ., In this work we investigate how the knot is formed and what are the dynamical mechanisms which drive the self-entanglement process ., To this end , we report on the first atomistically detailed numerical simulation of the folding of a knotted protein , based on a realistic description of the inter-atomic forces ., These simulations show that the knot is formed by following a specific sequence of contacts ., The comparison of the findings with those based on simplified folding models suggest that the productive succession of contacts is aided by a concerted interplay of amino acid interactions , arguably including non-native ones .

physics, protein folding, biophysics theory, biophysics

journal.pcbi.1002913

The HAMP Signal Relay Domain Adopts Multiple Conformational States through Collective Piston and Tilt Motions

To survive , bacteria must constantly monitor their environmental conditions and adapt to these by generating a response , such as a change in gene expression or motility ., In bacteria , signaling proteins are built from modular components that regulate input , output and protein-protein communication ., Many signaling proteins contain characteristic transmitter and receiver domains that promote information transfer within and between proteins ., Signaling pathways are assembled by arranging these domains in various configurations 1 , of which the simplest have two protein components: a sensor monitoring an environmental parameter , often located close to the membrane , and a cytoplasmic response regulator that mediates an adaptive response ( i . e . a change in gene expression ) ., The sensor typically contains an N-terminal input domain coupled to a C-terminal transmitter module ., In many two-component signaling pathways , transmembrane -helices position the sensor/transmitter at the periplasmic side of the membrane , with the transmitter oriented toward the cytoplasm , see FIG . 1-A ., Communication with the transmitter domain occurs via stimulus-induced conformational changes of the linker regions ., A typical linker region is the HAMP domain , originally identified in Histidine kinases , Adenylyl cyclases , Methyl-accepting chemotaxis protein and Phosphatases 2 ., This residue motif functions as a signal relay , converting the signal received into activation of the transmitter domain 3 ., HAMP sequences contain heptad repeats ( ) , in which residues and are typically hydrophobic , indicating that HAMP forms a coiled-coil complex ., HAMP domains exist as a single unit , known as the the canonical form 3 , but also occur in a sequentially repetitive fashion , known as the diverse form 4–6 ., In the canonical form , the domain can be coupled to many different types of receptors and output regulators , such as diguanylate cyclases and phosphodiesterases 4 ., As a repetitive domain , HAMP occurs both in intracellular signaling proteins 5 and transmembrane receptors 6 ., Their wide occurrence , yet high structural similarity , may indicate a versatile mechanism for signal propagation in prokaryotes ., The first structure of a HAMP domain was resolved by NMR spectroscopy from the cytoplasmic C-terminal domain of the non-signaling trans-membrane ( TM ) protein Af1503 from the highly thermophilic organism A . fulgidus ( PDB code 2L7H ) 3 ., Identification of this protein domain occurred through sequence similarity to known HAMP sequences 3 , 7 ., While lacking the periplasmic input and cytoplasmic output domains typically coupled to a canonical HAMP , the AF1503-HAMP domain shows activity when expressed in E . coli , substituting for the original HAMP domain in the chemotactic receptor Tar 8–10 ., The structure of Af1503-HAMP shows a dimeric coiled-coil complex comprising four helices in a parallel orientation , shown in FIG . 1-B ., The two monomers are labeled 1 and 2 ., Containing 58 residues , each monomer consists of two helices , labeled N and C , connected by a residue linker ., The hydrophobic core of a canonical coiled coil comprises layers of residues and in the same heptad repeat , referred to as knobs-into-holes or ( ) packing ., Instead , the Af1503-HAMP structure displays an unusual packing in which each layer consists of either residues or in the N-helices interacting with residues or in the C-helices , see FIG . 1-C ., As each helix contains two heptad repeats , the hydrophobic core of HAMP contains four layers ., Additional residues directly preceding the -residues in the C-helices , or directly following the -residues in the N-helices , contribute to the packing , and are therefore labeled or respectively ., The residues in the helices that do not have neighboring residues contributing to the packing are labeled ., This packing is therefore referred to as complementary packing ., The structure of Af1503-HAMP currently serves as the prototype structure of HAMP 3 , 11 ., HAMP functions as a signal relay domain between input and output domains of many bacterial sensor proteins , transmitting signals via conformational changes ., An extensive mutagenesis study on the HAMP domain of the Tsr chemotaxis receptor provided insights into the mechanism of signal transduction by HAMP 12 , 13 resulting in the dynamic bundle model ., In this model , HAMP signal transduction occurs through changes in the stability of the helical bundle , modulated by conformational changes in the linker connecting HAMP to the transmembrane helices or changes in the stability of the output domain 12 ., More importantly , the changes in stability of HAMP , induced by either input or output signals , cover a wide range of different conformations , indicating that HAMP function is more complex than an on-off switch 13 ., Several models exist to describe the functional motions involved in the signal transduction mechanism of HAMP , including the gearbox model 3 , the piston model 14–16 and a model describing helical tilting 11 , 17 ., Hulko et al . compared the complementary packing mode of the prototype structure and the knobs-into-holes packing of a typical coiled-coil structure , showing that a concerted helix rotation by would convert the conformation into the canonical packing 3 ., Ala291 in the prototype structure is an -residue in the second heptad repeat of the N-helix and contributes to the packing as an -residue ., Because small residues favor packing and large residues favor packing 18 , residue 291 of Af1503-HAMP was changed to explore the influence of the sidechain size on adenylyl cyclase activity , using a chimeric assay system 3 ., This mutation study revealed an inverse dependence of the activity on the volume of the hydrophobic sidechain at position 291 3 ., In particular , the A291V mutant reduced the activity to 62% compared to the wild type ( WT ) system and appeared to oscillate rapidly between two forms with presumably the packing and the packing ., Recently , structural data for most of these mutants became available 7 , 19 , revealing that there are several intermediate structures in the conversion between complementary and knobs-into-holes packing modes 7 ., The mutant A291F shows the highest structural diversity , as its crystal structure revealed an anti-parallel conformation , whereas in solution the mutant conformation is a mixture of parallel and anti-parallel conformations ., The parallel conformation revealed the knobs-into-holes packing 7 with the corresponding helical rotation ., Further evidence for helical rotation comes from the photoreceptor NpHtrII from N . pharaonis ., Upon excitation by light , the NpHtrII transmembrane helices perform a rotation and a displacement lateral to the membrane , as shown by electron paramagnetic resonance studies 20 ., A second model is known as the piston model ., Structural investigations on the aspartate chemoreceptor Tar in E . coli have shown that a transmembrane helix linked to a HAMP domain exhibits a piston-like motion inward to the cytoplasm upon binding of a signaling molecule to the periplasmic sensor domain 14 , 15 ., As the HAMP domain is directly connected to the transmembrane helix undergoing this inward motion , a piston-shift motion may play a role in HAMP mediated signal transduction ., A mutation study focusing on positioning the transmembrane helix directly preceding the HAMP domain in the Tar receptor further confirmed that these helices exhibit a piston-like motion , inward to the cytoplasm 21 ., Furthermore , a molecular simulation study looking into the position of the anchoring residues in the transmembrane helix of a chemotaxis receptor showed that downstream signaling activity was strongly correlated with a piston shift of 1 . 5 Å of the transmembrane helix 16 ., In Ref ., 11 , Falke et al . confirmed the NMR structure of Af1503-HAMP as a structural template for the Tar HAMP domain and proposed , based on activity studies of Tar , a pivot model in which an initial piston motion may be able to tilt the helices from different subunits of HAMP with respect to each other ., Helical tilting is also proposed as a model for signal relay based on in vivo cross-linking studies of a HAMP domain in the membrane based Aer sensor monitoring the intracellular redox potential 17 ., Interestingly , this study found that the N-terminal helix in one monomer tilts in concert with the C-terminal helix in the other monomer ., Molecular simulation can complement experiments by modeling the dynamical time evolution of biomolecular systems in atomistic detail ., A recent molecular dynamics study using a structural model of part of the Tar chemotaxis receptor elucidated the role of the connection between the transmembrane helices and HAMP in transmitting the signal from the sensor domain 22 ., These simulations showed that HAMP exhibits larger fluctuations and a helical tilt upon a downward piston shift of the second transmembrane helix 22 ., In this work , we aim to elucidate the nature of the signal transduction mechanism by HAMP , by investigating its equilibrium behavior via molecular dynamics ( MD ) ., In particular , we test the hypothesis that HAMP can adopt different conformations , of which one represents an active , signal-relaying configuration , and another an inactive , resting state ., To this end , we perform regular MD simulations on Af1503-HAMP in two conformations ., One conformation is the NMR structure , whereas an alternative conformation originated from the mutant A291F , which has a distinctly different packing ., In addition , we enhance the sampling with metadynamics , which applies adaptive biasing potentials in MD simulations , based on predefined collective variables ( CVs ) 23 ., These CVs constitute the motions the helices in HAMP exhibit with respect to each other: tilting , piston shift and rotation , based on the various models for the mechanism through which signals are relayed from the input domain , via HAMP , to an output domain ., We find that Af1503-HAMP can adopt three additional conformational states besides the NMR structure , and that these states can inter-convert via changes in the piston shift of the helices ., These conformational changes also directly lead to changes in the tilt angle between two HAMP monomers ., Finally , biasing the helical rotation does not lead to a significant conformational change ., This work supports the hypothesis that piston motions of the input domains connected to HAMP trigger the activation of HAMP by inducing piston motion in the output domain , most likely in combination with a tilting of the output domain helices ., Mutagenesis studies have shown that Af1503-HAMP has reduced activity upon altering the alanine at position 291 ., Increasing the volume of the hydrophobic sidechain at this position changes the packing in the hydrophobic packing from complementary to ( knobs-into-holes ) 7 ., In this section , we perform MD simulations on wild-type and the mutant A291F Af1503-HAMP domains , to investigate the differences in structure and dynamics of these conformations ., First we performed four 40 ns and four 60 ns MD simulations of the wild-type Af1503-HAMP domain , called WT hereafter , using the NMR structure ( PDB code 2ASW 3 ) as a starting point ., Visual inspection revealed no dissociation of the complex or unfolding of the -helical regions ., As a quantitative measure we calculated the RMSD of the helices with respect to the NMR structure , , and the number of helical hydrogen bonds , , shown in FIG . 2-A as a contour plot of the negative natural logarithm of the probability distribution of these two measures ., The profile displays a single minimum at =\u200a0 . 7 Å and around 50 ., In other representations , including the helical rotation , , the inter-helical tilt angles , and the helical piston motion , the WT simulations also display a single minimum ., The values of these collective variables are listed in TAB ., 1 . The helices within one monomer have a tilt angle with respect to each other , while tilting angles between monomers are around ., These angles are consistent with typical values observed in Ref ., 3 and reflect that the monomers are not exactly parallel , but have a tilted orientation with respect to each other ., Consequently , the HAMP domain resembles a cone with the tip at the C-terminal side , see FIG . 1-B ., Finally , the helical piston shift in the WT system is very small ., All these observations indicate that the structure resolved by NMR for the Af1503 HAMP domain is very stable as a single unit at room temperature ., By increasing the volume of the hydrophobic sidechain at position 291 , Ferris et al . have shown that the hydrophobic core can exhibit different packing modes that are in between the complementary packing and canonical packing 7 ., The A291F variant can adopt several conformations , including the packing , as shown by NMR spectroscopy 7 ., We performed MD simulations of this mutant , revealing that the parallel A291F structure is not stable in solution , see FIG . S1 in Text S1 for details ., The simulations showed either the onset of helical unfolding or relaxation to a conformation obtained by fusing the A291F variant to a C-terminal domain 7 ., We used this perturbed structure as a starting point to explore further the conformational space of the wild-type Af1503-HAMP ., We therefore prepared a structure in which positions of atoms are identical to the NMR structure of the A291F mutant but with the phenylalanines on position 291 changed to alanines , again yielding the wild-type sequence ., We performed 24 independent 50ns MD trajectories on this system , denoted as WT* ., Most of these trajectories relax to values of 1 . 2 Å or lower ., In one out of the 24 trajectories , the helical bundle changes to an “out-of-register” conformation with a mismatch of the hydrophobic layers ., This shifted register could be the result of a piston motion induced by asymmetric input from the sensor domains , pushing monomer 1 down with respect to monomer, 2 . However , there are several reasons to consider this conformation as misfolded rather than an alternative functional state of HAMP ., As already noted in Refs ., 11 , 14 , a register shift is too severe a change for a HAMP domain: the functional states of HAMP should closely resemble the Af1503-HAMP structure with only minor rearrangements 11 , 14 ., An out-of-register shift of the hydrophobic layers reflects a piston shift of 4–5 Å , which is larger than 2 Å determined from crystallography studies of the input domain 14 ., We therefore excluded this trajectory from further analysis ., FIG . 2-B displays the probability distribution as a function of and the RMSD of the helices with respect to the NMR structure of the A291F mutant , , revealing two minima , and ., The minimum is identical to the configurations sampled in the WT-labeled simulations , as indicated by the low value for ., The minimum deviates from the wild-type configuration , but is also different from the A291F conformation ., Note that this graph only gives an indication of the low free energy regions ., In the WT* simulations , transitions from to or vice versa occur only once in eight of the trajectories and not at all in the others , which is insufficient to give an accurate estimate of the free energy barriers separating the different states ., The simulations clearly show a relaxation from the A291F mutant structure with packing to conformations close to the structure of wild-type Af1503-HAMP , which may involve helical rotation , as postulated in the gearbox model 3 ., The rotation of a helix along its principal axis can be defined in different ways ., The program samCC can calculate several properties of helical bundles , including the Crick angle of a coiled-coil complex 4 ., The Crick angle is defined for each residue as the angle between the center of the bundle , the residue and the center of the helix ., This gives a measure for the rotation per residue ., Instead , we computed the rotation of the entire helix , treated as a single rigid body , by defining a rotational reference point on the helix and then calculate the angle between this reference point on a structure , the helical center of mass and a reference structure: the NMR structure of wild-type Af1503-HAMP ., This procedure is explained in detail in the Methods section ., In FIG . 2-C , we plot the time evolution of the four helical rotation angles of a single , typical WT* simulation , which ends in the state ., All helices start out with positive rotation values , an effect of aligning the conformation to a reference structure ., The rotation angles drop to zero after a few ns , indicating the fast relaxation to conformations similar to wild-type Af1530 HAMP ., During the fast relaxation , visual inspection revealed that the pairs of N and C-helices exhibit similar rotation , whereas an N-C pair rotate in opposite directions , in agreement with the gearbox model ., Upon reaching the state , the N-helices have rotation angles of and the C-helices fluctuate around , with respect to the reference structure ., Visual inspection of the trajectories show that piston and tilting motions contribute to the relaxation process ., The necessity of such motions can also be deduced from comparing the conformations of the wild-type HAMP and the A291F variant ., As the two conformations have different bundle shapes , conversion of one into the other will require tilting of the helices and piston shifts to realign the hydrophobic layers ., The new conformations in / differ from the WT conformation in the values for the piston shifts , as shown in FIG . 3-A , B ., FIG . 4 shows a schematic representation of the piston-shifted states ., indicates the conformations close to WT , without any piston shifts; =\u200a0 Å ., The new conformational state / is split up in two symmetrically related states ., Focusing on , this state exhibits an upward piston shift of 1 Å for the N-helix in monomer 1 ( N1 ) , and a downward piston motion of 1 . 5 Å of the C-helix in monomer 2 ( C2 ) ., Similarly , state reveals a downshift of C1 in combination with an upshift of N2 ., We show all possible piston combinations in FIG . S2 in Text S1 ., The piston shifts fall within the range of 1–2 Å , as experimentally determined 14 ., Strikingly , a piston shift of N1 is not correlated to piston shifts occurring for N2 ( see FIG . 3-A ) ., Similarly , the piston motions of the two C-helices are not correlated ., Changes in the four inter-monomer tilt angles are strongly correlated to each other , as explained in FIG . S3 and FIG . S4 in Text S1 ., We can therefore describe changes in the four inter-monomeric tilt angles by only one inter-monomer tilt angle , which describes the tilt angle between the helices of monomer 1 and of monomer, 2 . To determine whether other motions are related to the piston shift observed for states and , we plotted two-dimensional probability plots as a function of the N2 and C2 piston shifts and the tilt angle ( FIG . 3-B ) ., There is no difference in monomer tilt angle for the native conformation and the piston-shifted conformations and , because is between and for all states ., We investigated the rotation of helix N1 with the piston shift of the same helix in FIG . 3-C ., For the piston shift , two minima occur , which have similar values for the rotational angle ., Clearly , a piston shift seems to be uncorrelated to either changes in tilt or rotation of the helices ., In FIG . 3-A5 , we plotted the negative log probability distribution of the rotation of helices N1 and C2 ., This contour plot shows only one minimum and a small positive correlation , which seems in contrast with FIG . 2-C ., This figure shows one relaxation process , whereas FIG . 2-A5 shows the average relaxation to either or ., Although the WT* MD simulations occasionally visit a novel conformation , they only sample a small part of the conformational space and are inherently out of equilibrium ., To explore the equilibrium behavior we enhanced sampling by applying adaptive biasing potentials in the MD simulations , in the well-tempered metadynamics approach 23 , 24 ., As the biasing potentials are based on predefined collective variables ( CVs ) , described in the Methods section , the approach allows the identification of important CVs in conformational transitions ., First , we performed a metadynamics simulation , biasing the inter-monomer tilt angle ., In the first attempt , the range of was unlimited , resulting in values for of and higher ., At such a large tilt angle , the hydrophobic core is disrupted , leading to dissociation of the complex ., Once the complex has fallen apart , it is impossible to return to the intact state using only the inter-monomer tilt angle as a CV ., To prevent these severe changes we constrained the range of by adding a repulsive wall at ., Note that HAMP embedded in a sensory protein complex is very unlikely to explore very large tilt angles ., FIG . 5-A0 shows the time evolution of the biasing potential along ., After 35 ns , the shape of the profile does not change anymore ., At this point the negative biasing potential represents the free energy profile along and shows one broad free energy minimum ., The width of the minimum is consistent with the results from the conventional MD simulations ., Even though the biasing potential acts on one CV , we can obtain the free energy surface along other CVs by using a reweighting procedure 25 ., The resulting profiles are shown in FIG . 5-A1–A6 ., The patterns of piston motions as observed in the WT* simulations are partially reproduced ., Only the pair of helices N1-C2 exhibits piston motions , while does not change ( see FIG . 5-A1 , A2 ) ., When reaches , becomes more negative , see FIG . 5-A3 ( accordingly reaches 1 . 5 Å , see FIG . 5-A1 ) ., This shows that even though we bias the inter-monomer tilt angle , a spontaneous transition to the state can occur as well ., The reweighted free energy surface as a function of and for N1 in FIG . 5-A6 does not show such correlated motions ., Hulko et al . 3 postulated a mechanism , called the gearbox model , for relaying signals in HAMP via concerted rotation of the helices , thereby changing the packing of the hydrophobic layers ., Using metadynamics we can test this mechanism by biasing the rotation of one helix and observe the rotation of the other helices ., We performed a one-dimensional metadynamics simulation using as the CV ., The first attempt , in which the rotational angle was completely unconstrained , resulted in unfolding of the helix ., We therefore applied constraints to the range of , with a lower boundary at and an upper boundary at ., Recent structural studies revealed that concerted rotation does occur , but that the range is different per hydrophobic layer 7 , 19 ., This means that biasing the rigid body rotation of the entire helix will inevitably lead to unfolding , as some parts of the helices rotate differently than others ., FIG . 5-B0 shows the time evolution of the biasing potential and the resulting free energy surface ., From 6 ns onwards , the profile changes very little and reveals one broad minimum centered at , consistent with the observations for the conventional MD simulations of the wild type NMR structure ( see TAB . 1 ) ., The reweighted free energy surface in FIG . 5-B5 reveals a positive correlation between and , on which the bias was applied , similar to that observed in the WT* simulations ., FIG . 5-B1–B4 show that during this metadynamics simulation , not only the native state is visited , but also a piston-shifted state , with only one transition and one transition backwards ., This transition is not the result of the biasing potential on , but a spontaneous fluctuation in the piston mode of pair C1 and N2 ., This is revealed by the free energy surface as a function of and , in which is one broad minimum , completely uncorrelated to the changes in , see FIG . 5-B7 ., The WT* simulations revealed that Af1503 HAMP can adopt different conformations , which can be distinguished by the piston shift ., In the metadynamics simulations biasing the rotation and tilting these two conformations do not show up in the profile of the biasing potential , whereas they do appear spontaneously in the reweighted free energy surface ., If these states are truly metastable , a metadynamics simulation biasing the piston motion should in principle reveal them most efficiently ., We therefore performed a one-dimensional well-tempered metadynamics simulation on ., To prevent unfolding of the helices , we constrained the range of the piston shift to =\u200a−1 . 4 Å as a lower bound and =\u200a1 . 2 Å as the upper bound ., The resulting free energy profile is shown in FIG . 5-C0 and shows two free energy minima ., One minimum is located at =\u200a−0 . 25 Å and corresponds to the native conformation of the wild type , the state ., The other minimum is located at =\u200a1 . 1 Å and corresponds to the state ., FIG . 4 shows a representative conformation of the state ., The reweighted free energy surface as a function of of N1 and C2 in FIG . 5-C1 , C2 demonstrates that correlated piston shifts occurred only for the pair that contains the helix on which the bias was applied ., The other pair did not undergo piston motions ., FIG . 5-C3 shows the free energy profile as a function of the inter-monomer tilt angle and the piston shifts ., An increase in the tilt angle of to is correlated with an increase in the piston shift of C2 from 0 Å to −1 . 1 Å ., The free energy surface as a function of the helical rotation angle shows again that the change in rotation is uncorrelated to the change in piston , and furthermore that the change in rotation in N1 is positively correlated with the change in for C2 ., We have found a negative correlation between the piston shifts of the N-terminal helix of one monomer and the C-terminal helix of the other monomer , ( see FIG . 5-C1 ) ., To investigate this correlation further , we performed a two-dimensional metadynamics run , biasing both and ., FIG . 6-A1 shows the resulting two-dimensional free energy surface ., The profile reveals two minima that are very similar to the states and identified in the conventional MD study and the one-dimensional metadynamics simulation biasing a single piston shift ., The piston shift of the N1-helix is strongly anti-correlated with the piston shift of the C2-helix ., In FIG . 6-A2 , the reweighted free energy profile as a function of and shows that piston shifts in this helical pair are not correlated , as no change occurs for , when shows a piston shift ( see FIG . 6-A3 , A4 ) ., Biasing the piston shift of one helix resulted in enhanced piston shift of only one other helix , which has to be part of the other monomer and be at the other end of the protein chain ., We performed a two-dimensional metadynamics simulation on and to further investigate whether piston motions between N-terminal helices are truly not correlated ., The resulting free energy profile is shown in FIG . 6-B2 and reveals four minima ., Three minima , , and have also been observed as well in the conventional MD study and represent respectively states in which no piston shift has occurred , a piston shift has occurred in the N1-C2 pair , and a piston shift has occurred in the N2-C1 pair ., In addition , this free energy surface contains an extra minimum at ( =\u200a1 Å , =\u200a1 Å ) in which both helical pairs have undergone a piston shift ., Biasing one helical pair does not result in the inter-monomer tilt angle changing in a concerted way with the changes in piston shift ., In FIG . 6-A3 , A4 , for or , the inter-monomer tilt angle rests at ., The occurrence of the state goes hand-in-hand with an increase of the inter-monomer tilt angle from of WT to ( see FIG . 6-B3 , B4 ) ., This shows that , although we bias on piston shifts , changes in the inter-monomer tilt angle occur spontaneously ., To further explore the state , we performed MD simulations of this piston-shifted state , see FIG . S5 in Text S1 ., The state is only meta-stable , as it returns to either the state or the state within nanoseconds ., Our molecular dynamics simulations show that the structure of wild type Af1503-HAMP is very stable at room temperature , whereas the simulations of the A291F mutant show that the NMR structure of this variant is not at all stable at room temperature ., For A291F-HAMP , we found that the system either shows loss of helical structure or can relax to a conformation similar to A291F-HAMP fused to a DHp domain 19 ., The structure of the A291F mutant was suggested as an alternative conformation for HAMP , as increasing the volume of the hydrophobic sidechain at position 291 would change the packing in the hydrophobic core from complementary to ., If the packing would truly be an additional metastable state for HAMP , this structure would be as stable as the one assumed by wild-type Af1503-HAMP ., However , our simulations showed that this structure relaxes either to the native conformation , or to a conformation or with an upward piston shift of the N-terminal helix in one monomer and a downward piston shift of the C-terminal helix in the other monomer ., The metadynamics simulations aimed at exploring the equilibrium free energy landscape of HAMP revealed an additional stable state when the piston shift was biased ., This additional metastable state shows a strong similarity to the piston-shifted state found in the WT* simulations ., In this state , the N-terminal helix of one monomer moves up , and the C-terminal helix of the other monomer moves down ., Cysteine crosslinking studies of the HAMP domain in the Aer sensor protein , which senses the intracellular redox potential , resulted in a similar observation , in which a correlation between the N-terminal helix of one monomer and the C-terminal helix of the other monomer is observed 17 ., The metadynamics simulations sampled this correlated motion for both symmetry related pairs ., Moreover , the two-dimensional metadynamics simulation that biased both correlated motions uncovered the existence of an additional state in which both helical pairs in the piston shifted conformation ., These piston-shifted states also show an increased inter-monomer tilt angle ., Biasing directly on the tilt collective variable resulted in the spontaneous sampling of the state , showing that the piston shift and change in tilt angle are strongly coupled ., The role of helical rotation is less clear , as our results seem to indicate that changes in rotation occur independently with respect to changes in the piston shifts or tilt angles ., Directly biasing the rotation angle results in small changes of the other rotation angles , but not in visiting an additional conformational state ., As we calculated the rotation as a single value for an entire helix , we cannot directly compare our results with the gearbox model ., To this end , we extracted 40 snapshots from each piston-shifted state and computed several properties related to four-helical bundles for each residue , using the SamCC software 4 , 7 ., One of these properties is the Crick angle per residue , measured as the angle between the center of the bundle , the -atom of the residue and the point on the principal axis of the helix closest to the residue ., The Crick angles are known for an ideal helical bundle , so we can also measure the deviation from the ideal packing ., In FIG . 7 we show the deviation from the ideal Crick angle for all four piston states ., In all states the overall deviation is close to for the N-helices and for the C-helices and does not come close to zero in any of the states ., The curves representing the helices in the state are very similar to the curve measured for the NMR structure of wild-type Af1503-HAMP ( dashed lines ) ., The four states exhibit s

Introduction, Results, Discussion, Methods

The HAMP domain is a linker region in prokaryotic sensor proteins and relays input signals to the transmitter domain and vice versa ., Functional as a dimer , the structure of HAMP shows a parallel coiled-coil motif comprising four helices ., To date , it is unclear how HAMP can relay signals from one domain to another , although several models exist ., In this work , we use molecular simulation to test the hypothesis that HAMP adopts different conformations , one of which represents an active , signal-relaying configuration , and another an inactive , resting state ., We first performed molecular dynamics simulation on the prototype HAMP domain Af1503 from Archaeoglobus fulgidus ., We explored its conformational space by taking the structure of the A291F mutant disabling HAMP activity as a starting point ., These simulations revealed additional conformational states that differ in the tilt angles between the helices as well as the relative piston shifts of the helices relative to each other ., By enhancing the sampling in a metadynamics set up , we investigated three mechanistic models for HAMP signal transduction ., Our results indicate that HAMP can access additional conformational states characterized by piston motion ., Furthermore , the piston motion of the N-terminal helix of one monomer is directly correlated with the opposite piston motion of the C-terminal helix of the other monomer ., The change in piston motion is accompanied by a change in tilt angle between the monomers , thus revealing that HAMP exhibits a collective motion , i . e . a combination of changes in tilt angles and a piston-like displacement ., Our results provide insights into the conformational changes that underlie the signaling mechanism involving HAMP .

For survival , bacteria must constantly monitor their environmental conditions and adapt to these by generating a response ., Protein sensors enable bacteria to perceive their surroundings and are typically built from modular compounds that are connected by linker regions ., The HAMP domain is such a linker region that relays signals between different modules in a sensory cascade ., HAMP is a dimer comprising four helices in a parallel coiled-coil interaction motif ., One of the hypotheses explaining the mechanism of signal communication by HAMP is that the domain can adopt different stable conformations ., In this work , we used a molecular simulation approach to investigate this hypothesis at high atomic resolution ., We found that HAMP can adopt different conformations and that , in doing so , the helices shift and tilt with respect to each other ., Furthermore , we found that if one helix moves upward , the helix at the other end in the other monomer moves down .

biochemistry, biochemical simulations, computational chemistry, protein interactions, molecular dynamics, proteins, protein structure, chemistry, biology, computational biology, biophysics simulations, biophysics

journal.pgen.1002871

Mutational Signatures of De-Differentiation in Functional Non-Coding Regions of Melanoma Genomes

Sporadic cancer is mainly caused by the progressive accumulation of genomic mutations ., Therefore , a mechanistic understanding of cancer requires a comprehensive catalog of all somatic variants in a tumor genome ., Although the majority of somatic variants occur in non-coding regions of the genome , most studies have focused on interpreting genic mutations 1 , even when whole-genome data was generated 1–6 ., As a consequence , it is unclear if and how non-coding variants might contribute to cancer progression ., To comprehensively study functional consequences of somatic variants , one needs cell cultures made from the tumor ., First , though , one needs to know how representative the cell culture is compared to the original cancerous tissue ., Here we characterize these differences and use comparative and functional genomics methods to assess how mutations are distributed within melanoma genomes ., We used a combination of data produced by the Illumina GAIIx and HiSeq2000 platforms to generate over 5 . 4 billion 100 bp reads representing three different high-coverage genomes ( Figure 1A and S1 ) from the same 33 year old untreated male: two genomes represent a cutaneous melanoma sample , one of a laser capture microdissected metastatic tumor from the shoulder ( primary tumor is of unknown origin ) , and the other from a low-passage cell-culture derived from that tumor ., We also generated a matched “normal” genome from a blood sample ., Using our single nucleotide genotype calling methodology 7 , we were able to make confident genotype calls at 92 . 9% , 84 . 5% , and 95 . 6% of the tissue , cell culture , and normal genomes , respectively ., To accurately and comprehensively identify novel somatic single nucleotide variants ( SSNVs ) in the cell culture and tissue genomes we developed a new computational algorithm , which was validated and shown to have high sensitivity and specificity ( see Materials and Methods ) ., Utilizing published algorithms 8–10 , we were also able to identify somatic copy number changes and chromosomal rearrangements ., Comparing the somatic alterations identified in the tissue and cell culture genomes reveals their extent of relatedness ( Figure 1 and S2 ) ., In total , we identified 105 , 460 SSNVs in the tissue and 122 , 837 in the cell culture that were not present in the patients non-tumor DNA ., This number of somatic mutations is substantially higher than other published whole-genome cancer studies 1–6 ., If we examine genomic regions that have sufficient coverage to make a reliable call in both samples ( 81 . 1% of the genome ) , 95 . 2% of the sites are common ( 2 . 9% and 1 . 9% are unique to the tissue and cell culture , respectively ) ., The two melanoma samples are less concordant at the level of copy number variations ( CNVs ) relative to the normal genome ( Figure 1C and 1D ) ., In total , 118 Mb of the cell culture has somatic CNVs whereas only 63 Mb of the tissue does ., In support of these results , we found that aCGH CNV calls were highly concordant with our whole-genome sequencing-based calls ( Figure S3 ) ., One striking difference in the cell culture genome is that it includes a near-complete loss of one copy of chromosome 14 ( Figure S2 and S4 ) ., The additional CNVs in the cell culture genome may result from the low-passage culturing process ., This is a known phenomenon that has been previously documented in higher-passage hESC cell cultures 11 and a xenograft of a primary tissue cancer sample 6 ., As such , our CNV results are consistent with other reports and extend these findings to lower-passage tumor cell cultures ., Because non-normal CNV regions can influence SSNV calls , we recalculated concordance at non-CNV regions ., Focusing on these areas , there are 91 , 823 SSNVs in the union of both samples , and 96 . 1% are shared ( 2 . 0% and 1 . 9% are unique to the tissue and cell culture , respectively ) ( Figure 1B ) ., The SSNV mutational spectrum is reflective of UV damage , even for cell culture and tissue-specific calls ( Figure S5 ) ., We additionally made somatic insertion and deletion ( indel ) calls ( see Materials and Methods ) , and found that after CNV filtering there are 269 somatic indels shared between the tissue and cell culture , while the tissue has 127 unique indels and the cell culture has 160 ( Figure S6 ) ., This lower level of concordance , relative to SSNVs , between calls is not surprising , as previous studies show that indel calling is more difficult with short reads 12 ., Together , these results provide a high-resolution picture of the differences between a metastatic tissue sample and the cell culture derived from it ., We next compared mutations from another melanoma whole-genome study by applying our computational SSNV detection method to sequence data from metastatic melanoma ( colo-829 ) and matched normal ( colo-829BL ) cell lines 1 ., We identify more SSNVs than originally reported , and the bulk of our calls are concordant with those ( Figure S7 ) ., Importantly , we identify 448 of the 454 ( 98 . 7% ) Sanger-validated and 40 of the 43 ( 93% ) COSMIC calls in the colo-829 genome ., Variant calls that are specific to our algorithm are enriched for the characteristic melanoma UV mutational signature ( Figure S8 ) ., We observe 100% concordance with Sanger sequencing-based cross-validation of novel SSNV calls at 181 positions in the cell culture genome ( see Materials and Methods ) , which suggests that our SSNV detection algorithm has a low false positive rate ., Additionally , we randomly selected 96 cell culture-specific and 96 tissue-specific SSNVs for PCR amplification and Sanger sequencing ., Of the successful PCR and Sanger sequencing reactions , we observe 97 . 7% concordance and 98 . 7% concordance at tissue-specific and cell culture-specific positions , respectively ., Together , these results suggest our SSNV detection algorithm is both highly sensitive and specific ., Comparison of the colo-829 SSNVs to those from our melanoma sample shows commonly mutated genes , some of which are associated with melanoma pathogenesis ( Figure S9 ) ., For example , missense mutations ( D261N and H533Y ) were identified in ADAM29 , which encodes a member of the A Disintegrin And Metalloproteinase ( ADAMs ) family which are membrane anchored glycoproteins with several biological functions encompassing cell adhesion , cell fusion and signaling 1 ., Importantly , we recently reported that a systematic mutational analysis of all members of the ADAM family of membrane-bound metalloproteases showed that ADAM29 is often mutated in melanoma 13 ., Functional analyses have indicated that ADAM29 mutations affect adhesion of melanoma cells to specific extracellular matrix proteins , suggesting that mutated ADAM genes play a role in melanoma tumorigenesis 13 ., This study also identified a missense mutation ( R175C ) in PTK2B , which encodes the non-receptor protein tyrosine kinase PTK2B , also known as PYK2 or FAK2 , a focal adhesion protein that shares structural similarity with its paralog focal adhesion kinase 1 ( FAK1 ) ., PTK2B has been previously linked to metastasis via RhoC-dependent activation of FAK1 , MAPK , and Akt 14 ., As we previously reported a high prevalence of somatic mutations in PTK2B in metastatic melanoma 15 , these studies suggest that PTK2B may be a melanoma cancer gene and that further studies are required to more fully characterize the functional role of its mutations in melanoma ., For additional genic annotations , we include a supplementary file that outlines all coding mutations discovered in this study ( Table S1 ) ., Because metastatic tumor formation involves successive iterations of mutation , followed by selection and clonal expansion , the resulting cell population has undergone an evolutionary process commonly referred to as clonal evolution 16 , 17 ., When measuring the similarity of sequences across many species , the genome has clear signatures of intense selective pressure 18–23 ., Some regions reject mutations more than expected ., To determine if the selective forces operating on a metastatic cell over the span of cancer development are similar to those operating across species over millions of years , we compared somatic mutation accumulation in melanoma to evolutionary constraint ., For this analysis , we combined the SSNVs from our tissue sample with those we identified in the colo-829 cell line ., This resulted in 141 , 655 unique SSNVs , of which 99 . 3% are non-coding ., To determine if these mutations are uniformly distributed throughout the genome , we first measured mutation accumulation in functionally different regions identified by chromatin-based chromosomal segmentations 24 ( Figure 2A ) ., Such segmentations currently exist for nine different cell types ( Figure S10 ) , and we chose NHEK cells as our primary focus since these appear most similar to melanoma cells out of all nine cell types ( see below ) ., The enrichment results are consistent when the samples are analyzed independently ( Figure S11 ) ., There is a clear anti-correlation with evolutionary constraint ( Figure 2B and S12 ) ., However , there is also a strong anti-correlation with mutation accumulation and coding regions ( Figure 2C and S13 ) , which is expected due to transcription-coupled repair ( TCR ) ., Of note , the heterochromatin low signal regions ( state 13 in the chromosomal segmentations ) accumulate mutations roughly equal to random expectation ( Figure S14 ) , indicating that they may be suitable targets for estimating the background passenger somatic mutation rate ( which for this tumor we calculate as about 42 SSNVs per megabase ) ., The above results indicate that somatic mutations do not occur uniformly across the genome ., To eliminate the mutation suppression bias related to TCR in known genic areas , we specifically focused on regions of the genome less likely to be transcribed— windows that do not overlap and are greater than 10 Kb from annotated genes or transcription start sites ( TSSs ) ., We performed a multiple regression on mutation accumulation in bins of these regions using evolutionary constraint , GC content , and fraction of transcribed bases , which we obtained from a separate melanoma RNA-seq study 25 ., Adding the additional variables removes the correlation with evolutionary constraint ., Unsurprisingly , the fraction of non-coding bases transcribed ( one of the variables in the above-mentioned regression analysis ) is almost perfectly anti-correlated with enrichment for mutation accumulation ( Spearmans R\u200a=\u200a−0 . 97789; P<2 . 2e-16 ) ., These results suggest that TCR is a mechanism associated with preventing mutation accumulation in non-coding regulatory elements ., We next sought to examine the distribution of somatic mutations across experimentally-derived functional non-coding regions ., To do this , we compared our SSNV collection to broad classes of active regulatory elements identified by the DNaseI hypersensitive site ( DHS ) assay 26–30 ., This experiment was performed genome-wide on melanocytes—the precursor cell type to melanoma—as part of the ENCODE Project Consortium 31 ., We hierarchically partitioned melanocyte DHSs based on genic landmarks and calculated somatic mutation enrichment ( Figure 3A ) ., All DHS categories except for 3′ UTRs are significantly less enriched than random expectation ( horizontal line at 0 ) and compared to common SNPs from the 1000 genomes consortium ( grey points ) ., 5′ UTRs are the most depleted ., These results are consistent with the observed increase in mutation accumulation along the length of genes ( Figure S15 ) and are reproducible when the samples are analyzed independently ( Figure S16 ) ., Despite their distant location from known transcribed regions , intergenic TSS-distal DHSs are also significantly depleted for accumulating mutations ., To avoid confounding from transcription-coupled repair ( described above ) , we subsequently focus on intergenic TSS-distal DHSs ., We performed single linkage clustering of DHSs from 29 different cell states ( cell types and conditions ) identified by the ENCODE Project Consortium 31 to identify sites that are cell-type-specific , present in a combination of cell types , or ubiquitously present ., Out of all cell-type-specific DHSs , the most depleted for mutation accumulation are those specific to melanocytes , aortic smooth muscle cells ( ASMCs ) and H1 embryonic stem cells ( ESCs ) ., Ubiquitously present DHSs are even more depleted ., We next calculated the mutational load on all melanocyte DHSs by measuring mutation accumulation when these regulatory regions are active in all possible cellular contexts/combinations ., Unsurprisingly , mutation enrichment decreases as the melanocyte DHS is active in more cell types ( Figure 3C; yellow line ) ., As a control for this experiment we examined all combinations of non-melanocyte DHSs and found a similar trend ( Figure 3C; blue line ) , although not as depleted as the melanocyte DHSs ., These results indicate that regulatory regions are preferentially repaired in metastatic melanoma and that this occurs in a cell-type specific manner ., To further understand the relationship between cell type regulatory architecture and somatic mutation enrichment in metastatic melanoma , we clustered all 29 cell types based on their regulatory element signatures ( Figure 3D ) ., Note the relationship between melanocytes and the other two cell types where cell-type-specific DHS mutations are highly depleted ( ASMCs and ESCs ) ., ASMCs are derived from the same embryological layer—neural crest—as melanocytes , and ESCs are an undifferentiated pluripotent cell type ., Brain cell ( medulloblastoma ) -specific DHSs , which are also neural crest derived , show significant depletion as well ., The topology of the tree and the significant depletion for somatic mutation accumulation in regulatory regions specific to neural crest-derived and ESC cell types suggests that the metastatic melanoma cell utilized these regulatory programs ., These results imply that the regulatory architecture of the metastatic melanoma cell de-differentiated to a more basal cellular program that is visible in the pattern of mutations covering cell-type-specific regulatory regions ., In support of this hypothesis , a recent study found that human melanoma-initiating cells express a neural crest stem cell marker 32 ., To experimentally test the hypothesis of regulatory de-differentiation , we performed genome-wide DNase-Seq to identify DHSs in colo-829 and the cell culture sample sequenced in this study ., Generating trees using these two samples and DHSs from the other cell types shows that the two melanoma samples are closely related to each other and melanocytes ( Figure 4A ) ., However , focusing on gene regulatory status by only considering DHSs that overlap exonic regions shows a different tree topology ( Figure 4B ) ., Here , the melanomas are de-differentiated relative to the melanocyte sample ., It is known that highly transcribed genes accumulate fewer somatic mutations relative to more lowly transcribed genes 1 ., Thus , the extent of TCR depends on the level of transcription ., Recent studies show that non-coding functional elements are transcribed 33–36 ., So , one would expect that somatic mutation accumulation in these regions could be modulated by whether or not , and to what extent , they are transcribed ., To determine if TCR might operate at melanoma-specific TSS-distal non-genic regulatory regions , we calculated how many of these sites are transcribed ( Figure 4C ) ., We found that melanoma-specific regulatory regions are significantly more likely to be transcribed ( P<2 . 2 e−16; Fishers Exact Test ) relative to melanocyte-specific regulatory regions ., To further investigate this , we searched for a hallmark signature of TCR—repair events biased to the transcribed strand ., Focusing on SSNVs overlapping the melanoma regulatory elements we identified that occur within introns ( so that we can orient mutations relative to the transcribed strand ) , we observe a significant ( P\u200a=\u200a0 . 001605; exact binomial test ) strand bias ( Figure 4D ) ., These are the first results to our knowledge that demonstrate the regulatory architecture at non-coding regions in cancer genomes is de-differentiated and likely shaped by TCR ., Here we have used whole-genome sequencing to identify the somatic mutations in a metastatic melanoma tissue sample and a low-passage cell culture derived from the same patient ., We speculate that the mutational signatures in the metastatic cell indicate that regulatory architectures of the precursor cell it was derived from and other basal cellular programs were utilized during the path to metastasis—consistent with a tumorigenesis model of embryonic program redeployment ., A pathology-confirmed metastatic melanoma tumor resection , paired with a pheresis-collected peripheral blood mononuclear cells , was collected from a 33 year old melanoma patient enrolled in IRB-approved clinical trials at the Surgery Branch of the National Cancer Institute ., A portion of the fresh tumor was frozen and embedded in Optimal Cutting Temperature ( OCT ) embedding medium ., A melanoma cell line was derived from mechanically dispersed tumor cells , which were then cultured in RPMI 1640+10% FBS at 37°C in 5% CO2 for 9 passages ., Genomic DNA was isolated using DNeasy Blood & Tissue kit ( Qiagen , Valencia , CA ) ., Several quality controls were performed one of which was the use of cytopathology , to determine the percentage of melanoma antigen expressing cells ., The tissue culture line used in this study was evaluated by immunohistochemistry to have at least 75% of cells express melanoma-specific antigens ., This threshold was set as it has been reported to give sufficient purity to identify regions of homozygous deletion , hemizygous deletion , copyneutral LOH , duplication and amplification 37–39 ., Genotyping of the samples was performed to verify that they are derived from the same individual ., H&E stained sections of fresh frozen melanoma tissues are prepared for initial histologic assessment ., Sections are examined by a pathologist for the presence of tumor , estimation of tumor content , presence of inflammation and necrosis ., Tissues with less than 70% tumor and/or significant areas of inflammation and necrosis are subjected to LCM ., Laser capture microdissection ( LCM ) was performed in the Pathology Core Facility of MSKCC , New York , NY , using the Veritas Microdissection System ( Arcturus ) ., The Veritas system combines ultraviolet laser cutting and laser capture using an infrared laser source ., Fresh frozen melanoma tissues sectioned between 8 and 10 µm were transferred to PEN membrane slides ( MDS Analytical Technologies ) and sections were stained by using a modified protocol described previously 40 , 41 ., Briefly , sections were stained with hematoxylin as follows: slides were immersed in 70% ethanol for about 10 min followed by sequential dips in nuclease free water , Mayers hematoxylin solution for 30 sec , nuclease free water , 75% ethanol , 95% ethanol and finally dehydrated in absolute ethanol by 3 changes of 3 min each ., Multiple serial sections ( 10–20 ) of the tissue are used to maximize cell yields ., 5 , 000 to 10 , 000 cells were harvested in each LCM cap and material from 5–10 caps was pooled together to maximize yields ., DNA was extracted using DNeasy Blood and Tissue kit ( Qiagen ) following manufacturers instructions ., DNA was eluted in 35 ul of elution buffer ., DNA measurements were made using ND-1000 UV-Vis spectrophotometer from NanoDrop technologies ., We generated 5 , 409 , 104 , 173 100 base paired-end reads that pass the Illumina chastity filter and contain 32 or more Q20 Sanger-scaled quality bases for this study , which were partitioned among the genomes as follows: 1 , 042 , 502 , 044 for the cell culture , 1 , 588 , 246 , 159 for the tissue , and 2 , 778 , 355 , 970 for the normal ., Reads were aligned to the unmasked hg18 version of the human genome using BWA 42 with default parameters ., After removing molecular duplicate read pairs ( read pairs that map to the same position on the reference sequence are likely an artifact of sample preparation ) using samtools 43 and considering only reads with a mapping quality of Q30 or greater and bases with quality of Q20 or greater , we observe an average base coverage of 21 . 4× , 29 . 6× , and 47 . 7× for the cell culture , tissue , and normal genomes , respectively ( Figure S1 ) ., Within coding regions , we were able to make confident variant calls ( see details below ) at 64 . 3% , 85 . 2% , and 88 . 9% of the positions in the cell culture , tissue , and normal genomes , respectively ., Comparing territory that is callable in the cell culture and tissue results in 81 . 1% genome coverage ., For variant calling , only reads with mapping quality of Q30 or greater and bases with quality of Q20 or greater were considered ., We used two related algorithms to make single-position genotype calls in the normal and melanoma genomes ., For all genomes , we use a Bayesian genotype caller named Most Probable Genotype ( MPG ) that has been described previously 44 ., This genotype caller produces accurate calls in regions that satisfy whole-genome coverage and quality parameters as determined by a separate study 7 ., Namely , the MPG score must be equal or greater than 10 and the MPG score to base Q20 quality-coverage ratio must be equal to or greater than 0 . 5 ., To independently verify MPG calls , we compared genotypes to those called by the Infinium 1M quad SNP-chip platform ., The genotype concordance rate with the SNP-chip for the normal genome is 99 . 937% at 99 . 3% of the positions , excluding regions with hidden SNPs 45 and abnormal copy number ., A similar comparison performed on the cell culture genome results in 99 . 939% concordance at 91 . 2% of the positions ., To better identify variant positions in the cell culture and tissue genomes , we first developed a new algorithm similar to MPG , called Most Probable Variant ( MPV ) ., An important distinction between MPG and MPV is that the MPV score reflects the degree of confidence that a sample has a genotype different from the reference genome , whereas the MPG score reflects the degree of confidence in the genotype call itself ., MPV is a new option ( –score_variant ) in the MPG program and the executable source code is freely available for download from the following URL: http://research . nhgri . nih . gov/software/bam2mpg/ ., We optimized calling parameters for MPV by downloading and analyzing genome-wide tumor and normal data that was previously published for the colo-829 melanoma and colo-829BL normal cell lines 1 ., Using MPV with optimized parameters ( MPV score must be greater than or equal to 10 with no coverage ratio criteria similar to the MPG parameters ) on the cell culture genome allows us to identify more variant positions without dramatically sacrificing accuracy ( Table S2 ) ., Comparing MPV calls for the cell culture genome to the SNP-chip results in 99 . 79% concordance at 96 . 28% of the variant positions ., To identify novel somatic single nucleotide variant ( SSNV ) positions we compared the MPV-called genotype in either melanoma genome to the MPG-called genotype in the normal genome and then subtracted out any variants that are present in dbSNP129 or within ten bases of an indel identified by the MPV algorithm ., Loss of heterozygosity ( LOH ) variants were ignored since there is no novel somatically-acquired allele ., Running our analysis pipeline on our own samples resulted in 122 , 837 SSNVs in the cell culture genome and 105 , 460 in the tissue genome ., It is important to note that these two numbers are not comparable because they are not normalized across the common callable territory in the cell culture and tissue genomes ., Once we account for this , the somatic variant counts drop to 97 , 532 for the cell culture and 98 , 548 for the tissue ., We validated novel SSNVs by PCR amplifying the regions in the cell culture and normal genomes and then Sanger sequencing the products ., Of 192 randomly chosen positions ( 96 in coding regions , and 96 in non-coding regions ) , we were able to successfully PCR amplify and Sanger sequence 181 in both genomes ., Of these , we observed evidence for somatic variants concordant with the whole-genome data at 100% of the positions ., For further validations we randomly selected 96 cell culture-specific and 96 tissue-specific SSNVs for PCR amplification and Sanger sequencing ., Of the 78 successful PCR and Sanger sequencing reactions for the cell culture set , 75 ( 96% ) had genotype calls concordant with the whole-genome sequencing call ., For the tissue set , 43/73 ( 59% ) were concordant ., This result allowed us to focus on the 30 positions where the tissue-specific whole-genome calls were not concordant with the PCR and Sanger calls ., We found that by implementing three simple filters , we eliminated 29 of 30 discordant positions and 0 of 43 concordant positions , so that the concordance rate is 43/44 ( 97 . 7% ) ., The filters we implemented are: These filters removed 2 of 3 discordant cell culture-specific calls and 0 of 75 concordant calls , so that the concordance rate is 75/76 ( 98 . 7% ) ., We additionally looked at the mutation spectrum for all the common , tissue-specific , and cell culture-specific SSNVs ( Figure S5 ) ., All three mutation spectrums are enriched for the known G>A/C>T UV signature ., Together , these results suggest that our method is highly specific ., To estimate the extent of normal cell contamination in the cell culture and tissue samples , we calculated the fraction of reads with mapping quality of at least 30 supporting the acquired somatic allele at heterozygous positions and compared this to what would be expected in a completely homogenous cellular population with no normal cells ., Our analyses show that the cell culture has no normal contamination , while the tissue sample contains about 42% normal cells ( Figure S17 ) ., Other groups have used similar methods to estimate tumor sample purity 46 ., We first estimate copy number in non-overlapping 5 Kb tiles in the normal genome using the copySeq algorithm 8 ., We only consider tiles with 80% uniquely mappable k-mers ( which is 94% of the tiles ) to ensure accurate copy number estimation ., To detect amplifications and deletions in the cancer genomes we use the CNV-seq algorithm 9 , which compares the cancer to normal genome , with the following parameters: Somatic copy number alterations ( SCNAs ) are then defined over the 5 kb tiles called in the normal genome using CNV-seq results and copy number is adjusted based on the level of normal genome contamination , as described above ., Adjacent amplified or deleted 5 kb windows are merged and only regions where two or more windows are affected are retained ., To conservatively identify tissue or cell culture-specific CNVs , we filtered the CNV-seq calls in one sample by looking at the corresponding log2 ratio in the other sample ., Any CNV-called regions in one sample with a CNV log2 ratio<\u200a=\u200a−0 . 1 or >\u200a=\u200a0 . 1 in the other sample were considered CNVs even if they were not called by the CNV-seq algorithm ., Thus , these regions are not considered sample-specific , which result in a conservative set of CNV calls ., We made somatic insertion and deletion calls by extending our MPV and MPG scoring methodology ( see Single nucleotide variants section above ) to indel calls ., We first select all possible non-reference indel calls , irrespective of score threshold , across all three genomes using MPG on the normal genome and MPV on the tumor genomes ., After merging all possible calls , we then look at each genome independently and determine how well the reads support the indel call using MPG on the normal genome and MPV on the tumor genome , both with thresholds of 10 ., To find somatic indels we keep non-reference tumor calls that do not match the normal call at that same position ., Because CNVs can bias indel calls , we subsequently filter by retaining regions where the CNV log2 ratio >−0 . 1 and <0 . 1 such that a conservative set of indels outside CNV regions are compared ., The final indel results are summarized in Figure S6 ., We used an Agilent 180K aCGH array to look for CNVs in the tissue sample ., For gain/loss calls , we used the default Nexus 6 . 0 settings for Agilent 180K catalog arrays for mosaic tissue samples , and adjusted the minimum probe bin size to 10 instead of the default 3 for segmentation We used the BreakDancer algorithm 10 to detect chromosomal rearrangements ., In order to detect somatic events we require a score of 90 or greater in the cancer genome , which is consistent with parameters reported in a previous study 6 and no evidence in the normal genome ., We further filter the results in two ways ., First , by removing any somatic events that occur in any of ten normal genomes from an ongoing internal study ( data not shown ) ., Second , by requiring that the 2 kb region immediately surrounding each putative breakpoint is greater than 99% mappable according to the CRG 100mer alignability track available at the UCSC genome browser ., We used the Genome Structure Correction ( GSC ) method 47 to calculate enrichment statistics for SSNVs relative to other genomic features ., All results are based on 10000 samplings and reported as the log2 fraction of observed base overlaps divided by the mean of the null overlaps ., Error bars represent +/− one standard deviation from the mean of the null distribution ., We calculate enrichment or depletion only in situations where ten or more SSNVs overlap a particular set of genomic features ., If there are fewer overlaps , we consider the calculation unreliable and therefore ignore those comparisons ., Common SNP control data sets were constructed using 1000 Genomes calls 48 at positions that have a minimum of 5% minor allele frequency ( MAF ) and are concordantly called across four different centers ., Similar results are observed when using 20% MAF SNPs ( data not shown ) ., Chromatin segmentation data for nine different cell types was obtained from Ernst et al . 24 ., We ignored states 14 and 15 , which correspond to repetitive regions of the genome ., Variant calls are generally filtered out of these areas by the 1000 Genomes Consortium because they result in high false positives rates ., These two states combined occupy 0 . 27% of the genome on average over the nine different cell types , so ignoring them will have little effect on our analyses ., We divided genomic features into hierarchical and mutually exclusive categories based on the following hierarchical sequence of genic landmarks: coding regions , 5′ UTRs , 3′ UTRs , introns , intergenic transcription start site ( TSS ) -proximal ( within 5 , 000 bp of a TSS ) , and intergenic TSS-distal ( greater than 5 , 000 bp from a TSS ) ., All genic landmarks are based on the GENCODE annotation 49 in hg18 and can be downloaded from the UCSC Genome Browser genome . ucsc . edu ., We masked out all regions of the genome overlapping with , or within 10 , 000 bp , of any part of a gene or TSS ., For the remaining parts of the genome , we created 50 , 000 bp non-overlapping tiles and calculated the number of bases that overlap evolutionarily constrained regions ., Constrained regions are based on the GERP method 20 and the Enredo , Pecan , Ortheus ( EPO ) alignments 50 , 51 and are available at the Ensembl browser www . ensembl . org ., We discarded tiles with no constrained region overlap and sorted the remaining tiles by the fraction of constrained base overlaps ., Using this sorted list , we created ten equal-sized bins and calculated mutation accumulation enrichment ( see above ) for the tiles within each bin ., We used post-embargo ENCODE Consortium DHS data sets for the following 29

Introduction, Results/Discussion, Materials and Methods

Much emphasis has been placed on the identification , functional characterization , and therapeutic potential of somatic variants in tumor genomes ., However , the majority of somatic variants lie outside coding regions and their role in cancer progression remains to be determined ., In order to establish a system to test the functional importance of non-coding somatic variants in cancer , we created a low-passage cell culture of a metastatic melanoma tumor sample ., As a foundation for interpreting functional assays , we performed whole-genome sequencing and analysis of this cell culture , the metastatic tumor from which it was derived , and the patient-matched normal genomes ., When comparing somatic mutations identified in the cell culture and tissue genomes , we observe concordance at the majority of single nucleotide variants , whereas copy number changes are more variable ., To understand the functional impact of non-coding somatic variation , we leveraged functional data generated by the ENCODE Project Consortium ., We analyzed regulatory regions derived from multiple different cell types and found that melanocyte-specific regions are among the most depleted for somatic mutation accumulation ., Significant depletion in other cell types suggests the metastatic melanoma cells de-differentiated to a more basal regulatory state ., Experimental identification of genome-wide regulatory sites in two different melanoma samples supports this observation ., Together , these results show that mutation accumulation in metastatic melanoma is nonrandom across the genome and that a de-differentiated regulatory architecture is common among different samples ., Our findings enable identification of the underlying genetic components of melanoma and define the differences between a tissue-derived tumor sample and the cell culture created from it ., Such information helps establish a broader mechanistic understanding of the linkage between non-coding genomic variations and the cellular evolution of cancer .

Here we investigate the relationship between somatic variants and non-coding regulatory regions ., To do this , we develop a new algorithm for identifying single nucleotide somatic variants in whole-genome sequencing data and apply it to a metastatic melanoma sample and a cell culture derived from this sample ., Our results show that the two genomes are similar at the level of single nucleotide changes and more variable at larger copy number changes ., We further observe that patterns of somatic mutation accumulation in non-coding regulatory regions suggests that the metastatic melanoma cells de-differentiated into a more basal regulatory state ., That is , by simply looking at mutation accumulation across cell-type-specific non-coding functional regions , one can clearly see patterns that are indicative of cell state de-differentiation ., Results from genome-wide functional regulatory region experimental mapping support this observation .

medicine, dermatology, functional genomics, genome sequencing, skin neoplasms, genome analysis tools, malignant skin neoplasms, personalized medicine, melanomas, comparative genomics, biology, genetics, genomics, genetics and genomics, human genetics

journal.pcbi.0030255

Determination of the Processes Driving the Acquisition of Immunity to Malaria Using a Mathematical Transmission Model

Plasmodium falciparum malaria continues to be a major cause of human morbidity and mortality , especially in Africa , but varies greatly in endemicity across the continent and elsewhere 1 ., The consequent variation in levels of acquired immunity and age-specific disease patterns complicates malaria epidemiology and means that control policies that are optimal for one setting are not easily translated to other settings ., In highly endemic areas where clinical immunity develops rapidly 2; there is concern that interventions which reduce transmission could also affect the development of immunity 3–6 ., A delay in the acquisition of immunity beyond early life has the potential to change the spectrum of serious clinical symptoms 7 , 8 and the lifetime risk of disease 4 ., While the processes that determine the acquisition of immunity to P . falciparum clearly impact on the epidemiology of the disease , they are complex and poorly understood due to the unclear relationship between immunological markers and functional immunity 9–11 ., However , there is evidence to suggest that both clinical ( anti-disease ) immunity and anti-parasite immunity develop at different rates ., For example , in people who emigrate from malaria endemic settings , clinical disease appears to emerge only in those who remain away for at least 3–5 y 7 , 12 ., Furthermore , these emigrants also present clinically with lower parasite densities than those who travel from non-endemic areas , suggesting that an additional component of immunity that regulates parasite densities may be longer-lived ., This hypothesis is also supported by analysis of age-stratified anti-malarial antibody seropositivity rates which gives estimates of half-lives that span decades 13 ., There is also evidence to suggest that acquired immunity does not only depend on exposure but is also influenced independently by age ., For example , there is evidence for an age-dependent exposure-independent maturation of the antibody response to malaria 14 , and this may in part explain the observation that the proportion of severe malaria cases presenting with severe malarial anaemia is more closely associated with age than with transmission intensity 15 , 16 ., Immune responses which affect subpatent parasitaemia may influence malaria transmission , but high rates of subpatent infection in high transmission areas suggest that acquired immune mechanisms capable of complete parasite clearance rarely develop in naturally exposed populations , so we allow for the possibility of subpatent tolerance ., Here we develop a mathematical model to better understand the impact of the development of immunity on observed epidemiological patterns , and also aspects of the immunology which might be inferred from the epidemiology such as time scales of acquisition and loss ., Whilst a number of malaria transmission models have been developed in the past which incorporate immunity 17–25 , each do so in different ways and hence make comparison between model structures difficult ., In contrast , we systematically explore the impact of immune responses at different points of the hosts natural history of infection which are then tested by comparing model output with epidemiological observations ., Our results demonstrate that more than one type of age- and transmission intensity-specific response are necessary to predict malaria epidemiological patterns , in line with current immunological understanding 7 , 9 , 10 , 26 ., We first developed an age-structured transmission model for malaria in which acquired immunity acts at three different stages of a hosts history of infection:, 1 ) susceptibility to symptomatic disease ( severe and clinical cases ) upon infection or re-infection , assuming susceptibility decreases with cumulative exposure to infectious bites ( e . g . , as a result of antibody-mediated strain-specific immunity ) ;, 2 ) natural recovery from asymptomatic to undetectable infection ( i . e . , effective clearance of parasites ) , which increases with cumulative exposure to infectious bites after a delay during childhood representing maturation of the immune system ,, 3 ) natural clearance of undetectable subpatent infection , assuming increased tolerance and slower clearance of such infection ., Each response , which we call an immunity function , is allowed to change with age and malaria transmission intensity ( commonly expressed as the entomological inoculation ( EIR ) ) and hence represents the acquisition and loss of immunity dependent upon exposure ., The first two immunity functions incorporate a memory component ( i . e . , allow for gradual loss in the absence of reinfection ) 27 , whereas the final immunity function ( associated with regulation of parasite density ) is assumed independent of acquired immunity , as subpatent parasites ( if any ) are kept subpatent by an effective immune response ., Figure 1A shows the patterns of parasitaemia and clinical disease by age observed in northern Tanzania ., These data were collected from 24 villages at three different altitude levels ( <600 m , 600-1200 m , and >1200 m ) and in two different regions 28 ., In one of the regions ( region 2 ) , estimates of malaria transmission intensity as measured by the EIR were also collected ., These varied by altitude with the highest transmission intensity occurring at low altitude ( 56 infectious bites per person years ( ibbpy ) , range 28–108 at <600 m , 3 ibppy , range 0 . 4–7 . 6 at 600-1200 m , and 0 . 12 ibppy , range 0 . 01–0 . 032 at >1200 m ) ., Although these data were not available in the other region , the patterns of parasite prevalence by age and altitude are similar ., Clinical data from severe malaria admissions to district , regional , and referral hospitals serving the Usambara mountain region ( region, 2 ) are shown in Figure 1B 15 ., Figure 1C–1D shows the prevalence of parasiteamia by age in locations on the north bank and south bank of River Gambia , The Gambia 29 ., Transmission in The Gambia is highly seasonal , and transmission intensity differs between the settings with higher intensity on the south bank ., The estimates are presented separately for the dry and wet seasons , with higher prevalence observed during peak transmission in the wet season ., The corresponding patterns predicted by different versions of the model are shown in Figure 2 ., If the model does not incorporate immunity at any point , we observe a rise in the prevalence of parasitaemia or clinical disease which saturates at older ages ( Figure 2A and 2B ) ., This clearly does not match the decline in both parasitaemia and clinical disease at older ages observed in data ( Figure 1 ) ., Allowing the model to incorporate immunity that results in increased persistence of subpatent infections ( immunity function, 3 ) gives rise to profiles that either peak too early in life and decay too rapidly at high EIRs or which saturate for low EIRs ( Figure 2C and 2D ) ., Allowing the model to incorporate immunity resulting in more rapid recovery from asymptomatic infections or symptomatic disease ( immunity function, 2 ) gives rise to patterns of parasitaemia that match those observed reasonably well ., However , the patterns of symptomatic disease decay too slowly with age ( Figure 2E–2F ) ., Finally , allowing the model to incorporate immunity that reduces the proportion of infections that result in clinical disease ( immunity function, 1 ) results in patterns of clinical disease that closely match those observed in the data but fails to reproduce the decline in parasitaemia with age ( Figure 2G–2H ) ., Other discrepancies between the model predictions and observed patterns of parasitaemia and disease by EIR and inconsistencies in lifetime episodes were also observed for each immunity function ( see Protocol S1 ) ., We next considered combining the different functions to identify which combination best reproduces the observed age-prevalence patterns in Figure 1 ., Combining immunity functions 1 and 2 ( i . e . , allowing a reduction in the proportion of infections that give rise to clinical disease and an increase in the rate of recovery from asymptomatic infection to subpatent infection ) reproduces well the age-prevalence of parasitaemia and severe disease observed in the study data ( Figure 2I and 2J ) ., It also reproduces the observed decrease in clinical cases in older ages as the EIR is increased ( see Protocol S1 ) ., Adding the third immunity function ( increasing persistence of subpatent infection ) results in patterns that more closely resemble those observed if this function alone drives immunity ( Figure 2C and 2D ) and therefore lessens the agreement between model predictions and observed data ., The age-prevalence patterns in Figure 2I and 2J resemble but do not exactly match those observed in data ( Figure 1 ) ., There are many reasons for not expecting an exact match: estimates of EIR are imprecise , and quoted values are averages over surveys and locations within altitude ranges; there may be random variation and unaccounted factors , such as bias in data sampling among age groups; and parasite density and detection at a given age may differ among sites ., However , we note that the model predicts age-parasitaemia curves which saturate with age for medium-to-low EIR , which is not observed in data ., Adjusting parameters does not seem to alter this feature ., However , if natural recovery from infection ( e . g . , from asymptomatic to subpatent ) is solely determined by age ( via physiological processes , provided there is exposure on which infection is conditional ) , we obtain patterns closer to those observed ( Figure 3 ) ., This suggests that parasite immunity in non-naïve individuals may be controlled by physiological development rather than by the amount of natural exposure ( provided there is exposure ) 7–9 , 14 , 15 , 30 ., An alternative way of testing the immunity functions ( conditional on the remaining model structure and assumptions being valid ) is to compare the predicted mean infectivity by age , which may be regarded as the probability of carrying gametocytes ( although not all gametocyte carriers will be infectious ) , with the observed age-prevalence of gametocytes ., The patterns predicted by our best model ( incorporating immunity functions 1 and, 2 ) closely match the patterns observed in northern Tanzania and The Gambia ( Figure 4 ) ., Since the model parameters were fixed or fitted to asexual parasite data , these results are an independent test of the models ability to reproduce observed epidemiological patterns ., Our determined half-lives of clinical and parasite immunity were 5 y and 20 y , respectively ., By varying these parameters , we explored whether patterns of age-prevalence can inform possible bounds for these parameters ., Reducing the half-life for the duration of clinical immunity below 5 y results in a sharp increase in the proportion of all infections that are symptomatic cases and , in addition , results in less-pronounced age-prevalence peaks which begin to deviate from those observed in data ., Increasing the duration of clinical immunity does not substantially change age-prevalence patterns but does have an impact on the proportion of infections that are symptomatic cases ( Figure 5A and 5B ) ., Reducing the half-life for the duration of parasite immunity below 20 y similarly has an impact on the age-prevalence curves and at very low values ( <10 y ) gives rise to curves that saturate rather than decline at older ages ., The proportion of infections that are asymptomatic and parasitaemic is also increased ., However , increasing the duration of parasite immunity has little impact on either outcome ( Figure 5C and 5D ) ., Our results demonstrate that , while distinct models can explain patterns of parasitaemia observed in individuals aged 0–5 y , in order to reproduce full age-prevalence patterns of parasitaemia and clinical disease observed in endemic malaria settings at least two distinct acquired immunity processes are required:, 1 ) an early age ( or early exposure ) reduction in clinical susceptibility , and, 2 ) a process of parasite immunity that increases the rate of natural recovery from infection and which develops substantially later in life ( late childhood to adolescence ) ., Adopting one of these processes in isolation does not reproduce observed patterns of age-prevalence of asexual parasitaemia , disease , and infectivity ( gametocytaemia ) across different endemicities ( as measured by EIR ) ., Moreover , while both clinical and parasite immunity were allowed to vary with age and EIR , the model in which natural recovery from infection ( e . g . , asymptomatic to subpatent ) is determined solely by age better matches observed patterns than a model in which this is also determined by the intensity of exposure ( EIR ) ., This suggests parasite immunity in non-naïve individuals may be controlled by physiological processes rather than by amount of exposure ( provided there is exposure ) ., These findings agree with the current view that parasite immunity may require ageing to develop , but subsequently can persist without high antibody titres and therefore be maintained by occasional infrequent boosting 7–9 ., Peaks in parasitaemia above 30 y of age present across endemic levels in eastern Tanzania might reflect malaria-HIV co-infection 31 and are not expected to be captured by the model ., Incorporating a prolonged duration of ( subpatent ) infections , i . e . , continual reinfection that prolongs infection and boosts an immune response that allows parasitaemia to persist at subpatent levels , worsened the model predictions ., However , we cannot exclude that an overall immune-modulated increase in duration of infections takes place , as suggested by recent hypotheses from within-host models 32 and in transmission models with fewer immunity components 17 , 18 ., This is because the increase in duration of subpatent infection with increasing EIR could be weaker than considered by us ., Furthermore , interpretation of this immunity function may depend on our model structure: we assume ( via immunity functions 1 and, 2 ) that a host returning to the noninfected state ( SH ) is likely to rapidly become asymptomatic with subpatent parasitaemia upon reexposure ( i . e . , is immune to symptomatic and to patent asymptomatic infection ) , tantamount to frequent subpatent infection but with recovery and reinfection modelled explicitly ., Other models 17 , 18 assume persistent asymptomatic infection ( though patency status may not be specified ) which may be regarded as an implicit way of modelling this reinfection cycle ., Our model additionally allowed us to explore what age-prevalence patterns can tell us about the duration of clinical and parasite immunity ., Our results suggest that clinical immunity has shorter memory ( with a half-life of the order 5 y or more ) , while parasite immunity is effectively everlasting ( with a half-life of 20 y or more after onset in adolescence ) ., These durations are in line with evidence that migrating adults returning to endemic areas tend to become more sensitive to clinical attack but have lower parasite levels than children 8; they are also in line with immunological studies in which one postulated mechanism of clinical immunity ( antibodies to parasite phospholipids ) has been shown to have a rather short half-life 33 , 34 ., There are limitations in the epidemiological data that are available to inform model parameters ., In particular , there are few and uncertain estimates of EIR by altitude range 28 , 35 , as mentioned earlier ., Furthermore , EIR estimates were not obtained from the same villages that were parasitologically surveyed , and the local history of interventions ( which might affect the EIR ) is not known ., Therefore , discrepancies between observed and estimated EIR values are to be expected , especially in low-transmission areas where mosquito sampling is more difficult ., The model presented here clearly makes a number of simplifying assumptions ., One of the main limitations is that the immunity functions , whilst generated based on current immunological understanding , could not be constrained by data ., Further data on the way in which immunity develops and on the factors driving its development could help to refine these functions ., The model also does not allow for partial immunity to reinfection , which would be relevant from the point of view of treating or vaccinating against pre-erythrocyte stages ., While sterilising or partial pre-erythrocyte immunity are likely to be rare 7 , it could be useful to extend the model to explore this possibility ., Thirdly , we have not explicitly modelled the effects of parasite genetic diversity and have thus , strictly speaking , treated infections as monoclonal ., However , the widely accepted hypothesis that immune development is regulated by antigenic variation and cumulative exposure to inoculations of differing parasite strains 20 , 22 , 26 , 32 , 36 is analogous to our definition of immunity levels in terms of cumulative exposure with finite memory ., Our model is therefore consistent with theories in which immunity is strain-specific whilst integrating other aspects of acquired immunity development supported by cross-sectional data and current immunological understanding ., This age-structured malaria transmission model shares many features with existing models 17–25 but is novel in the way it combines epidemiological and immunological processes ., Previous models have considered immune responses of types similar to those studied here ( especially immunity that acts on the duration of asymptomatic and subpatent infections ) 17 , 18 , 21 , 22 , 24 , whilst others have represented acquired immunity through increased ability to reduce blood-stage parasite density 18 , 22 , 23 , 25 ., Clearly , it is never possible to determine whether the structural assumptions behind any model represent the true processes generating the observed data , and it is likely that more complex model structures could also generate similar patterns ., One alternative method that could be employed is to track parasite density rather than infection alone ., Such an approach explicitly acknowledges variation in parasite load between individuals , and this variation may influence the development of immunity ., However , such an approach also has its limitations ., In particular , the distinction between disease and asymptomatic and subpatent infection requires definition of arbitrary parasite density thresholds for becoming diseased once infected and for detection by microscopy ., Our assumption that susceptibility and recovery vary continuously via dependence on cumulative exposure is , however , analogous to the effect of immunity in bringing parasite density below such thresholds ., A second alternative method for incorporating immunity into mathematical models is to explicitly model strains and hence incorporate long-lasting strain-specific immunity ., As noted above , our assumption that immunity develops with exposure and has finite memory essentially reproduces the patterns that would be obtained from such a model ., The model does not imply that parasite density or strain-specific immunity are unimportant; as indeed there is strong evidence to support both playing a role in the development of immunity ., Rather , our simpler model structure which implicitly incorporates these processes through immunity functions allows us to explore the timescales over which clinical and parasite immunity develop and are lost as well as the role of ageing and exposure on these functions ., Few previous models have been consistent in checking that they can reproduce the patterns of infection observed across a range of endemicities ., By validating output against such patterns , we have sought to develop a model that is both informative about the impact of immunity on falciparum malaria epidemiology and also forms a solid basis with which to explore the impact of interventions ., Having a robust framework which adequately captures the development of immunity with exposure and age is particularly important in exploring the impact of interventions such as insecticide treated nets ( ITNs ) and intermittent preventive therapy ( IPT ) in infants and children for which there is the potential to delay immunological development ., We model a human population with continuous age structure in which individuals of a given age can be in one of the following states: susceptible or not infected ( SH ) , latent infection ( EH ) , infected with symptomatic disease ( including severe and clinical cases ) ( DH ) , asymptomatic with detectable parasites ( AH ) , and asymptomatic infection with undetectable ( subpatent ) parasite density ( UH ) ., The main distinction between states DH and AH is that individuals in state AH do not prompt treatment that leads to a change in infection state ., The state UH is included to account for the fact that measured parasitemia often decays with age , while highly sensitive parasite detection techniques suggest parasitemia continues increasing with age nearing 100% in highly endemic areas 37 ., In tandem , we consider a mosquito population whose individuals can be susceptible ( SM ) , exposed ( latent ) ( EM ) , or infectious ( IM ) ., Figure 6 shows the transitions between states in each population ( without displaying ageing ) ., Susceptible humans move to latent infection at rate Λ , the force of infection on the human population ., Individuals remain in this state for a mean duration 1/h ( the mean latent period ) ., A proportion φ develop disease whilst the remainder ( 1−φ ) move to the asymptomatic infection category ., A proportion f of symptomatic cases ( DH ) receive effective drug treatment and recover at rate rT , while the remaining cases recover naturally without treatment at rate rD ., If clinical treatment or natural recovery is fully successful at removing parasites ( with probability φ ) , the host returns to the susceptible state and otherwise moves to the asymptomatic state ., Asymptomatic infections become subpatent at rate rA , and these subpatent infections are cleared at rate rU with individuals returning to the susceptible state ., Those in the asymptomatic state may additionally develop disease through superinfection at rate φΛ ., Each human infection state , namely DH , AH , and UH , has a specific level of infectivity ( transmission of mature gametocytes ) to biting mosquitoes ., The full equations for this model and further parameter definitions are given in Protocol S1 ., Table 1 summarises variables , parameters , and the values used to generate the model outcomes presented in Results ., Sensitivity analyses of model output to these parameters are presented in Protocol S1 ., In our analysis , we focus on results obtained once endemic levels are reached ., Model outputs are generated by fixing the EIR or by fixing mosquito density ( m ) and calculating the EIR via the equations describing the mosquito section of the parasites transmission cycle ( see Protocol S1 ) ., We ignore any possible dependence of infectivity in the different infection states ( DH , AH , UH ) on age and EIR because this is currently less-well-understood 38 ., For simplicity , we assume that the rate of natural recovery from clinical disease ( rD ) in the absence of treatment is identical to that from asymptomatic infection ( rA ) , and that the rate of recovery of treated cases ( rT ) is determined by treatment only ., Unknown parameters ( Table 1 ) were estimated by running the model over a wide range of plausible values and excluding values which lead to epidemiological patterns that clearly failed to visually match observed patterns ., Our aim was to identify model structures and parameters values based on their ability to reproduce patterns and relationships ., Given the many uncertainties in model structure , large number of parameters , and limited data available , it would have been very difficult to implement a more formal and rigorous statistical approach ., Rather , we have focused on qualitative comparison and understanding ., The sensitivity analyses to key parameters ( in Protocol S1 ) give an idea of uncertainty and ranges of parameter values that might be expected on the basis of this model and datasets ., To explore the impact that acquired immunity can have on patterns of age prevalence in endemic settings , we extend the basic transmission model above to incorporate immunity acting at three different stages of a hosts history of infection ., Mathematical details of the functions , described in brief below , are given in Protocol S1 ., 1 . Susceptibility to symptomatic disease , φ ( immunity function 1 ) ., We assume that individuals are born with maternally acquired immunity which is determined by the endemic level of disease and decays with a half-life dM ., Following birth , clinical immunity accumulates due to exposure at a rate dependent on the force of infection in the population , Λ ., This acquired immunity decays with a half-life dS ., The schematic for this model is shown in Figure 7A ., Susceptibility to symptomatic disease is then assumed to decrease in a nonlinear way as levels of clinical immunity increase ., The overall dependence of susceptibility φ on age and EIR resulting from this model is shown in Figure 7B ., 2 . Rate of natural recovery from asymptomatic to undetectable infection , rA ( immunity function 2 ) ., The parasite immunity level associated with this response is similarly assumed to accumulate at a rate dependent on the force of infection in the population , Λ ., The onset of parasite immunity is further assumed to have an age-related delay with mean dl , and any maternal immunity is lost during this period ., Parasite immunity then decays with half-life dA ., The schematic for this model is shown in Figure 7C ., The recovery rate rA is assumed to increase with levels of parasite immunity through a nonlinear function which saturates at higher levels of immunity ., The overall dependence of recovery on age and EIR resulting from this model is shown in Figure 7D , where change with age follows from age-dependent exposure ( see Protocol S1 ) ., As an alternative , we also consider a model in which parasite immunity is determined only by age ( given some exposure to infection ) and not by EIR ., 3 . Rate of natural clearance of undetectable infection , rU ( immunity function 3 ) ., We assume that the duration of undetectable infection is boosted by continual reexposure and therefore not directly dependent on age ., The onset of immunity is therefore dependent on the force of infection , Λ , and decays with half-life dU as in previous models of superinfection 17 , 18 ., A schematic for this is shown in Figure 7E and the resulting recovery rate as a function of the force of infection in Figure 7F .

Introduction, Results, Discussion, Materials and Methods

Acquisition of partially protective immunity is a dominant feature of the epidemiology of malaria among exposed individuals ., The processes that determine the acquisition of immunity to clinical disease and to asymptomatic carriage of malaria parasites are poorly understood , in part because of a lack of validated immunological markers of protection ., Using mathematical models , we seek to better understand the processes that determine observed epidemiological patterns ., We have developed an age-structured mathematical model of malaria transmission in which acquired immunity can act in three ways ( “immunity functions” ) : reducing the probability of clinical disease , speeding the clearance of parasites , and increasing tolerance to subpatent infections ., Each immunity function was allowed to vary in efficacy depending on both age and malaria transmission intensity ., The results were compared to age patterns of parasite prevalence and clinical disease in endemic settings in northeastern Tanzania and The Gambia ., Two types of immune function were required to reproduce the epidemiological age-prevalence curves seen in the empirical data; a form of clinical immunity that reduces susceptibility to clinical disease and develops with age and exposure ( with half-life of the order of five years or more ) and a form of anti-parasite immunity which results in more rapid clearance of parasitaemia , is acquired later in life and is longer lasting ( half-life of >20 y ) ., The development of anti-parasite immunity better reproduced observed epidemiological patterns if it was dominated by age-dependent physiological processes rather than by the magnitude of exposure ( provided some exposure occurs ) ., Tolerance to subpatent infections was not required to explain the empirical data ., The model comprising immunity to clinical disease which develops early in life and is exposure-dependent , and anti-parasite immunity which develops later in life and is not dependent on the magnitude of exposure , appears to best reproduce the pattern of parasite prevalence and clinical disease by age in different malaria transmission settings ., Understanding the effector mechanisms underlying these two immune functions will assist in the design of transmission-reducing interventions against malaria .

Whilst it is clear that natural immunity to malaria infection develops in those living in malaria-endemic regions of the world , the precise way in which it is acquired and the duration of immune memory are less-well-understood ., We used a mathematical model that mimics malaria transmission between humans and mosquitoes in endemic settings to explore what epidemiological data , and in particular the prevalence of malaria in different aged individuals , can tell us about how immunity might develop ., We explored three different parts of the transmission cycle at which immunity could act:, 1 ) reducing the likelihood that an infected person develops symptomatic disease;, 2 ) increasing the rate at which infection is cleared , and, 3 ) increasing the duration of low-level ( subpatent ) infections that would continue to boost the immune system and hence protect against further disease ., Our results show that the first two mechanisms together give rise to patterns of malaria by age group that are consistent with those observed in different malaria endemic settings in Africa ., Our model also suggests that immunity to symptomatic disease lasts for at least five years , develops faster if there are higher levels of infection in the population , and increases with age ., On the other hand , our model suggests that immunity that helps to clear infection lasts longer ( 20 years or more ) , develops later in life , and does not depend on the amount of transmission in the population .

mathematical models, public health and epidemiology, malaria, epidemiology

journal.pgen.1000641

The Evolutionary Origin of Man Can Be Traced in the Layers of Defunct Ancestral Alpha Satellites Flanking the Active Centromeres of Human Chromosomes

Active human centromeres are made of great ape-specific alpha satellite DNA ( AS ) , comprised of ∼171 bp tandem monomers forming nearly identical higher order repeats ( HORs ) and represented by the “new” suprachromosomal families ( SFs ) 1 , 2 and 3 ., They are surrounded by much less homogeneous HOR-free “monomeric” AS ( SF4 and SF5 ) often disrupted by transposon insertions 1 , 2 ., SF4 is usually composed of a single M1 class of monomers with no evidence of higher-order periodicities ., SF5 is formed by two types of monomers , R1 and R2 , alternating irregularly ., R2 is similar to M1 ( class A ) , and R1 represents the first appearance of novel class B monomers , which bind CENP-B protein and presumably have invaded the A-arrays before the great ape divergence 1 ., High identity and high copy number of HORs are presumably maintained by an active process called homogenization , which is driven by homologous recombination mechanisms such as unequal crossover and/or gene conversion ., The monomeric AS is older than the HOR arrays and resembles AS of lower primates 1 ., Divergence patterns and transposon distribution suggest that the “old” domains were once homogenous , but at some point homogenization had stopped and accumulation of sequence divergence and of transposable elements commenced 2 ., Thus , old AS arrays are likely the remnants of the centromeres of our primate phylogenetic ancestors , once active and homogenous , but obsolete and degrading since centromeric function and homogenization have shifted to the new AS 1 , 2 ., Furthermore , analysis of the human X chromosome short arm ( Xp ) pericentromeric region , the first one sequenced in its entirety , has revealed an age gradient , with most distal Xp AS domain dating to early primate evolution , the HOR domain to the time of great ape divergence and the domains in between being of interim age 3 , 4 ., Assuming that the succession of AS layers on the long arm ( Xq ) side is symmetrical , it was proposed that the primate X chromosome centromere “evolved through repeated expansion events involving the central functional AS domain , such that ancestral centromeric sequences were split and displaced distally onto each arm” 4 ., Previously , we proposed the existence of a kinetochore-associated recombination machine ( KARM ) that homogenizes only the active centromere , a model that accounts well for the above observations 1 , 2 ., Accumulating evidence suggests that topoisomerase II , a DNA decatenating enzyme , is an important part of this machine ., In mitosis , it resides in the kinetochore 5–7 and plays a crucial role in resolution of the recently discovered chromatin PICH threads that connect chromatid centromeres 8–12 ., The enzyme introduces double strand breaks into human AS arrays 13–15 , and in dicentric chromosomes its activity is observed only in the active centromere 16 ., As topoisomerase II breaks are known to initiate homologous recombination 17–19 , the enzyme is a likely candidate for KARM function ., Here we present a complete analysis of AS layers of chromosomes 8 , 17 and X and for the first time provide comprehensive comparisons of the entire layer patterns on both arms of one chromosome and between different chromosomes ., As expected , the succession of multiple layers appears to be largely symmetrical around the centromere ., More surprisingly , the layer structure is to a large extent shared between non-homologous chromosomes , supporting a model of genome-wide expansion events that give rise to new centromeres on many chromosomes within an evolutionary short period of time ., Primate comparisons reveal that each major taxon in the human lineage corresponds to a separate “suprachromosomal” centromeric layer , providing a complete record of human ancestry ., Comparisons of inter- and intra-species divergence within a layer suggest that , after relocation of the centromere , the dead arrays experienced an unusual burst of mutability ., Highly informative structure and their potential role in “centromeric speciation” 20 , 21 should make centromeric layers extremely useful for phylogenetic analysis ., In human chromosomes 8 , 17 and X the pericentromeric regions of both chromosome arms have been sequenced almost completely , starting from the surrounding euchromatic regions and into the HOR arrays that constitute current centromeres ., We used the genomic builds of these chromosomes ( see Table S1 for reference sequences ) to identify and extract all AS monomers and analyzed them using a cladistic approach 3 , 4 , 22–24 based on construction of monomeric phylogenetic trees ( Figure 1; see Text S1 for details ) ., This resulted in identification of a number of distinct AS domains in each centromere ( listed in Table 1 and shown in different colors in Figure 2 ) ., The main criteria to assign differently located arrays to the same-color suprachromosomal layer was their structural similarity and ability to “mix well” on phylogenetic trees with each other , but not with the other layers ., Our results do not contradict previous partial analysis of AS on these chromosomes 3 , 4 , 24 and throughout the genome 1 , 2 ., However , a few important new features were noted ( Table 1 and Text S1 ) and the entire complex pattern of AS relationships was revealed for the first time ., Figure 2 shows that same-color AS layers are shared by both arms of one chromosome , as well as by three different chromosomes ., However , two solitary domains , grey ( H4 ) and olive-green ( H1H2 ) , were observed ., To find out if the counterparts of solitary domains were present elsewhere in the genome , we scanned the databases and found the arrays of sequences that mix well ( Table S2 and Text S1 ) on chromosomes 1 , 3 , 4 , 5 and 18 ( grey ) and 5 and 7 ( olive-green ) ., The yellow and blue layers corresponded to previously characterized SF4 ( M1 ) and SF5 ( R1R2 ) , respectively 25 , 26 ( see Table 1 and Text S1 ) ., The genome-wide distribution of these families was documented previously 1 , with additional examples provided in Table S6 ., The fact that same-color arrays from different chromosomes mix on phylogenetic trees ( Figure 1C ) confirms that , contrary to the new AS , the old AS had no chromosomal specificity and was homogenized genome-wide 1 , 22 within a “suprachromosomal” layer ., The layers identified in Figure 2 showed no significant mixing with each other on phylogenetic trees ( Figure 1 ) ., However , within some layers two or more closely related sub-domains could be discriminated ( Figures S1 and S2 ) ., These sub-layers mixed with each other to some extent ( Figure S3 ) and thus could not be formally identified as individual layers within the framework of this study ., The nature and significance of this finer structure deserve further investigation ( see Text S1 for details and discussion ) ., The layer pattern depicted in Figure 2 shows the structures partially symmetrical around the current centromere with most of the layers being shared between chromosomes ., Same-color domains on different sides of the centromere mix well on phylogenetic trees ( Figure 1B ) ., This confirms that in many cases a new centromere arises in the midst of the old one , by amplification of a new AS variant , and moves the remnants of the old centromere sideways ., However , emergence of evolutionary new centromeres 27 , 28 and chromosomal rearrangements may predictably cause partial asymmetry ., The same layers on different chromosomes combined with structural discontinuity in the succession of layers ( e . g . monomeric – dimeric – monomeric ) prove that the sequences seeding new centromeres were not picked up independently on each chromosome , but rather have spread by rounds of interchromosomal exchange and subsequent amplification , as was shown previously for the new AS families 1 , 29 , 30 ., An alternative scenario of genome-wide homogenization with occasional exclusion of some domains or their parts is detailed in the “Discussion” section ., Such a scenario would not support structural discontinuity and frequent symmetry and , therefore , cannot be solely responsible for the patterns described ., However , it could explain the sub-layers noted above ., Interpretation of the layer structure proposed above allows a number of predictions ., Hence , we moved on to verify it by phylogenetic , transposon distribution and divergence pattern analyses ., One would expect the extant primate taxa to share a certain number of ancestral layers with humans in a collinear succession up to the layer corresponding to the last common ancestor of humans and this particular primate ., Ancestral layers may be followed by some primate-specific ones , corresponding to evolution of a primate branch after divergence from the human lineage ., As this prediction is a particularly valuable tool in the hands of molecular anthropologists , we tested it by demonstrating the human lineage layers identified in this work in the genomes of various primate species ., Samples of each layer were used to identify the most closely related sequences in the incomplete builds and collections of shotgun primate sequences in the GenBank ., The AS monomers were extracted and aligned and phylogenetic trees constructed independently of human sequences ., Consensus monomers of the branches were matched to consensus monomers of human layers and “mixing tests” were occasionally performed to verify the identity of the layers ., As expected , each primate genome contained counterparts to some human lineage layers as well as the sequences specific for a particular primate branch ., The results are summarized below and in Table 2 , reference sequences are listed in Table S6 ., Notably , we failed to find any AS among abundant genomic sequences available for lemurs and tarsiers ( not shown ) ., Only grey layer ( H4 ) sequences were found in the genome of a New World monkey ( NWM ) Callithrix jacchus , along with NWM-specific AS including S3S4 satellite 1 ., The Old World monkeys ( OWM ) had grey ( H4 ) , red ( H3 ) and olive-green ( H1H2 ) layers plus a number of OWM-specific satellites including the one based on S1 and S2 types 1 ., Therefore , we concluded that the grey layer belonged to an ancestor that pre-dated the NWM/OWM separation while the red and olive-green layers originated later , but were already present in a common ancestor of humans and OWM ., Additionally , the yellow-striped ( V1 ) and yellow ( M1 ) layers , but no newer types , were found in gibbons ., This indicated the existence of at least 2 entirely extinct taxa in the human lineage , the one with “red” ( H3 ) centromeres between NWM and OWM , and the one with “yellow-striped” ( V1 ) centromeres between OWM and gibbon ancestors ., Recent analysis of HOR-like AS in gibbons 22 and our preliminary data ( see Table S6 and Text S1 ) indicate the abundance of yellow-derived ( M1 ) gibbon-specific AS sequences ., In orangutans , the blue layer ( R1R2 ) was present together with a number of the blue-derived orangutan-specific families composed of R1R2 dimers and longer HOR-like R1/R2 repeats ( Table S6 and Text S1 ) ., Contrary to expectations based on the old hybridization data 1 , after an extensive search ( over 11 . 4 Gb of WGS sequences screened ) , we failed to find new SFs in orangutans ., Thus , the new AS is , in fact , specific to African apes , not great apes , as it was supposed previously ., As expected , in gorilla and chimpanzee genomes , all the above layers plus the three new SFs 1 , 2 and 3 were present ( not shown , see Table S6 for reference sequences ) ., As described above , certain types of AS sequences that were absent in some primate WGS reads were readily detectable in WGS collections of other primates ., However , the conclusions based on the absence of findings should be treated with some degree of caution , as it is possible that the WGS reads were not comprehensive ., To get another estimate of the age of AS layers identified in this work , we typed L1 retroposons integrated therein , as described previously 2–4 ., The age of the oldest L1 elements found in an AS layer would indicate the time when it stopped homogenization and became available for insertions 2 ., Table 2 ( see also Table S3 and Text S1 for details ) shows that the oldest L1 elements were identified as follows: PA3 in the blue layer; PA3 and just one PA4 in the yellow; PA4 in the yellow-striped; mostly PA4 and just two copies of PA5 in the olive-green layer ., PA5 was the oldest L1 repeat in the red layer and PA7 in the grey ( numerical number of L1 family increases with age 31 ) ., In order to relate these results to living primates phylogeny we scored the L1 elements in the genomes of various primates looking for the elements active at the time of divergence of respective taxon with human lineage ., In each genome the youngest major L1 repeat shared with humans was identified as follows: PB3 and PA15 ( were active simultaneously 31 ) for lemurs; PA8 for tarsiers; PA6 for NWM and PA5 for OWM ., Gibbons had just a few PA3 and abundant PA4 , orangutans had abundant PA3 , gorillas and chimpanzees had abundant PA2 ( Table 2 and Table S4 ) ., Superimposing the above two sets of data , it can be concluded that the blue layer was already available for insertion shortly after orangutan divergence ( PA3 was still active ) ., The yellow layer , which was exposed to only a residual PA4 activity , if any , and a lot of PA3 activity , started to accumulate L1s between gibbon and orangutan divergence from the human lineage ., The yellow-striped layer got its oldest L1s way after OWM divergence ( PA4 is abundant ) and the olive-green layer right before or right after that , as it still had PA5 activity ., The red layer belonged to a more distant OWM - human ancestor ( PA5 abundant ) and the grey one to a common ancestor of OWM and NWM ( PA7 present ) ., The age of the layers may be roughly estimated as follows: new AS 7 myr , blue ( R1R2 ) 14–16 myr , yellow ( M1 ) 16–18 myr , yellow-striped ( V1 ) 18–23 myr , olive-green ( H1H2 ) 23–26 myr , red ( H3 ) 26–40 myr , grey ( H4 ) 40–58 myr ., As a temporary classification ( see Table 1 ) , we propose:, ( i ) To term the AS forming centromeres of monkeys in human lineage “ancient AS” ( types H1–H4; without suprachromosomal family names ) ,, ( ii ) to keep the term “old AS” only for lower ape layers , namely V1 ( yellow-striped; SF6 ) , M1 ( yellow; SF4 ) and R1R2 ( blue; SF5 ) and , ( iv ) to apply the term “new AS” to African ape-specific SFs 1 , 2 , and 3 1 ( see Text S1 for details ) ., It is expected that the closer the layer is to a current centromere the younger it is and the less is the divergence between monomers ( or dimers ) within the array ., Table 3 shows that in all cases the pattern of divergence does not contradict this prediction ., Divergence figures for same-color domains on one chromosome and on different chromosomes are in remarkable concordance ., We next tested if the divergence in each layer would match its proposed age ., Orthologous comparisons of the grey ( H4 ) domains from a number of primate species 4 , 24 ( Table S5 ) yield a “normal” mutation rate of about 0 . 2% per million years ( 0 . 17%–0 . 21%; see Text S1 ) , similar to surrounding euchromatin 4 , 24 and L1 repeats 31 ., However , the age of the layers calculated from intra-array divergence figures , using this rate , clearly contradicts our primate and L1 dating ( Table 4 ) ., This “calculated age” is about twice as old and out of line with all accepted taxon age estimates 32 ., Conversely , mutation rate calculated using intra-array divergence and the age of layers estimated from primate and L1 dating , ranges 0 . 4%–0 . 6% and is 2 to 3 times higher than normal ., To explain this discrepancy we propose that a period of hypermutability occurs in homogeneous AS array after the centromere moves away and homogenization stops ., The high mutation rate may somehow be caused or conditioned by near perfect identity of centromeric repeats and gradually subsides upon accumulation of mutations and repeat divergence ., The normal rate applies only to the “long dead” arrays ., The high rate applies to the “hypermutability” stage ( “freshly abandoned” centromeres ) and possibly to the “homogeneous” stage of array evolution ( active centromeres ) ., In the latter case , it may drive the rapid concerted evolution of homogenous HOR domains 24 ., Hypermutability may be caused by low fidelity DNA synthesis , if secondary structures in AT-rich satellite or high density of replication origins 33 cause fork stalling and trigger repair mechanisms using error-prone DNA polymerases 34 ( see Text S1 for more details ) ., The record of human evolutionary lineage revealed in the centromeres appears to be relatively consistent and easy to interpret ., There is remarkable concordance between primate studies , L1 dating and divergence grading ., However , our conclusions have to be verified further upon availability of more comprehensive primate sequencing data ., Notably , living primates faithfully represent the succession of major taxa in the human lineage ., The only two entirely extinct taxa are represented by red ( H3 ) and yellow-striped ( V1 ) layers ., The controversial fossil record 32 , 35 offers at least three candidate extinct families: Propliopithecidae ( Catopithecus and Aegyptopithecus; 33–35 myr ) , Pliopithecidae ( Proconsul; 17–27 myr ) and Dryopithecidae ( Dryopithecus; 9–14 myr ) ., The first two match the red ( 26–40 myr ) and yellow-striped ( 18–23 myr ) layers pretty well and the third is not supported by a separate AS layer , which may mean it is a sister clade of either African apes or Pongidae ( orangutan family ) ., The exact positions of extinct taxa in human lineage are hotly debated 35 and comprehensive analysis of AS layers in extant primates and man would help to resolve this problem ., The possibility to number and locate the extinct ancestral taxa on the evolutionary tree and to distinguish the ancestor from the descendant even in two-species comparisons is unique to the AS record ., As there is every reason to believe that centromeric layers are not limited to AS and primates , the method has vast potential for phylogenetic studies ., Figure 2 displays imperfect symmetry of AS layers around the current centromere ., One can potentially explain it in two different ways ., The process creating layers can be asymmetrical in nature , and the elements of symmetry may appear randomly as a matter of coincidence ., Alternatively , the process may be intrinsically symmetrical , but the symmetry is imperfect for a number of random historical reasons like formation of evolutionary new centromeres , chromosomal rearrangements , etc ., A possible scenario for the asymmetrical process would be continuous genome-wide homogenization and concerted evolution of active centromeres , identical on all chromosomes ., From time to time for various reasons ( inversion , insertion of a long mobile element , amplification of another tandem repeat , etc . ) parts of individual arrays get cut off the bulk of the array and hence are excluded from homogenization ( a “segment cut off scenario” ) ., As KARM presumably brings efficient homogenization only to the current centromere , and only one centromere per chromosome can be maintained , the cut off part looses both homogenization and centromeric function and becomes a dead segment ., An epigenetic mark 20 is likely to be involved in the stable choosing of only one of the segments as a centromere , in this scenario ., If two segments on both arms of one chromosome die at about the same time , they would form a symmetrical structure ., Dead segments in different chromosomes , excluded from homogenization at the same time , would form a “same-color” suprachromosomal layer ., However , such coincidences and hence the elements of symmetry should be rare ., Also , structurally discontinuous patterns like a succession of monomeric – dimeric – monomeric layers are hardly possible ., Therefore , this process cannot be solely responsible for centromeric evolution ., A mechanism for the symmetrical process was described as the only one possible for the new chromosome-specific SFs 1 , 29 , 30 , which are represented by structurally different HORs on each chromosome ., It includes a series of interchromosomal transfers and amplification events , facilitated , as we propose , by KARM , which is also responsible for homogenization ( “interchromosomal transfer/amplification scenario” ) ., In most cases , new variants come from another location , insert into the active centromere , split and inactivate it by luring the kinetochore to the new array and move the remains sideways as a result of self-expansion ., Potentially , this process could be solely responsible for the layer pattern revealed in chromosomes 8 , 17 and X . However , depending on the extent of symmetry and interchromosomal similarity of the layer patterns in the rest of the genome , some combination of the two scenarios may appear to be parsimonious ., Notably , in this work we studied only the centromeres with SF2 ( chromosome 8 ) and SF3 ( chromosomes 17 and X ) HOR domains ., However , our unpublished preliminary results show that SF1 centromeres are flanked by the same types of old and ancient AS sequences ( see chromosome 7 sequences in Tables S3 and S6 ) ., Each new expansion of an AS variant covered many chromosomes and occurred in a relatively short time , as the order of layers is more or less conserved between chromosomes ., Obviously , an expanding variant had to possess some sequence novelty , which attracted the kinetochore 1 , 36 ., For instance , a new or better fitting protein-binding site might make a satellite repeat a more attractive centromere , as may be exemplified by evolutionarily recent recruitment of CENP-B protein to primate centromeres 26 , 37 , 38 ., Initially new sequence variants may arise in poorly homogenized areas such as dead segments , borders of current centromeric arrays , etc ., A successful variant has to accidentally insert into a current centromere , win over the kinetochore and self-expand ., Apparently , once such “better” centromeric sequence appeares on one chromosome , it has a good chance to invade other chromosomes very quickly ., Together with the tendency of a new variant to integrate/amplify in the current centromere , not in the old layers , it suggests that KARM may take part in interchromosomal transfer and/or integration and amplification in new locations ., When a neocentromere is formed on unique DNA , KARM may be used to seed and amplify centromeric repeats at the new site ., It is also likely that only a sequence integrated in the current centromere may easily acquire the centromeric epigenetic mark 20 ., Additional details of specific scenarios are provided in Text S1 ., The centromere is remarkable for its plasticity ., Centromeric DNA and proteins are subject to phylogenetic variation very much unlike other components of chromatin and cell division machinery 21 ., Here we show that constant generation of new AS variants and perhaps their competition for centromeric function resulted in serial waves of AS expansion in the course of primate evolution ., Each wave led to emergence of new underlying sequences in active centromeres of many chromosomes ., It was demonstrated previously that in the genomes of monkeys , A-type AS , as a rule , is the same in all chromosomes and hence is homogenized throughout the whole genome 1 , 22 ., On the contrary , the new AB-type AS which is present in the genomes of African apes is chromosome-specific and , as a rule , is effectively homogenized only within one chromosome ., According to our model , an AS layer unites the arrays which, ( i ) have a common origin ,, ( ii ) were active centromeres at the same time , and, ( iii ) at that time were homogenized throughout the genome as a single entity ., Points i and ii are also valid for the new SFs , but point 3 is not , otherwise SFs and AS layers are the same ., This difference reflects a shift from genome-wide to chromosome-specific homogenization ., Centromeric function per se can be performed by unique sequences as exemplified by neocentromeres 27 , 39 and the centromeres of budding yeast 40 ., However , natural centromeres of all higher organisms are made of highly repeated sequences , hinting perhaps at some additional function ., Two such functions , not mutually exclusive and perhaps even interrelated , have been discussed ., Cohesion of centromeres , tension sensing and signalling to the spindle assembly checkpoint , may be provided by the formation and resolution of PICH threads 8–12 ., If PICH threads are formed by recombination intermediates , they may just show the proposed kinetochore-associated recombination machine ( KARM ) at work ., Vast satellite arrays provide phenotypically silent DNA to form the threads , which , therefore , are indifferent to breakage , occasional erroneous repair , etc ., On the other hand , the features of primate highly repeated centromeres , such as, ( i ) tandem structure prone to recombination ,, ( ii ) putative possession of its own recombination machine ,, ( iii ) presence of a divergent dead zone that provides a good source of new sequence variants , and finally, ( iv ) alleged propensity to go through hypermutability periods , seem to constitute a special “plasticity” adaptation evolved to ensure that from time to time a new centromere would arise in a stochastic manner ., The concept of centromeric speciation 21 speculates on possible evolutionary benefit of such an adaptation ., It suggests that centromere plasticity may play a role in generation of new species by providing partial reproductive isolation of a karyological variant with a new centromeric layer ., Establishment and separation of incipient species may proceed via mechanisms described for other types of chromosomal speciation 41 ., The NCBI website ( www . ncbi . nlm . nih . gov/ ) was used to extract human centromeric regions and the BLAST server ( www . ncbi . nlm . nih . gov/blast/Blast . cgi ) to search for AS-containing human and primate clones and WGS reads ., AS monomers were identified by PERCON similarity search 2 , extracted and classed into monomeric types and SFs by a Bayesian classifier 2 and aligned by CLUSTALW 42 ., After manual inspection , a small number of evidently abnormal monomers were discarded ., Phylogenetic trees were constructed using the PHYLIP 3 . 65 package ( http://evolution . genetics . washington . edu/phylip . html ) ., DNA distance matrix was calculated using the F84 method and trees were constructed by UPGMA and neighbor-joining methods ., Similar neighbor-joining trees were obtained with the MEGA4 package ( www . megasoftware . net ) using the same monomer sets , and most major branches were verified by an interior branch test ( >90% ) in 500 replicates as described 4 ., To identify a group of monomers as distinct AS domain we used four main steps:, ( i ) The “branching test” shows that monomers are on the same branch of the monomer phylogenetic tree and hence are closely related and have presumably arisen by amplification of a single ancestral sequence;, ( ii ) the “compact residence test” verifies that all these closely related monomers are concentrated in separate array ( s ) without significant interspersion with monomers from other branches;, ( iii ) the “structural test” evaluates the divergence in the group , reconstructs the ancestral sequence by derivation of an appropriate group consensus and places it in AS classification by establishing the relationships with other AS monomeric types ., ( iv ) Finally the “mixing test” shows that the monomers from differently located domains mix on one branch of a monomer tree and hence have arisen from a single ancestor and were once homogenized as one entity ( i . e . belong to one AS “layer” ) ., Layer-specific AS sequences , in human and primate genomes , were searched by using samples of layers as queries in a BLAST search of human or primate NCBI databases , including WGS assemblies and trace collections of shot-gun sequences ., WGS reads containing AS were used as such without any experimental verification of their centromeric location ., Non-AS repeats were identified by RepeatMasker ( www . RepeatMasker . org ) ., The same program was used for L1 classification ., Mutation rates were calculated using Jukes and Cantor formula 43 ( see Text S1 ) .

Introduction, Results, Discussion, Materials and Methods

Alpha satellite domains that currently function as centromeres of human chromosomes are flanked by layers of older alpha satellite , thought to contain dead centromeres of primate progenitors , which lost their function and the ability to homogenize satellite repeats , upon appearance of a new centromere ., Using cladistic analysis of alpha satellite monomers , we elucidated complete layer patterns on chromosomes 8 , 17 , and X and related them to each other and to primate alpha satellites ., We show that discrete and chronologically ordered alpha satellite layers are partially symmetrical around an active centromere and their succession is partially shared in non-homologous chromosomes ., The layer structure forms a visual representation of the human evolutionary lineage with layers corresponding to ancestors of living primates and to entirely fossil taxa ., Surprisingly , phylogenetic comparisons suggest that alpha satellite arrays went through periods of unusual hypermutability after they became “dead” centromeres ., The layer structure supports a model of centromere evolution where new variants of a satellite repeat expanded periodically in the genome by rounds of inter-chromosomal transfer/amplification ., Each wave of expansion covered all or many chromosomes and corresponded to a new primate taxon ., Complete elucidation of the alpha satellite phylogenetic record would give a unique opportunity to number and locate the positions of major extinct taxa in relation to human ancestors shared with extant primates ., If applicable to other satellites in non-primate taxa , analysis of centromeric layers could become an invaluable tool for phylogenetic studies .

The primate centromere evolves by amplification of alpha satellite sequences in its inner core , which expands and moves the peripheral sequences sideways , forming layers of different age in the “pericentromeric” area ., The expanding centromere model poses two main questions: ( 1 ) whether the succession of layers is symmetrical on both sides of the centromere , and ( 2 ) whether different chromosomes share the same layers ., We have analyzed and dated the layers on both sides of human chromosomes 8 , 17 , and X and shown that they were largely symmetrical on one chromosome and largely shared and arranged similarly in non-homologous chromosomes ., The layer pattern revealed that genome-wide waves of expansion of new satellite variants have occurred repeatedly in the human evolutionary lineage ., The layers which are likely to be the relic centromeres of our common ancestors with primate taxa follow each other in chronological order ., The two layers that do not match any living primate indicate the two completely extinct ancestral taxa aged 26–40 and 18–23 million years ., These could be Propliopithecidae ( Cathopitecus and Egyptopithecus ) and Pliopithecidae ( Proconsul ) , aged 33–35 and 17–27 million years , respectively ., The possibility to reveal and date extinct ancestors makes the analysis of satellite layers a unique tool for the reconstruction of primate phylogeny .

evolutionary biology/human evolution, molecular biology/molecular evolution, molecular biology/centromeres, molecular biology/chromosome structure, genetics and genomics/chromosome biology, evolutionary biology/genomics, molecular biology/bioinformatics

journal.pcbi.1005441

Peptide probes derived from pertuzumab by molecular dynamics modeling for HER2 positive tumor imaging

Human epidermal growth factor receptor 2 ( HER2 ) is a prominent target in breast cancer diagnosis and treatment , as approximately 20–30% of patients with breast cancer overexpressing the HER2 receptor 1 , 2 , a 185-kD transmembrane glycoprotein with 1 , 255 amino acids 3 ., The HER2 gene is a proto-oncogene that maps to chromosome 17q21 ., HER2 contains four domains ( I , II , III , and IV ) that comprise a ligand-binding extracellular portion , a single transmembrane helix , a tyrosine kinase domain closely related to the Janus kinases , and a C-terminal tail with a number of tyrosine phosphorylation sites that serve as a scaffold for adaptor molecules and enzymes in facilitating downstream signaling 4 ., The heterodimerization of HER2 with any of the other three HER family receptors results in autophosphorylation of the terminal carboxyl segment and initiates a variety of signaling pathways that regulate cell growth , proliferation , and metastasis 5–7 ., Currently , a number of therapeutic approaches have been developed to antagonize the effects of HER2 overexpression; these approaches include the humanized monoclonal antibodies trastuzumab and pertuzumab 8 ., Trastuzumab demonstrates clinical benefits in the treatment of HER2-positive breast cancer , in both early and metastatic stages ., One year of trastuzumab therapy is recommended for all patients with HER2-positive breast cancer who are also receiving chemotherapy 9 ., However , as trastuzumab becomes a routine therapy , resistance can develop following an initial robust response; a lack of response to initiation has also been observed among patients 10 , 11 ., The other antibody drug , pertuzumab , has received US Food and Drug Administration approval for the treatment of HER2-positive metastatic breast cancer ., Trastuzumab and pertuzumab bind to different epitopes in the extracellular domain of HER2 , and their mechanisms of action differ ., Pertuzumab binds the pocket of domain II , inhibits HER2 dimerization with other receptors , and leads to slowed tumor growth ., Trastuzumab , on the other hand , binds to subdomain IV 12 , and works by inhibiting the PI3K/Akt , Mirk , and hKIS pathways and promoting proteolytic cleavage of the extracellular domain 13 ., However , both drugs have been shown to stimulate the antibody-dependent cellular cytotoxicity mechanism 14 ., It has been commonly recognized that , compared to antibody drugs , small peptides are cost-effective , have good tissue and membrane permeability , high target specificity , and low toxicity ., Moreover , specific modifications to targeting peptides can be employed to provide diverse biosensing functions; this strategy has been leveraged to develop a method by which to detect metastatic tumor cells in primary tumors 15 ., The science of molecular dynamics ( MD ) has been widely applied to chemical physics , materials science , and the modeling of biomolecules—such as interactions between ligands and receptors 16 , 17—by simulating the physical movement of atoms and molecules based on a family of molecular mechanics force fields ., The MM/GBSA method is often used to estimate the free energy of solute–solvent interactions ., In this method , a generalized Born ( GB ) model is used to approximate the Poisson–Boltzmann equation , based on modeling a molecule as a set of spheres ., The accessible surface area ( SA ) approximates the experimental value of the averaged behavior of many highly dynamic solvent molecules between the transfer free energy and the surface area of a solute molecule 18 ., The one-bead-one-compound ( OBOC ) 19–21 library method can be used to systematically synthesize and screen the peptide library of a target protein ., It is a simple means of rapidly identifying small molecules that bind with high affinity to receptor molecules ., The strategy has been modified by many researchers to overcome the several limitations inherent in the original approach 22 ., Our research group has previously advanced it to a lab-on-chip system that embraces the whole peptide screening process—from single bead trapping to the final sequencing of peptides—by using MALDI–TOF–MS 23 , 24 ., In this study , we use a combination protocol comprising MD , MM/GBSA binding free energy calculation 25–27 to derive peptides that interact with HER2 protein based on the HER2/pertuzumab crystal structure ( PDB entry: 1S78 ) ., In silico mutations were performed to screen for peptides with the lowest binding free energy , and OBOC peptide library screening was then carried out ., Both the binding free energy calculation and the OBOC library screening found the peptide 58F63Y to have the highest affinity to HER2 ., 58F63Y , together with five other peptides , were selected for further analysis and experimental validation ., All results show that the peptide 58F63Y binds most favorably to HER2 , with a dissociation constant ( KD ) of 536 nmol/L ., The results of ex vivo and in vivo experiments using mouse xenografted tumors confirm that this peptide has strong affinity and high specificity to HER2 ., Binding free energy decomposition analysis 28–30 and distances calculation using Pymol found that there are more paired residues with low binding free energy and distances of less than 5 Å , which may explain the high affinity ., Compared to other peptides that target HER2 31 , peptide 58F63Y is unique in that it is acquired based on simulation using a different primary model with binding sites on domain II of the HER2 protein ., Given its low toxicity , this peptide may be used as an alternative probe in the diagnosis and treatment of HER2-positive breast cancer and contributes to the HER2-targeting peptide library ., Pertuzumab is a monoclonal antibody marketed by Genentech for the treatment of HER2-positive breast cancer ., Pertuzumab binds to HER2 at the center of domain II , sterically blocking the pocket essential to receptor dimerization and signaling ., The HER2/pertuzumab crystal structure was obtained from the Protein Data Bank ., In this structure , the soluble extracellular domain of HER2 32 was crystallized in complex with the Fab fragment of the disulfide anti-HER2 monoclonal antibody pertuzumab 33 ., Based on the calculation of distances for all residues between HER2 and pertuzumab and the selection of those within 5 Å , we found a peptide fragment of 20 residues in length ( sequence: EWVADVNPNSGGSIYNQRFK ) with a beta folding layer structure , named 4665 , that plays an important role in the interactions ( Fig 1A ) ., The HER2/fragment 4665 complex was chosen for further simulation analysis ., MM/GBSA free energy was calculated based on 500 snapshots from 7 to 10 ns of MD simulation trajectories ( Fig 1B ) for each complex , as described in the Materials and methods section ., The results show that the predicted binding free energy between 4665 and HER2 is –48 . 53 kcal/mol with the van der Waals ( ΔEvdw ) contribution being a main component ., To improve the affinity of 4665 against HER2 , we undertook single and double mutations and performed MD simulations to estimate the binding affinity ., The properties of the interacting amino acids in both HER2 and peptides , as well as the space among the interactions , were considered ., As shown in Fig 1C , Glu46 , Trp47 and Lys65 have low energy contributions to the HER2/4665 complex ., In addition , Asn52-Asn54 has almost no contributions to the binding ., Therefore , mutations were made mostly in two beta strands in 4665: one is from Val48 to Val51 with high hydrophobicity , and the other is from Ser55 to Phe64 ., The basic rules are that mutations should favor electrostatic and van der Waals interactions , and do not cause steric overlap ., Briefly , the van der Waals ( ΔEvdw ) contribution is a main component in HER2/4665 interactions as suggested from the free energy calculation ., As a consequence , residues in the strand of Ser55-Phe64 were preferably mutated to nonpolar amino acids ( Ser55 , Gly56 , Gly57 and Gln62 ) to increase the van der Waals ( ΔEvdw ) contribution ., Moreover , residues with large side chains are mutated into amino acids with similar side chain groups , such as Val48 , Ala49 , Gly56 and Gly57 ., We also took into account of the inter spaces between the peptide and HER2 ., For example , Ser58 is located on a beta strand that is close to the HER2 fragment Phe256-Lys314 but with a large spatial distance , so Ser58 was mutated to the residues with larger side chain groups or nonpolar amino acids ., Arg63 is also located on a beta strand that is closer to the HER2 fragment Phe256-Lys314 ., Considering its distance ( 3 . 9 Å ) to Phe257 , Arg63 is mutated into residues with side chain groups no larger than benzol methyl or nonpolar residues ., Finally , we first carried out 59 single mutations based on the 4665 sequence by performing MD simulations , and binding free energies were calculated for each mutant ., Seventeen single mutations with binding free energies < –48 . 53 kcal/mol were selected to create combinations of double mutations , thus resulting in another 56 mutants ., All mutations and their binding free energies are shown in Tables 1 and 2 ., Among these mutations , 34 sequences have lower binding free energies than the original 4665 peptide; peptide 58F63Y ( sequence: EWVADVNPNSGGFIYNQYFK ) is the lowest , and so it is expected to bind most tightly to HER2 ., The results of our previous work show that when receptor–ligand interactions are similar—save for only a few residue differences—computational binding free energy calculations can closely reflect the relative affinity of peptide binding 31 , 34 ., In the current study , to determine whether the above in silico screening correctly identified a peptide with affinity among the highest ones , the OBOC peptide library approach ( Fig 2A ) was later performed ., We designed the peptide library based on the calculated binding free energies of peptides/HER2 with single and double mutations , using MD simulations ( Tables 1 and 2 ) ., Mutations with binding free energies lower than the wild type 4665 ( ΔGtot < –50 kcal/mol ) were selected , and the intersection of single and double mutations from 55 to 64 was used in library construction ( Fig 2B ) , resulting in a library of 5184 sequences ., Biological screening of the OBOC peptide library is routinely carried out as described in the Materials and methods section ., Three positive beads were identified , and following MALDI–TOF–MS/MS analysis , one of them was found to be the same as the peptide 58F63Y , which has the lowest binding free energy with MD simulation ( Fig 2C ) ., After combining the results , peptide 58F63Y was selected for further experimental validation ., Another double mutant 55V63Y and its single mutations 58F , 63Y , and 55V—as well as the original wild type 4665—were also selected to facilitate better comparison ., The sequences of these six peptides were aligned using Clustal Omega; the results are shown in S1 Fig . Surface plasmon resonance imaging ( SPRi ) —which has been previously used to estimate interactions between molecules for the purposes of disease diagnosis , drug discovery , and peptide screening 23 , 24 , 35–37—was used in this study to estimate the dissociation constants of peptides binding to HER2 , as described in the Materials and methods section ., The dissociation constant was calculated from kinetic constants obtained by fitting association and dissociation curves to real-time binding and washing data ., While none of the four peptides shows any affinity to the HSA protein ( S2 Fig ) , Fig 3 indicates that the KD values of the peptides 4665 , 58F , 63Y , 55V , 58F63Y , and 55V63Y with HER2 protein are 9 . 86 μmol/L , 1 . 32 μmol/L , 1 . 54 μmol/L , 64 . 6 μmol/L , 0 . 536 μmol/L , and 8 . 16 μmol/L , respectively ., We can see that , to some extent , the KD values agree with the binding free energies from the simulation , which range from –48 kcal/mol to –66 kcal/mol ., This finding is consistent with that of our previous work , in which it was found that computational binding free energy from MM/GBSA can be used to estimate the relative affinity of peptide binding ., Based on the SPRi results , three peptides ( 58F , 63Y , 58F63Y ) with the highest affinity , as well as the starting peptide 4665 , were chosen for later confocal fluorescence imaging analyses ., Four tumor cell lines ( SKBR3 , MCF7 , MDA-MB-468 , and 293A ) , each with a different HER2 expression level , were used in confocal fluorescence imaging analysis to confirm the binding specificity of peptides to the HER2 protein ., Among these cell lines , the expression of HER2 was found to be high in SKBR3 , medium in MCF7 , and low in MDA-MB-468 and in 293A ., As shown in Fig 4 and S3 Fig , when treated with Cy5 . 5-peptides , fluorescent intensities are strong in SKBR3 cells , but weaker in MCF7 cells and absent in MDA-MB-468 and 293A cells ., Especially , SKBR3 cells treated with Cy5 . 5-58F63Y show the strongest fluorescence , thus indicating that 58F63Y binds with the highest affinity to HER2 ( Fig 4B ) ; this finding is consistent with the aforementioned MD calculation and SPRi analytical results ., All these results confirm that peptides can specifically bind at the cellular level to the extracellular domain of the HER2 protein ., Peptide 58F63Y ( which had the highest affinity ) and the wild type 4665 were chosen for subsequent in vivo studies ., Toxicity of peptides 58F63Y and 4665 to HUVEC and SKBR3 cells was also measured ., 4665 and 58F63Y shows no toxicity to both cell lines ( S4 Fig ) ., To investigate the affinity and specificity of peptides to HER2-positive tumors in vivo , nude mice bearing subcutaneous SKBR3 tumor xenografts were intravenously injected with Cy5 . 5-labeled peptides and Cy5 . 5 as the control; they were then subjected to whole-body optical imaging , using a small animal in vivo imaging system ( CRI Maestro 2 ) ., Fig 5A shows clear differences between the tumor images of mice with Cy5 . 5–58F63Y or Cy5 . 5–4665 and those of the control mice ., The intensities are plotted in bar charts ( Fig 5C ) that indicate that binding affinity increases 5 . 09-fold for 58F63Y and 3 . 52-fold for 4665 , relative to the control ., Moreover , fluorescence images of the dissected organs of the experimental mice , taken 30 min postinjection with Cy5 . 5-labeled peptides or control Cy5 . 5 , were acquired for further examination ., Both images and quantification in the bar charts ( Fig 5B and 5D ) show that tumors treated with 58F63Y and 4665 have significantly high fluorescence signals , compared to those in the controls ., Among all the organs , the kidney was found to have the highest background signal for both peptides and control , probably due to the toxic effect of Cy5 . 5 38 , 39 ., Taken together , all the results demonstrate that 58F63Y has high specificity and affinity for HER2-positive tumors ., To better understand the reason of the high affinity of 58F63Y to HER2 , a more detailed in silico analysis was performed ., Another four mutants—as well as the original wild type 4665—were also selected for a more comprehensive comparison ., To verify peptide structure stability after binding to HER2 , the root mean square deviation ( RMSD ) values of the backbone atoms of the initial structure and of successive simulated structures were calculated for all six HER2/peptides complexes ., S5 Fig shows that for all six complexes , the RMSD values become stable after about 5000 ps in the MD trajectories , thus indicating the convergence of each peptide and the complex structures towards an equilibrium state ., A glance at the results of the free energy decomposition analysis of the peptides found that more residues have low energies in the mutants than in 4665 ., Specifically , for 4665 , except Tyr60 , no other residue contributes energy < –2 . 5 kcal/mol ( Fig 6A ) ., However , in the mutants ( Fig 6B–6F ) —besides Asn54 and Ser55 in all mutants—the following were found to contribute energy < –2 . 5 kcal/mol: Tyr60 and Gln62 in 58F; Ser58 , Ile59 , and Tyr63 in 63Y; Gln62 in 55V; Ser58 , Phe59 , Gln62 , and Arg63 in 58F63Y; and Asn52 and Ile59 in 55V63Y ., That is to say , each mutant has more than three residues that are favorable to HER2 binding ., From S6 Fig , we can see that all six peptides overlay in the same pocket of the HER2 protein , thus indicating that the binding sites of these peptides remain the same as those in the wild type ., MM/GBSA binding free energy calculation and decomposition analysis of the HER2 protein reveals that fragment 236–314 has major interactions with these peptides , and that for each mutant , more than three residues have binding free energies < –3 kcal/mol ( S7 Fig ) ., Together , the beneficial changes of each mutant contribute to a lower binding free energy than that in wild type 4665 ., Models from PyMOL were used to better visualize the key interacting residues in peptides/HER2 complexes ( S8 Fig ) ., In Fig 7 , 18 residues pairs have distances of less than 5 Å in 58F63Y/HER2 ( 236-314 ) , compared with 10 pairs in 4665/HER2 ( 236-314 ) ., As a summary , more residues pairs with lower binding free energy and close distances in the 58F63Y/HER2 complex may contribute to the high affinity of 58F63Y ., In summary , based on the crystal structure of HER2/pertuzumab , we acquired a peptide that was 20 residues in length ( i . e . , 58F63Y ) that targets the HER2 protein; these findings were derived through the use of mutations and the computational calculation of the affinity with a combination protocol of molecular dynamics modeling , MM/GBSA binding free energy calculations , as well as the screening of an OBOC peptide library based on the mutations from the in silico modeling ., This work proves that MM/GBSA binding free energy can be used to reflect the relative affinity of peptide binding closely ., The peptide 58F63Y has a KD value of 536 nmol/L and binds to HER2 at the same site as the parent fragment of pertuzumab ., Through confocal fluorescence imaging and in vivo and ex vivo studies , the peptide was found to have high affinity and specificity for the extracellular domain of HER2 ., We expect this peptide to serve as an alternative probe that can be used in combination with others to improve the early detection , diagnosis , and targeted therapy of HER2-positive breast cancer ., The primary sequences of the pertuzumab fragment from 46 to 65 ( named 4665 ) and its mutants 58F , 63Y , 55V , 58F63Y , and 55V63Y were aligned by using the Clustal Omega program , which is available on EMBnet website ( http://www . ebi . ac . uk/Tools/msa/clustalo/ ) ., The model for the HER2/4665 complex derives from the crystal structure of the HER2 extracellular region and pertuzumab in the RCSB PDB ., The model for HER2/4665 was constructed based on the crystal structure of HER2/pertuzumab , by keeping related amino acids in pertuzumab ., Other models were constructed based on HER2/4665 ., The AMBER03 force field was used to investigate the potentials of the complexes in the following molecular mechanics minimizations and MD simulations 40 ., Missing atoms were added by using the tleap program ., The whole system was solvated with TIP3P 41 water molecules in a truncated octahedron box with a minimum solute box-edge distance of 12 Å 42 ., Then , the largest negative coulombic potential around the protein was randomly neutralized with counter-ions Na+ placed on the grids ., The numbers of water molecules and Na+ ions in each system are listed in the supporting information ( S1 Table ) ., To remove poor-quality contacts between the complex and the solvent molecules , three-step energy minimization was performed by using the sander module of AMBER12 prior to undertaking the MD simulations ., First , the whole protein was fixed and the water molecules and counter-ions were minimized; second , the backbone atoms of the protein were fixed and the side chains were minimized using the same settings as above; third , the whole system was minimized without any constraints ., The first two stages consisted of a 5 , 000-cycle steepest descent and a 2 , 500-cycle conjugate gradient minimization; the final step consisted of 10 , 000 cycles of steepest descent and 5 , 000 cycles of conjugate gradient minimization ., The SHAKE 43 method was applied to constrain covalent bonds related to hydrogen atoms , with a tolerance of 10−5 Å ., Particle Mesh Ewald 44 was employed to adequately deal with long-range electrostatic interactions , and in the MD simulations , the cutoff distances for nonbond energy interactions were set to 12 Å 45 ., Then , the entire system was gradually heated from 0 to 310 K in seven steps 46 , 47 over 60 ps 46 , 48 in the NVT ( canonical ensemble ) ., Finally , 10 ns MD simulations were implemented with a 2 fs 49 time step under the constant temperature of 310 K . During the sampling process , the trajectories were saved every 0 . 2 ps , and the conformations generated from the simulations were used in further analysis ., MM/GBSA 50 serves as an effective computational tool in analyzing biomolecular interactions ., When used with knowledge-based energy terms , MM/GBSA can help determine the binding free energies of all systems , based on the calculation of the average free energies of solvation ( ΔGbind ) between targeted protein and ligands over trajectories of MD simulation ., The MM/GBSA method can be summarized as the following equation ., In which , ΔGbind represents the binding free energy in solution consisting of the molecular mechanics free energy ( ΔEMM ) , the conformational entropic effect to binding ( −TΔS ) in the gas phase , and the solvation free energy containing polar contribution ( ΔGGB ) and nonpolar contribution ( ΔGSA ) ., The ΔEMM term includes ΔEele ( electrostatic ) and ΔEvdw ( van der Waals ) energies and was calculated by the sander module of AMBER12 ., The polar contribution was calculated by using the GB 51 mode , with solvent and the solute dielectric constants set to 80 and 4 , respectively ., Additionally , the nonpolar energy was estimated , with a solvent-probe radius of 1 . 4 Å: ΔGSA = 0 . 0072 × ΔSASA 52 , by the LCPO method 50 based on the SASA model 53 ., For each ligand–protein , 500 snapshots were taken from 7 to 10 ns on the MD trajectories ., Due to the low prediction accuracy and the high computational cost 54 , 55 upon the nmode module in AMBER12 as well as their similar values in analogical system 31 , 34 , the entropic contribution was ignored in the calculation of the predicted total binding free energy ( ΔGtot* means that ΔGtot does not contain -TΔS energy ) ., The specific inhibitor-residue interaction spectra were generated by using MM/GBSA decomposition analysis 28 , 56 undertaken through the mm_pbsa program of AMBER12 ., Four kinds of energy were found—namely , ΔEvdw , ΔEele , ΔGGB , and ΔGSA—and each contributed to the binding interaction of each ligand–residue pair ., The ΔEvdw and ΔEele energy terms were calculated by the sander module of AMBER12 ., The polar contribution ( ΔGGB ) to solvation energy was calculated by using the GB module and the parameters for the GB calculation were developed by Onufriev et al . 57 ., The nonpolar solvation contribution ( ΔGSA ) part was computed based on the SASA determined through the ICOSA method 52 ., All energy components were calculated by using 500 snapshots extracted from the last 3 ns of the MD trajectories ., After undertaking the decomposition process , the free energy contribution could be allocated to each residue from the association between the receptor and the ligand ., Graphic visualizations and presentations of protein structures were generated by using PyMOL 58–60 ., The OBOC peptide library was designed based on binding free energies derived from the MD simulations of single and double mutants that are lower than the HER2/4665 complex ., These mutations are in the 4665 fragment of pertuzumab from 55 to 64: EWVADVNPNX55X56X57X58X59X60N61X62X63X64K ., According to computational calculations , the V , S , and M mutants have lower energies for X55 ., This is also the case for G , M , and Y for X56; G , A , and V for X57; S , F , and H for X58; I and R for X59; Y and W for X60; Q and F for X62; R , W , Q , and V for X63; and F and R for X64 ., The result is a 3*3*3*3*2*2*2*4*2 = 5184 library capacity ., The OBOC library synthesis and screening was performed as per the previously used method 61–64 ., Briefly , HBTU ( 4 mmol ) and Fmoc-amino acid ( 4 mmol ) reagent was dissolved in 0 . 4 mol/L N-Methyl morpholine in N , N-dimethylformamide and coupled with the solid phase supporting materials for 40 min during the coupling step ., A 20% piperidine was used to remove the Fmoc group for 10 min in the deprotection step ., During the OBOC library synthesis , the amino acid coupling process was carried out in the “split” step , while the deprotection process was carried out in the “pool” step ., After elongation , a trifluoroacetic acid cleavage reagent was introduced to cleave the side chain protection group of each residue ., Afterwards , the solid phase supporting materials were incubated with 5% milk , then with HER2/biotin complex , and then with monodispersed magnetic streptavidin microspheres ., Each step was performed in an incubator at 37°C for 2 h , and followed by three washes with PBS ., The HER2 protein was biotinylated using a biotinylation kit ( Solulink Inc . , USA ) ., Positive beads with dark colors were picked out for the in situ chemical cleavage before MALDI–TOF–MS/MS analysis , and a 30 mg/mL cyanogen bromide solution was used overnight ., Peptides were synthesized using Fmoc strategy solid phase peptide synthesis 65–67 ., Unsophisticated peptides were purified using a Hitachi HPLC system ( L-7100 , Japan ) on a TSK gel ODS-100V reversed-phase column ., Peptides were eluted with a linear gradient of 5–80% acetonitrile containing 0 . 1% trifluoroacetic acid at a flow rate of 2 mL/min within 25 min ., Peptides were then subjected to MALDI–TOF–MS ( Bruker Daltonics ) analysis ., Purified peptides were dried in vacuum desiccators and then stored at –20°C until further use ., 2- ( 1H-benzotriazole-1-yl ) -1 , 1 , 3 , 3-tetramethyluronium hexafluorophosphate was purchased from GL Biochem ( China ) ., Trifluoroacetic acid and fluorescein 5-isothiocyanate were acquired from Sigma-Aldrich Co ., LLC ( USA ) ., N-Methyl morpholine and N , N-dimethylformamide were acquired from a Beijing chemical plant ( China ) ., For SPRi analysis , a cysteine residue linked to the amino terminal of all peptides was used for interacting with a bare gold chip bearing a 47 . 5-nm thickness ., First , 1 μL peptides at 1 mg/mL was added to the gold surface of the chip and incubated overnight at 4°C ., The chip was then washed with PBS and deionized water three times , and 5% nonfat milk was applied to block overnight at 4°C ., After the chip was washed again with PBS and water , it was dried with nitrogen for later use ., Human serum albumin ( HSA ) protein ( Sigma-Aldrich Co . LLC ) was used as the control ., HER2 ( Sino Biological Inc . , China ) and HSA proteins ( Sigma-Aldrich Co . LLC ) were dissolved in PBST and diluted to 10 , 5 , 2 . 5 , 1 . 25 , or 0 . 625 μg/mL ., The SPRi analytical procedure was carried out on the prepared SPRi chip by running PBST buffer for baseline stabilization , followed by the protein sample , a PBST running buffer for washing , and finally 0 . 5% H3PO4 in deionized water for regeneration ., This cycle was repeated for each concentration of HER2 and HSA protein at 20 , 10 , 5 , 2 . 5 , 1 . 25 , and 0 . 625 μg/mL ., Real-time binding signals were recorded and analyzed through the use of a PlexArray HT system ( Plexera LLC , Bothell , WA , USA ) ., The dissociation constant was calculated by fitting the association–dissociation curves ., Four cell lines ( SKBR3 , MCF7 , MDA-MB-468 , and 293A ) were seeded at a density of 3000 cells/mL into culture dishes and allowed to culture overnight with 5% CO2 at 37°C ., Cell nuclei was stained with 1 mM Hoechst 33342 in 200 μL cell culture medium and incubated at 37°C for 15 min ., Then , cells were incubated in culture medium with 50 μM Cy5 . 5-labeled peptide at 4°C for 20 min ., Finally , cells were washed three times with cold PBS for observation ., An Olympus FV1000-IX81 confocal-laser scanning microscope was used for confocal fluorescence imaging ., An FV5-LAMAR 633 nm laser was used as the excitation source , and the emission wave length was collected at 690 nm ., Hoechst 33342 was excited by a FV5-LD405-2 405 nm laser and collected within the range of 422–472 nm ., All microscope parameters were identical for all observations of the binding ability of the various peptides ., All animal experiments were conducted in compliance with the Beijing University Animal Study Committee’s requirements vis-à-vis the care and use of laboratory animals ., The Beijing University Animal Study Committee approved the experimental protocols ., Five to six-week-old Balb/c female nude mice were subcutaneously administered approximately 1 × 107 SKBR3 cells into the right hind leg , to establish xenografted tumors ., Thereafter , tumor size was periodically measured with a caliper , and mice with tumors of 6−8 mm in diameter were selected for the following small animal experiments ., Cy5 . 5–NHS ( Lumiprobe ) was used to label peptides ., Either Cy5 . 5-peptides or the control Cy5 . 5 ( 1 μM , 200 μL ) was intravenously injected into tumor-bearing nude mice via the tail vein ., The mice were anesthetized and fluorescence signals measured using the small animal in vivo imaging system 30 min postinjection ., Three mice were used for each peptide and for control ., Near-infrared fluorescence imaging of tumor-bearing nude mice were taken with an exposure time of 50 ms , using the Cy5 . 5 filter sets ( excitation: 673 nm; emission: 707 nm ) , and the intensities were quantified using the same software ., Then , fluorescence images of the main organs and of tumors dissected from nude mice were individually taken as above .

Introduction, Results/Discussion, Methods

A high level of HER2 expression in breast cancer correlates with a higher tumor growth rate , high metastatic potential , and a poor long-term patient survival rate ., Pertuzumab , a human monoclonal antibody , can reduce the effect of HER2 overexpression by preventing HER2 dimerization ., In this study , a combination protocol of molecular dynamics modeling and MM/GBSA binding free energy calculations was applied to design peptides that interact with HER2 based on the HER2/pertuzumab crystal structure ., Based on a β hairpin in pertuzumab from Glu46 to Lys65—which plays a key role in interacting with HER2—mutations were carried out in silico to improve the binding free energy of the hairpin that interacts with the Phe256-Lys314 of the HER2 protein ., Combined the use of one-bead-one-compound library screening , among all the mutations , a peptide ( 58F63Y ) with the lowest binding free energy was confirmed experimentally to have the highest affinity , and it may be used as a new probe in diagnosing and treating HER2-positive breast cancer .

Many therapeutic approaches , including the human monoclonal antibodies trastuzumab and pertuzumab , target the human epidermal growth factor receptor 2 ( HER2 ) of any breast cancer that features HER2 overexpression ., Compared to these antibodies , peptides have many advantages , including lower cost , easier synthesis , high affinity , and lower toxicity ., Here , we first designed peptides that interact with HER2 protein based on the HER2/pertuzumab crystal structure ( PDB entry: 1S78 ) , using a combination protocol of molecular dynamics modeling , molecular mechanics/generalized Born solvent-accessible surface area ( MM/GBSA ) binding free energy calculations ., Then , combined with the peptide library screening , six peptides were selected for further analysis and experimental validations ., The results of ex vivo and in vivo experiments confirmed that one peptide ( 58F63Y ) in particular has a strong affinity and high specificity to HER2-overexpressing tumors ., This may due to more paired residues and lower binding free energy in peptide 58F63Y and HER2 complex based on free energy decomposition analysis and distances calculation ., While both in silico and in vitro screenings point to the same high-affinity peptide , the findings suggest that in silico screening based on calculated binding free energies is rather reliable ., Additionally , based on the calculation of binding free energies among mutants , we can reduce the library capacity of one-bead-one-compound screening ., In summary , we present a rather simple and rapid means of deriving a peptide with a clear binding site to its target protein .

fluorescence imaging, chemical characterization, medicine and health sciences, breast tumors, crystal structure, condensed matter physics, cancers and neoplasms, oncology, peptide libraries, molecular biology techniques, crystallography, thermodynamics, research and analysis methods, solid state physics, imaging techniques, breast cancer, proteomics, binding analysis, molecular biology, molecular biology assays and analysis techniques, free energy, physics, biochemistry, biochemical simulations, library screening, biology and life sciences, physical sciences, computational biology

journal.pgen.1004686

Cell-Autonomous Progeroid Changes in Conditional Mouse Models for Repair Endonuclease XPG Deficiency

If DNA damage , either inflicted from exogenous or endogenous sources , cannot be repaired , this has detrimental consequences for an organism ranging from transcription blocks , permanent cell cycle arrest and mutations , to cell death ., In the end , this unrepaired DNA damage contributes to the onset and progression of the aging process , as well as to cancer 1–3 ., Cells are equipped with a set of elaborate DNA repair mechanisms integrated into a complex DNA damage response machinery that jointly attempt to fix the unrepaired DNA 4 ., One such DNA repair mechanism is the Nucleotide Excision Repair ( NER ) pathway that removes a wide category of helix-distorting lesions , such as those induced by UV and bulky chemical adducts , in a tightly coordinated process involving over 30 proteins 5–7 ., NER can be divided into two subpathways based on the mode of damage recognition ., The Global Genome ( GG- ) NER subpathway specifically involves the XPC and XPE protein complexes , and probes the entire genome for lesions that disrupt base-pairing 5 , 7–9 ., Transcription-Coupled ( TC- ) NER , on the other hand , detects helix-distorting lesions that stall transcription in the transcribed strand of expressed genes , and hence enables resumption of transcription ., TC-NER is independent of XPC and XPE and specifically involves proteins such as CSA , CSB and UVSSA 8 , 10 , 11 ., After lesion recognition , the subsequent ‘cut-and-patch’ core repair reaction encompasses local opening of the DNA helix and lesion verification , performed by the TFIIH complex together with XPA ., Both correctly position the structure-specific endonucleases ERCC1/XPF and XPG for strand-specific excision of the lesion as part of a 22–30 bp oligonucleotide 5 , 7 , 12 ., Finally , the gap is filled by repair synthesis and closed by ligation 5 , 7 , 12 ., Multiple NER proteins have been attributed additional roles , both in DNA repair pathways other than NER , and in transcription regulation ., For instance , TFIIH is an essential component of the general transcription machinery 13 , 14 , but also other NER factors , including XPG and CSB , have been implicated in transcription regulation 15–18 ., The 5′ endonuclease ERCC1/XPF participates in the repair of interstrand crosslinks 19 , 20 and subpathways of DNA double-strand break repair 21 ., XPC , CSB , and XPG have been individually implicated in promoting base excision repair ( BER ) of oxidative DNA damage 22–30 ., TFIIH and XPG are , together with CSB , thought to be involved in the early steps of Transcription-Coupled Repair ( TCR ) , and XPG interacts directly with both CSB and RNA Polymerase II 31 ., Although still controversial , there are accumulating reports that TCR not only directs NER to blocked transcription but may also recruit BER for preferential repair of oxidative DNA damage in transcribed strands 32–34 ., Such a mechanism might be related to the roles of both CSB and XPG in promoting BER more globally ., If correct , it could explain the much greater consequences for the organism of TCR defects compared to defects in NER alone ( see below ) 35 ., A number of rare , autosomal recessive disorders resulting from mutations in NER genes underscore the importance of genome maintenance for the prevention of cancer as well as aging 3 ., NER-associated diseases are characterized by sun ( UV ) hypersensitivity and include xeroderma pigmentosum ( XP ) , UV-sensitivity syndrome ( UVSS ) , Cockayne syndrome ( CS ) , cerebro-oculo-facio-skeletal ( COFS ) syndrome , XPF-ERCC1 ( XFE ) progeroid syndrome , trichothiodystrophy ( TTD ) and disorders that combine the symptoms of these syndromes , including XP/CS 35–39 ., XP originates from defects in GG-NER or total NER activity and is characterized by an over 2000-fold increased risk of cancer in sun-exposed skin and , to a much lesser extent , in internal organs 36 ., XP patients may also develop progressive neurological symptoms and neuronal degeneration depending on the severity of the total NER deficiency 36 , 39 , 40 ., UVSS is characterized by skin UV hypersensitivity without actually developing skin cancer ., UVSS results from the selective loss of TC-NER function as a consequence of mutations in the proteins involved in detection of UV-induced transcription-blocking DNA lesions , i . e . UVSSA , CSA , and CSB 11 , 35 , 41–44 ., Mutations in CSA and CSB , however , generally cause CS , a heterogeneous multisystem disorder that , in addition to UV-sensitivity , is characterized by severe growth failure and cachexia , accelerated aging features , short lifespan , and progressive sensori-neuronal abnormalities 38 , 45 ., The severe symptoms of CS cannot be explained by the loss of TC-NER function as they do not occur in fully NER-deficient XP patients and TC-NER deficient UVSS patients ., Therefore , CS symptoms have been linked to additional , yet incompletely , defined functions of CSA and CSB in DNA repair , transcription regulation , other processes , or a combination of deficiencies 3 , 46 , 47 ., The same applies for mutations in the down-stream NER factors XPB , XPD , XPF , ERCC1 and XPG that cause combined XP/CS , or severe developmental/degenerative multisystem disorders such as COFS and XFE that share multiple features with severe CS forms 35 , 48–51 ., Thus CS symptoms can result from mutations in multiple proteins that operate together in NER , but the symptoms caused by these mutations cannot be explained by NER deficiency alone , raising questions about the identities of these non-NER activities underlying CS symptoms and the extent to which different symptoms reflect deficits of different cellular processes 35 , 46 ., Mutations in the structure-specific NER 3′-endonuclease XPG are rare , with less than 20 patients and 25 mutant alleles described so far 52–55 , and cause a spectrum of disease phenotypes varying from XP to XP/CS and COFS 53 ., Point mutations that selectively eliminate XPG nuclease activity cause XP , while C-terminal truncations , destabilizing point mutations , and mutations that abolish the interaction between XPG and the basal transcription factor TFIIH cause XP/CS and COFS 52–58 ., These data support the notion that a deficient function of XPG outside NER is responsible for the severe CS symptoms 15 , 52 , 53 , 55–57 ., For most NER disorders , mouse mutants have been generated that mimic the genetic defect found in patients , and to various extents reproduce XP and CS-like features as well as the progeroid hallmarks found in the corresponding human syndrome 59–62 ., Accordingly , Xpg-null ( Xpg−/− ) mice were found to develop a severe phenotype characterized by growth deficiency and very short lifespan , resembling severe XP/CS 63 ., In contrast , Xpg mutant mice carrying amino acid substitutions that selectively abolish the nuclease function of XPG ( XpgE791A and XpgD811A ) show severe UV-sensitivity but normal lifespan , hence , reproducing the XP phenotype 64 , 65 ., In addition , a mutant XPG construct containing a C-terminal truncation lacking the last 360 amino acids that was made to mimic the genotype of some XP-G/CS patients , developed a growth deficiency and short-living phenotype resembling that of Xpg−/− mice , albeit somewhat milder 65 ., Yet another C-terminal truncation mutant lacking the last 180 amino acids showed a normal lifespan , but produced a CS-like growth-deficient short-living phenotype after crossing with Xpa−/− mice that are already fully NER-deficient 65 , 66 ., Significantly , the same conversion of a normal lifespan into a short-living mouse model is observed after crossing CSA- or CSB-deficient CS mice with total NER- ( Xpa−/− ) or GG-NER ( Xpc−/− ) deficient mouse models 67– , but not by crossing NER-deficient XpgD811A with Xpa−/− mice 66 ., Together these data indicate a deleterious synergistic interaction between NER deficiency and loss of non-NER activities that underlie CS ., Furthermore , they show that the C-terminus of XPG could play a role in the CS symptoms , and that the XPG-deficient Xpg−/− mice may reproduce the phenotype of Xpa−/−Csb−/− , Xpc−/−Csb−/− or Xpa−/−Csa−/− double mutant mice 62 , 66–69 ., In previous analyses we clearly observed progeroid characteristics in many NER mutant mouse models including Xpa/Csb , Xpb , Xpd , and Ercc1 mutants 51 , 68 , 70–72 , yet the occurrence of progeroid features in Xpg−/− mice has hitherto been poorly established , mostly due to their very short lifespan ., Since we are particularly interested in the effect of Xpg deletion on organ-specific aging , we generated a conditional Xpg mutant ., As genetic background can have a significant effect on phenotype development , we first re-examined the pathological characteristics of Xpg−/− mice in a C57BL6/FVB F1 hybrid background , as was previously described for Ercc1 mutant mice 51 , 72 ., In this hybrid background Xpg−/− mice lived longer and presented progeroid features including cachexia and osteoporosis with pronounced degenerative phenotypes in both liver and brain ., We next studied the effect of liver- and forebrain-specific inactivation of Xpg , showing that the observed phenotypes are indeed due to lack of XPG protein ., Together our data show that , consistent with previous data in ERCC1- and CSB/XPA-deficient mice , Xpg−/− mice develop a multisystem progeroid degenerative phenotype ., To generate a Cre-inducible Xpg knockout allele we flanked the third exon of Xpg with LoxP sites ( Figure 1A ) ., Deletion of this exon causes a frame shift and a premature translational stop immediately after exon 2 at amino acid residue 89 ( instead of the full-length 1170 ) ., After transfection to 129 ES cells and selection of properly targeted clones ( Figure 1B and Materials and Methods ) , two independent transfected clones were used to generate germ-line transmitting chimeras ( Figure 1C ) ., Heterozygous males , carrying the conditional Xpg allele , were crossed to females ubiquitously expressing Flp for excision of the Neomycin cassette and to yield mice that are heterozygous for the floxed Xpg ( Xpgf ) allele ., Xpgf/+ mice were backcrossed and maintained in FVB/N background ., To generate Xpg mice carrying a knockout allele ( Xpg− , Figure 1A ) , Xpgf/+ mice were crossed to Cag-Cre mice , which ubiquitously express Cre recombinase from germline 73 , yielding heterozygous Xpg+/− animals ( Figure 1C ) ., Xpg+/− animals were>10 times backcrossed into C57BL6 or FVB/N genetic backgrounds ., Unless otherwise stated , experiments were performed with Xpg−/− mice in the C57BL6/FVB F1 hybrid background obtained from intercrossing C57BL6 Xpg+/−×FVB/N Xpg+/− animals to minimalize background specific effects ( see below ) ., The presence of a premature stop codon in the Xpg− allele was confirmed by sequencing Xpg cDNA from liver of Xpg−/− mice ( Figure S1A ) ., Accordingly , Western immunoblot analysis with an antibody raised against the central spacer region ( R-domain ) of XPG shows the absence of XPG protein product in mouse dermal fibroblasts ( MDFs ) isolated from Xpg−/− mice ( Figure 1D ) ., Next , we tested DNA repair deficiency of Xpg−/− MDFs ., In accordance with complete NER deficiency , Xpg−/− MDFs showed an almost 10-fold hypersensitivity to UV , similar to fully NER-defective MDFs derived from Xpa−/− mice ( Figure 1E ) 74 ., In addition , Xpg−/− MDFs were hypersensitive to treatment with Illudin S ( Figure 1E ) , consistent with the loss of TC-NER function 75 , and were deficient in UV-induced unscheduled DNA synthesis in line with loss of GG-NER activity ( Figure 1F ) ., Also , recovery of RNA synthesis after UV exposure was almost completely abolished in Xpg−/− MDFs , further demonstrating loss of TC-NER activity ( Figure 1G ) ., Xpg−/− MDFs showed no increased sensitivity to potassium bromate ( KBrO3 ) which causes oxidative DNA lesions , and a minimal increased sensitivity to the cross-linking agent cisplatin ( Figure S1B ) ., In view of a significant effect of genetic background on embryonic lethality and lifespan in ERCC1-deficient mice 51 , 72 , we examined whether a similar genetic background effect occurred in Xpg−/− mice , by comparing birth frequencies and lifespan of Xpg−/− mice in a C57BL6 , FVB/N or a C57BL6/FVB F1 hybrid background ., In C57BL6 background birth frequencies were below Mendelian expectations ( ∼8% , Table 1 ) , whereas in the FVB/N and C57BL6/FVB F1 hybrid background birth frequencies were Mendelian and near-Mendelian , respectively ( Table 1 ) ., Also the lifespan of Xpg−/− animals was strongly dependent on genetic background , with C57BL6 Xpg−/− mice showing a lifespan of 3 weeks , and Xpg−/− animals in FVB/N and C57BL6/FVB F1 hybrid background living for 15–18 weeks ( Figure 2A ) ., Further analysis of C57BL6/FVB F1 Xpg−/− mice showed that they had the same size and weight as wild type and heterozygote littermates at late embryonic stage ( E17 . 5; Figure 2B and C ) ., However , after birth , Xpg−/− mice showed reduced growth and weight gain compared to controls , and stopped growing at 6–8 weeks when their body weight was about 65–70% of that of wild type littermates ( Figure 2C ) ., From 10–11 weeks , body weights declined and the Xpg−/− mice became progressively cachectic ( Figure 2C and D ) ., At 14 weeks all Xpg−/− mice were severely runted ( Figure 2D ) , and the mice died a few weeks thereafter between 15–18 weeks of age ( Figure 2A and E ) ., The growth deficiency was paralleled by the development of kyphosis ( Figure 2D and F ) ., In addition , Xpg−/− mice progressively developed neurological symptoms , including clasping of the hind-limbs when lifted by their tails ( Figure S2A ) , and at a later time point fine tremors ( Figure 2E ) ., Accelerating rotarod and grip strength tests in 14-week old Xpg−/− mice revealed severe motor deficits and muscle weakness at this age ( Figure S2B and C ) ., Computed tomography ( CT ) confirmed severe kyphosis in Xpg−/− mice at 16 weeks of age ( Figure 2F ) ., To further examine skeletal abnormalities and the occurrence of osteoporosis as observed in other NER-deficient mouse models 68 , 76–79 , we measured several bone parameters using femoral bones ., Analysis of bone strength revealed decreased strength of the Xpg−/− femoral bone at 14–16 weeks ( Figure 2G ) ., Micro-CT analysis showed that the thickness of the trabeculae and cortex of the femoral bones was significantly smaller in Xpg−/− compared to wild type mice at 14–17 weeks , but not yet at 7 weeks ( Figure 2H ) ., Overall , these data indicate a progressive increase of age-related features such as osteoporosis ., Weight loss and reduced size of Xpg−/− mice was associated with reduced weight of internal organs ( Figure S3A ) and with a strong reduction in the amount of subcutaneous fat ( Figure S3B ) ., The Xpg−/− mice previously reported by Harada et al . 63 showed developmental abnormalities of the gastro-intestinal tract ., These gastro-intestinal abnormalities were proposed to be a major contributor of the post-natal growth failure and short lifespan ( <3 weeks ) of their animals 63 ., However , in contrast to their data , the gastro-intestinal tract of our Xpg−/− mice had a normal size and macroscopic appearance , and showed a normal histological appearance in HE-stained sections ( Figure 3A ) ., Furthermore , staining for the proliferative cell marker Ki-67 indicated that the number of proliferative cells in the intestinal epithelium was similar between Xpg−/− and wild type mice ( Figure 3A ) ., In accord with normal function of the gastro-intestinal tract we found that food intake per gram body weight was similar between wild type and Xpg−/− animals ( Figure S3C ) ., The liver is a central organ in many aspects of metabolic control , including regulation of circulating glucose levels and detoxification , and it plays a key role in regulation of IGF1-somatotrophic axis signaling ., Previous studies have shown that ERCC1/XPF-deficient mice develop multiple liver abnormalities , in particular anisokaryosis resulting from polyploidy , and intranuclear inclusions 72 , 80–83 ., Analysis of HE-stained liver sections of our Xpg−/− mice revealed mild anisokaryosis , and increased mean nuclear size at 14 weeks , but not at 4 weeks of age ( Figure 3B ) ., In addition , sporadically , hepatocytes had intranuclear inclusions ., These liver nuclear changes are a well characterized phenomenon in the aging liver , and indicate that Xpg−/− mice show features of accelerated aging in the liver similar to Ercc1Δ/− mice 72 , 82 ., Liver cells of ERCC1-deficient and other progeroid NER-deficient mouse mutants display changes in gene expression that encompass a downregulation of catabolic and oxidative metabolism and an upregulation of antioxidant and stress defense pathways , suggestive of a compensatory survival response to cope with increased DNA damage 51 , 68 , 84 , 85 ., To determine whether Xpg−/− liver cells also display a ‘survival-like’ stress response , we determined expression levels of selected antioxidant and somatotrophic genes by real-time PCR ., Indeed , mRNA levels from a subset of antioxidant effector genes , including Nqo1 , Srxn1 , Gstt2 and Gsta1 , were significantly increased in liver homogenates of young ( 7-week old ) Xpg−/− animals compared to controls ., At 14 weeks of age , we observed a similar significant increased expression of Nqo1 and Gsta1 while mRNA from the other antioxidant effector genes tested showed unaltered expression ( Figure 3C ) ., Expression levels of Nrf2 , which is a potent inducer of the antioxidant response element ( ARE ) , were unaltered , in line with the fact that NRF2-activation is largely achieved by post-translational mechanisms 86 ., As increased expression of antioxidant genes could be an indication of increased genotoxic stress , we also checked the expression of the p53-responsive kinase inhibitor p21 , a master regulator of cell survival and death 87 , which is generally increased after DNA damage and was previously shown to be elevated in livers of Ercc1 mutant mice 88 ., Expression levels of p21 doubled at the age of 7 weeks and were massively increased at the age of 14 weeks , indicative of genotoxic stress caused by the absence of XPG ., To determine changes in somatotrophic gene expression we examined mRNA levels of Ghr , Igf1r , Igf1 and Igfbp3 ., We found a two-fold suppression of Ghr and Igf1r mRNA expression at 7 weeks , and a significant downregulation of Ghr mRNA levels at 14 weeks of age ( Figure 3D ) ., Together the data indicate that the Xpg−/− liver in part reproduces gene expression changes observed in other short-living NER-deficient mice , which we refer to as a survival-like stress response ., Finally , consistent with other NER-deficient progeroid mice 51 , 85 , we found significantly reduced steady-state blood glucose levels in Xpg−/− mice ( Figure 3E ) ., The occurrence of neurological abnormalities and impaired motor behavior in Xpg−/− mice ( Figure 2E and S2 ) , as well as the abundant neurodegenerative features in ERCC1-deficient and combined XP/CS mouse models 89–91 , prompted us to investigate the central nervous systems of Xpg−/− animals for neurodegenerative changes ., Macroscopically , the brains and spinal cords of Xpg−/− mice showed a normal appearance , albeit somewhat smaller ., In addition , the gross histological organization analyzed in thionin-stained sections appeared normal in all brain regions ., As a first step to examine the occurrence of neurodegenerative changes , we examined the brains of 4- and 14-week old Xpg−/− mice immunohistologically for glial acidic filament protein ( GFAP ) expression , which outlines reactive astrocytosis in response to neuronal injury ., A mild increase in GFAP immunostaining occurred in patches in multiple nervous system areas at 4 weeks ( Figure 4A and S4A ) ., Instead , at 14 weeks , Xpg−/− mice showed a prominent ubiquitous increase in GFAP staining throughout the entire central nervous system including spinal cord , indicative of widespread astrocytosis ( Figure 4A and S4A ) ., Double-labelling of GFAP and the microglia cell marker Iba-1 showed that the increased GFAP staining was paralleled by microglia activation , characterized by increased Iba-1 immunoreactivity and the transformation of resting microglia cells into activated cells with thicker processes and larger cell bodies ( Figure S4B and C ) ., Next , to determine whether Xpg−/− central nervous system cells experience genotoxic stress , we studied the expression of the transcription factor p53 , which is activated by multiple types of DNA damage and is expressed in neurons and macroglia of many NER-deficient mouse models including mice defective in Ercc1 , Csa or Csb 89–91 ., Immunohistochemistry revealed p53-positive cells in all central nervous system regions ., Analysis of the p53 density in neocortex and cerebellum indicated an increase in number of p53-positive cells in brains of 14-week old compared to 4-week old Xpg−/− mice ( Figure 4B ) ., Similar to our findings in other NER mutant mice 89–91 , double labelling of p53 with neuronal ( NeuN ) and astrocytic ( GFAP , S100β ) markers , indicated that these p53-positive cells include neurons , astrocytes ( GFAP+ or S100β+; Figure S4D ) , and oligodendrocytes ., Although not systematically investigated , we also noted that , as in other NER-deficient mice , in neocortex and cerebellar cortex the majority of p53-positive cells were neurons , while in spinal cord a large proportion of p53-positive cells were astrocytes ( Figure S4E ) ., To obtain evidence for the occurrence of neuronal death , we analyzed calbindin staining in cerebellar cortex where it outlines Purkinje cells and enables easy detection of the degeneration of these cells 90–92 ., Calbindin staining revealed degeneration and loss of Purkinje cells in 14-week old Xpg−/− mice ( Figure 4C ) ., Also , calbindin staining revealed sporadic Purkinje cells with abnormal dendritic morphologies and , more frequently , Purkinje cells with swellings in their proximal axon ( Figure 4C ) ., Axonal swellings ( also designated torpedoes or axonal spheroids ) are a common feature in neurodegenerative disorders and aging 93 , that is also well documented for Purkinje cell axons of CS and XP/CS patients 94 ., The presence of axonal pathology indicates that many surviving Purkinje cells in 14-week old Xpg−/− mice display compromised health ., Notably , few small axonal swellings occurred in Purkinje cells axons in 4-week old Xpg−/− mice ., To further examine the extent to which neurons in Xpg−/− mice show compromised health , we examined the morphology of the Golgi apparatus in motor neurons ., In a previous study in Ercc1 mutant mice we noted that motor neurons displayed a variety of morphological abnormalities of the Golgi apparatus , and we proposed that these abnormalities reflect a heterogeneity of cellular deficits resulting from stochastic DNA damage 90 ., Immunostaining for the cis-Golgi marker GM130 showed that motor neurons in Xpg−/− mice developed the same heterogeneity in morphological abnormalities of the Golgi apparatus as observed in Ercc1Δ/− mice ., Double labelling of GM130 and p53 indicated that only a small subset of neurons with abnormal Golgi apparatus is p53 positive ., This variability in p53 expression further illustrates the heterogeneity of degenerative events that may occur in Xpg−/− neurons ( Figure S4F ) ., TUNEL staining to determine the amount of apoptotic cells showed a significant increase in both the cerebrum and the cerebellum at 4 as well as 14 weeks of age ( Figure 4D ) ., Finally , real-time PCR in Xpg−/− cerebellum revealed an upregulation of the p53-responsive kinase inhibitor p21 consistent with the activation of genotoxic stress pathways ( Figure 4E ) , and increased expression of several oxidative stress response genes ( Figure 4E ) ., In addition to the brain and spinal cord , we also investigated the retina , as retinal degeneration is a frequent symptom of CS and XP/CS patients 38 , that is also reproduced in CSA- and CSB-deficient mice 23 ., TUNEL staining revealed cell death in both the inner and outer nuclear layers of the retina of 4- and 14-week old Xpg−/− mice ( Figure 4F ) ., Hence , Xpg−/− mice display loss of photoreceptor cells as well as degeneration of the retinal circuitry ., Together these data indicate the occurrence of widespread progressive degenerative changes in Xpg−/− nervous system , strongly resembling the phenotype of ERCC1-deficient mice ., Transgenic expression of ERCC1 in the liver has been shown to alleviate growth deficiency and to extend lifespan of ERCC1-deficient mice 83 , suggesting that liver abnormalities are an important determinant of the reduced lifespan of these mice ., To determine the importance of liver pathology in the runting and the reduced lifespan of our Xpg−/− mice , we generated mice with liver-specific inactivation of the Xpg gene by crossing Xpgf/+ mice carrying the floxed Xpg allele with heterozygous Xpg+/− mice that also express the albumin-Cre recombinase transgene that drives Cre expression specifically in hepatocytes 95 to yield Xpgf/−/Alb-Cre+ mice , hereafter designated Alb-Xpg mice ., This Alb-Xpg mouse also has the advantage that it allows to study the effect of liver XPG-deficiency in the absence of abnormalities in other tissues ., A cohort of Alb-Xpg mice was allowed to reach the age of one year ., All Alb-Xpg mice displayed normal growth and weight gain , and none of them died prematurely ( Figure 5A ) ., Livers of Alb-Xpg mice had an increased size compared to wild type , while brain , kidney and spleen displayed unaltered size and weight ( Figure S5A ) ., Albumin and glucose blood levels were the same as in control mice ( Figure S5B and C ) ., Histological analysis revealed anisokaryosis with karyomegaly in the liver of Alb-Xpg mice analyzed at 26 and 52 weeks ( Figure 5B ) ., The observed karyomegaly was more prominent than that observed in 14-week old Xpg−/− mice , and cells with intranuclear inclusions were more frequent ., In addition , we identified p53-positive cells , increased cell death and increased cell proliferation in Alb-Xpg liver consistent with a progeroid degenerative phenotype ( Figure 5C–E ) ., Furthermore , real-time PCR showed that livers of Alb-Xpg mice displayed a massive induction of the DNA damage response gene p21 as well as increased expression of several antioxidant effector genes ( Figure 5F ) ., We also observed a trend of reduced expression of Ghr and Igf1r ( Figure 5G ) , hence , reproducing gene expression changes determined in livers of Xpg−/− mice ., Activation of the Nrf2 antioxidant response genes , reduction of the IGF1 axis , and increased proliferation shown by Ki67-staining are all consistent with liver regeneration after tissue damage 96 ., As an additional control we showed that the expression of these genes is unaltered in livers from Emx1-Xpg mice that are XPG-deficient in the dorsal forebrain ( see below; Figure S5D and E ) ., Together the data from the Alb-Xpg mice show that progeroid and gene expression changes in the liver triggered by the absence of XPG are not sufficient to explain the runted short-living phenotype of Xpg−/− mice ., Our neuropathological analyses of Xpg−/− mice uncovered severe neurodegenerative changes at 14 weeks of age , compatible with neurological and motor deficits in these mice ., Importantly , the presence of p53 in glia and abundant astrocytosis and microgliosis in the white matter indicate that abnormalities in Xpg−/− mice are not limited to neurons , but also involve glia cells ., This is consistent with our findings in CS mouse models 62 , 91 , and with the neuropathological changes found in CS and XP/CS patients that is dominated by white matter pathology , in addition to neuronal , glial and vascular pathology , and , in severe cases , developmental abnormalities 39 , 94 , 97–99 ., In previous studies , using a Cre-lox approach with CamKIIα-Cre and L7-Cre transgenic mice that drive Cre expression in post-mitotic forebrain neurons and Purkinje cells , respectively , we showed that neuron-specific deficiency of ERCC1 or combined deficiency of XPA and CSB was sufficient to trigger stochastic degeneration of these neuronal populations 89 , 91 , 92 ., These studies showed that neurodegenerative changes in ubiquitous ERCC1- and XPA/CSB-deficient mice are not a consequence of developmental abnormalities , vascular problems , or degenerative changes in other organs ., Furthermore , these neuron-specific mice enabled us to follow the degenerative process beyond the normal lifespan of the short-living ubiquitous ERCC1- and XPA/CSB-deficient mice 89 , 91 , 92 ., In the present study we therefore used a similar approach , but with an Emx1-Cre transgenic line that drives Cre expression in progenitor cells of the dorsal telencephalon , to achieve inactivation of the Xpg gene not only in excitatory neurons of the neocortex and hippocampus , but also of astrocytes and oligodendrocytes in these brain areas 100 ., Analysis of a cohort of Emx1-Xpg mice that were allowed to age for one year revealed no early death and showed that body weights were indistinguishable from that of wild type littermates until the age of 30 weeks ., At older age the mean weight gain of Emx1-Xpg mice was significantly smaller than in control littermates ( Figure 6A ) ., This difference in weight was associated with a proportional reduced weight of internal organs ( Figure S6A ) ., Basal blood glucose concentrations were the same as in controls , indicating that reduced weight of old Emx1-Xpg mice is not a consequence of reduced energetic intake ( Figure S6B ) ., To obtain a crude impression of the development of neurological symptoms , we examined the time of onset of clasping of the hind limbs when animals were lifted by their tails ., This abnormality developed in Emx1-Xpg between 14–24 weeks of age and was not observed in control littermates ( Figure 6B ) ., The Emx1-Xpg mice did not develop tremors and deficits in accelerating rotarod performance ( Figure S6C ) ., Both the absence of these motor deficits and the delayed onset of clasping in comparison to the Xpg−/− mice can be explained by the selective inactivation of Xpg in neocortex and hippocampus , avoiding the bulk of circuitries controlling motor behavior in mice 101 ., Macroscopic inspection of brains of Emx1-Xpg mice at 26- and 52-week already revealed that the neocortex was considerably smaller ., Histological analysis confirmed that the neocortex was thinner , and showed that also the hippocampus was dramatically smaller , while other brain regions were unaltered ( Figure S6D ) ., GFAP immunohistochemistry showed a very strong increase in GFAP staining indicative of astrocytosis in both cortex and hippocampus ( Figure 6C ) , and no changes in other brain regions ., As in the Xpg−/− nervous system , astrocytosis was paralleled by microgliosis , identified by immunohistology with an Iba-1 antibody ., Staining for Mac2 ( also known as galectin-3 ) , to outline phagocytosing microglia cells 102 , revealed very high levels of Mac2-positive cells in the corpus callosum and the fimbria fornix of Emx1-Xpg mice ( Figure 6D and S6D , E ) , and a moderate amount of phagocytosing microglia in the neocortex and hippocampus ( Figure 6D ) ., Remarkably , the presence of Mac2 could not be explained by axonal degeneration of cortical and hippocampal neurons solely , as we did not observe this phenomenon in our CamKIIα-Ercc1 mice 89 ., Furthermore , the capsula interna of Emx1-Xpg mice , which contains the descending corticofugal axons and shows severe axonal degeneration , did not show this dramatic increase of Mac2 staining ( Figure 6D and S6D ) ., Hence , the presence of high levels of Mac2 labelling may reflect the same phenomenon that we observed in our CS mice , i . e . the presence of phagocytosing microglia in the absence of axonal degeneration 91 ., Accordingly , we also found

Introduction, Results, Discussion, Materials and Methods

As part of the Nucleotide Excision Repair ( NER ) process , the endonuclease XPG is involved in repair of helix-distorting DNA lesions , but the protein has also been implicated in several other DNA repair systems , complicating genotype-phenotype relationship in XPG patients ., Defects in XPG can cause either the cancer-prone condition xeroderma pigmentosum ( XP ) alone , or XP combined with the severe neurodevelopmental disorder Cockayne Syndrome ( CS ) , or the infantile lethal cerebro-oculo-facio-skeletal ( COFS ) syndrome , characterized by dramatic growth failure , progressive neurodevelopmental abnormalities and greatly reduced life expectancy ., Here , we present a novel ( conditional ) Xpg−/− mouse model which -in a C57BL6/FVB F1 hybrid genetic background- displays many progeroid features , including cessation of growth , loss of subcutaneous fat , kyphosis , osteoporosis , retinal photoreceptor loss , liver aging , extensive neurodegeneration , and a short lifespan of 4–5 months ., We show that deletion of XPG specifically in the liver reproduces the progeroid features in the liver , yet abolishes the effect on growth or lifespan ., In addition , specific XPG deletion in neurons and glia of the forebrain creates a progressive neurodegenerative phenotype that shows many characteristics of human XPG deficiency ., Our findings therefore exclude that both the liver as well as the neurological phenotype are a secondary consequence of derailment in other cell types , organs or tissues ( e . g . vascular abnormalities ) and support a cell-autonomous origin caused by the DNA repair defect itself ., In addition they allow the dissection of the complex aging process in tissue- and cell-type-specific components ., Moreover , our data highlight the critical importance of genetic background in mouse aging studies , establish the Xpg−/− mouse as a valid model for the severe form of human XPG patients and segmental accelerated aging , and strengthen the link between DNA damage and aging .

Accumulation of DNA damage has been implicated in aging ., Many premature aging syndromes are due to defective DNA repair systems ., The endonuclease XPG is involved in repair of helix-distorting DNA lesions , and XPG defects cause the cancer-prone condition xeroderma pigmentosum ( XP ) alone or combined with the severe neurodevelopmental progeroid disorder Cockayne syndrome ( CS ) ., Here , we present a novel ( conditional ) Xpg−/− mouse model which -in a C57BL6/FVB F1 hybrid background- displays many progressive progeroid features , including early cessation of growth , cachexia , kyphosis , osteoporosis , neurodegeneration , liver aging , retinal degeneration , and reduced lifespan ., In a constitutive mutant with a complex phenotype it is difficult to dissect cause and consequence ., We have therefore generated liver- and forebrain-specific Xpg mutants and demonstrate that they exhibit progressive anisokaryosis and neurodegeneration , respectively , indicating that a cell-intrinsic repair defect in neurons can account for neuronal degeneration ., These findings strengthen the link between DNA damage and the complex process of aging .

neuroscience, animal models, physiological processes, developmental biology, model organisms, organism development, dna, research and analysis methods, mouse models, aging, biochemistry, cell biology, nucleic acids, physiology, genetics, biology and life sciences, dna repair, molecular cell biology, genetics of disease

journal.pbio.1002386

Social Evolution Selects for Redundancy in Bacterial Quorum Sensing

Quorum sensing is a mechanism of bacterial cell—cell communication that relies on the production , release , and group-wide detection of extracellular signal molecules called autoinducers ., Quorum sensing enables populations of bacteria to coordinate changes in gene expression 1 , 2 ., Bacteria often use quorum sensing to orchestrate the release of public goods ( e . g . , enzymes or surfactants ) whose functions benefit the entire community 2 , and to direct other cooperative behaviors such as transitions to more efficient modes of growth 3 ., The cooperative nature of quorum sensing is susceptible to exploitation by mutant genotypes that do not contribute to cooperation but benefit from it 2 , 4–6 ., Despite their immediate advantage over the wild-type , exploiting “cheater” genotypes will be eliminated in structured populations due to their negative effect on the average fitness of the community 5 , 7–11 ., In bacteria , population structure can naturally arise in biofilms , where bacteria can grow without significant mixing 10 , or during the formation of growth bottlenecks upon invasion into a new environment 11–14 ., Many bacterial species employ multiple quorum-sensing systems that impinge on the activity of a shared transcriptional regulator ., Each of the quorum-sensing systems encodes a specific receptor and autoinducer production gene with no or limited crosstalk 15 ., In several species such as B . subtilis 16 , V . harveyi 17 , and its pathogenic relative , V . cholerae 18 , 19 , the quorum-sensing systems are arranged in a parallel , seemingly redundant , architecture ., That is , all the quorum-sensing autoinducer receptors funnel information into the same signal transduction pathway ., It is unclear what the adaptive benefit is of harboring multiple , rather than a single , quorum-sensing autoinducer—receptor pair when the pairs function in parallel ., Here , we combine modeling and experiments in B . subtilis and V . harveyi to show that a strain that has accumulated an additional quorum-sensing system reduces its cooperative investment in the presence of its ancestor , but resumes full cooperation in a clonal population ., We show that this facultative cheating strategy requires a specific system integration design criterion; the novel receptor must have a dominant repressive effect on the ancestral quorum-sensing response in the absence of the novel autoinducer ., We show that , additionally , this particular network design often leads to synergistic activation of the quorum-sensing response by the different autoinducers ., We hypothesized that social interactions between different genotypes may contribute to the adaptive role of redundant quorum-sensing networks ., This hypothesis can be approached by comparing the social behavior of a wild-type species possessing multiple quorum-sensing systems to the behavior of mutant strains harboring varying numbers of quorum-sensing systems ., To explore this idea , we first examined the ComA-directed quorum-sensing network of B . subtilis 16 , 20–22 ., This network is composed of a single ComP-ComX system and multiple paralogous Rap-Phr systems , each encoding its own specific autoinducer ., The ComP receptor phosphorylates ComA when ComP is bound to the ComX autoinducer , while Rap receptors repress ComA only when their corresponding Phr ligands are not bound ( Fig 1A ) ., All the autoinducers therefore positively control ComA activity , but through different regulatory interactions ., The Com and Rap systems also differ with respect to their population genetics patterns ., The ComP-ComX system exhibits significant genetic variability within the population and different alleles form distinct orthogonal signaling pherotypes 20 ., These different pherotypes often display partial cross-inhibition , such that an autoinducer from one strain inhibits the response of another strain to its cognate autoinducer 23 ., Nonetheless , only a single ComP-ComX system is encoded in each B . subtilis isolate ., By contrast , all B . subtilis strains encode multiple Rap-Phr systems ., The exact number varies between strains , likely due to the association of some Rap-Phr systems with mobile elements 24 ., It was previously shown that deletion of any Rap-Phr system has only a small effect on ComA activity , while deletion of the phr genes or overexpression of the rap genes led to repression of quorum sensing 16 , 25 ., It is therefore unclear why so many quorum-sensing systems regulate ComA , and specifically , why Rap-Phr paralogs proliferate in the genome , while ComP-ComX is unique ., To address this question , we constructed strains in which we added to the wild-type strain a novel ComP-ComX system ( ExtraCom strain , comQXPRO-H-1+ 20 ) or we either added ( ExtraRap strain , rapPphrP+ 26 ) or we deleted ( MinusRap strain , ΔrapFphrF ) a Rap-Phr system ( Fig 1B , see methods and S1 File for strain construction details ) ., The autoinducing signals produced by the introduced ExtraRap and ExtraCom systems differed from those made by the paralogs present in the parent strain with no cross-activation 20 , 26 ( S1 Fig ) ., We note , however , that the comXRO-H-1 autoinducer cross-inhibits the endogenous ComP168 receptor 23 ., We examined the behavior of these strains under surface-swarming motility conditions , which strictly require the production and release of a ComA-dependent surfactant called surfactin 27 , 28 ( Fig 1C , see methods for details on swarm motility protocol ) ., Unlike the ΔcomA mutant , the above quorum-sensing variant strains exhibited robust swarming , reaching a similar cell yield as the wild-type after 48 h ( Fig 1C , S1A Fig , p = 1×10−5 , F ( 4 , 12 ) = 8 . 1 , n = 16; two-way ANOVA for difference between genotypes when including ΔcomA , p = 0 . 26 , F ( 3 , 9 ) = 8 . 1 , n = 12 without ΔcomA ) ., Altering the number of quorum-sensing systems therefore does not significantly affect the fitness of the bacteria in clonal populations ., Surfactin may function as a costly public good during swarming , allowing “cheater” strains to exploit the wild-type in coculture ., In agreement with this possibility , we found that the ΔcomA mutant strain regained its ability to swarm when cocultured with the wild-type , and in so doing , dramatically increased its relative frequency in the population ( Fig 1D and S1B Fig; p = 10−4 , two sample t test , n = 42 , see methods for details of the competition experiments and the wild-type competition against itself in Fig 1C that was carried out as a control ) ., When we performed similar coculture experiments between the wild-type and the different quorum-sensing variants , we found that the strain carrying an additional Rap-Phr system was strongly selected for over a strain lacking it ., In contrast , the ExtraCom strain was out-competed by the wild-type ., Moreover , the fitness advantage of the ExtraRap strain over the wild-type was similar to that of the “cheater” ΔcomA mutant at low frequency ( p = 0 . 21 , t ( 4 , 32 ) = 0 . 8 , linear regression comparison of the intercepts at zero frequency ) and approached neutrality as its frequency increased ( Fig 1D , p = 0 . 13 , t ( 2 , 16 ) = 1 . 06 , linear regression comparison of the intercepts at zero at a frequency of one ) ., Similar results were obtained for wild-type exploitation of the MinusRap strain ( S1C Fig ) ., In contrast to the selection of the ExtraRap strain , the ExtraCom strain remained close to neutral with respect to the wild-type at low frequency , but its competitive disadvantage increased with increasing frequency ( Fig 1D , p = 10−8 , t ( 2 , 16 ) = 11 , linear regression of slope ) ., Our results are therefore in agreement with the observed population genetics data for the two systems—selection for genomic proliferation of Rap-Phr systems and against proliferation of the ComP-ComX system ., To gain further insight into our results , we mathematically modeled cellular growth and quorum-sensing signaling dynamics during swarming ( Fig 2 ) ., In the model , we assume a simplified ancestral strain encoding a single ComP-ComX and a single Rap-Phr system ., We explored the growth and social dynamics of this ancestor and its corresponding ExtraRap- and ExtraCom-derived strains during swarming ( see methods and S1 File for description of the model and its assumptions ) ., Strikingly , the model was able to capture qualitatively the experimental results we obtained above both in clonal and social conditions ( Fig 2A and S2 Fig , compare with Fig 1B and 1C ) ., The findings underpin how selection depends on the particular circuit design of the two quorum-sensing systems ., The model also provides simple explanations for the frequency dependence of the ExtraRap system and the difference in selection for and against the ExtraRap and ExtraCom strains , respectively ., When a derived “Extra” strain is at low frequency , the concentration of the novel autoinducer it produces is very low compared to those of the ancestral autoinducers , which are produced by all the members in the population ( Fig 2A , left insets ) ., In this scenario , the level of quorum-sensing response of the “Extra” strain depends on the activity of the unliganded form of the novel receptor ., In the ExtraCom system , the novel ComP receptor is inactive in the absence of its cognate autoinducer ., The ancestral network will therefore not be affected by the presence of the novel ComP system , which leads to equal quorum-sensing activation of the ancestral and ExtraCom strains ( Fig 2A and 2C ) ., In contrast , in the ExtraRap strain , the autoinducer-free novel Rap receptor represses ComA ., Repression is dominant and overpowers activation of ComA by the shared ancestral quorum-sensing system ( Fig 2A ) ., Activated ComA levels will therefore be lower in the ExtraRap strain than in the ancestral strain ( Fig 2B ) , leading to selection of the ExtraRap strain due to exploitation of the ancestral strain ., As the frequency of the derived “Extra” strain increases , so does the concentration of its corresponding novel autoinducer ( Fig 2A , right insets ) ., In the case of the ExtraRap strain , accumulation of the novel autoinducer leads to partial de-repression of ComA to a level that approaches that of the ancestor ( Fig 2B ) , and this condition occurs as the ExtraRap strain approaches fixation ., Therefore , the ExtraRap strain acts as a cheater at low frequency but returns to full cooperation when fixed in the population ., In the case of the ExtraCom strain , accumulation of the novel autoinducer leads to a corresponding increase in its quorum-sensing response ., In the specific experimental case we examined , the novel ComX system ( ComXRO-H-1 ) in the ExtraCom strain cross-inhibits the ancestral ComP168 receptor , leading to a strong reduction in ComA activity in the wild-type , ancestral strain ( Fig 2C ) ., The ancestral strain therefore acts as a cheater with respect to the ExtraCom strain ., Our modeling framework also allows us to explore a theoretical case in which no cross-inhibition occurs ., In this situation , the ancestral strain maintains a constant level of ComA activity ( S3 Fig ) ., The net selective effect , with or without autoinducer cross-inhibition , is against the ExtraCom system , although selection is stronger when autoinducer cross-inhibition occurs ( S3 Fig ) 9 ., Thus , while autoinducer cross-inhibition naturally exists in the B . subtilis system we are studying , this feature is not strictly required for selection against accumulation of a novel quorum-sensing system ( S3 Fig ) ., Our model also predicts how the novel autoinducer will act with respect to the ancestral autoinducer in that the model provides us with information about what type of regulatory input—output gate is established ., An additional ComP-ComX system leads to formation of an OR-like ( additive ) regulatory gate for the two ComX autoinducers with respect to their control of ComA activity ( Fig 2E ) ., Thus , a single autoinducer is sufficient to elicit a strong quorum-sensing response ., In contrast , the repressive activity of an additional Rap system leads to formation of an AND-like ( multiplicative ) gate between it and the ancestral Phr or ComX system ., Thus , the simultaneous presence of both autoinducers is required to elicit a strong response ( Fig 2D and S4A Fig ) ., Our model predicts that the different architectures of the two quorum-sensing systems lead to differential investment in cooperative behavior by the ancestral and derived strains as well as to distinct regulatory input—output gate structures ., These features result in the observed patterns of selection ., To address these predictions experimentally , we introduced a YFP transcriptional reporter for ComA activity ( PsrfA-YFP ) into the wild-type , the ExtraCom , and the MinusRap strains ., We cocultured each reporter-containing strain with a reporter-free counterpart and measured gene expression as a function of frequency ( Fig 3A and 3B and S5A Fig ) ., To minimize the effect of changes in frequency and spatial distribution , we performed these assays in minimal medium using a surfactin production-deficient mutant of the sfp gene ( sfp− 27 ) , which has reduced quorum-sensing-associated cost ., Similar results were obtained when gene expression was measured during swarming ( S5B Fig ) ., The absence of surfactin in the minimal growth medium did not significantly affect expression of the PsrfA-YFP reporter construct in the cocultured strains ( S5B Fig ) ., We found that , when the MinusRap and wild-type strains are cocultured , the MinusRap strain maintained a constant ComA activity , irrespective of its frequency in the population ( F ( 1 , 16 ) = 0 . 41 , p = 0 . 53 , n = 18 , linear regression of the slope ) ., In contrast , at low frequency , the wild-type exhibited low level ComA activity , which increased with increasing frequency of the wild-type ( F ( 1 , 15 ) = 96 , n = 17 , p < 10−7 , linear regression of the slope ) ., At high frequencies , wild-type ComA activity approached the activity level of the MinusRap strain ( Fig 3A , t test , p = 0 . 43 for interception of best-fitted lines at a frequency of 1 ) ., When the ExtraCom strain was cocultured with the wild-type , ComA activity was the same in both strains when the frequency of the ExtraCom strain was low ( t test for linear regression of lines , p = 0 . 26 for interception at frequency of zero ) ., ComA activity in the ExtraCom strain increased with increasing frequency ( Fig 3B , p = 10−11 F ( 1 , 14 ) = 405 , for a zero slope ) ., In accordance with the expected effects of crossinhibition ( Fig 2 and S3 Fig ) , the ComA activity of the wild-type strain decreased dramatically with increasing frequency of the ExtraCom strain ( Fig 3B , p = 10−6 F ( 1 , 15 ) = 54 , for a zero slope ) ., We next measured the resulting regulatory gate structure of the response of ComA to addition of multiple autoinducers ., We constructed a strain constitutively expressing the rapF and rapC receptor genes but not their respective phrF and phrC autoinducer-production genes ., We found that the ComA response was significant only if both the PhrF and PhrC autoinducers were present , showing an AND-like gate structure ( Fig 3C , S6A and S6B Fig ) ., Likewise , an AND-like response occurred for PhrF and ComX regulation of ComA ( S5 Fig ) ., In contrast , regulation of ComA by the two ComX autoinducers in the ExtraCom strain was additive as expected for an OR-like response ( Fig 3D , S6C , S6D and S6E Fig , methods ) ., The experimental results support the role of social interactions in selection for accumulation of Rap-Phr systems coupled with selection against accumulation of ComP-ComX systems in B . subtilis ., In order to generalize these results , we formulated a generic model of selection with respect to quorum-sensing-dependent public goods ( S1 File ) ., This model suggests that two design criteria are necessary and sufficient for the invasion of a strain carrying an additional quorum-sensing system into a population lacking it:, 1 ) Dominant repression: The ligand-free novel receptor should act negatively to overpower the quorum-sensing response of the ancestral system , and, 2 ) Facultative operation: The addition of the novel autoinducer should restore the quorum-sensing response to levels similar to that of the ancestor ., The combination of these two features allows the invading strain to perform facultative cheating—cheat the ancestor in coculture ( criterion #1 ) but resume cooperation when it is fixed in the population ( criterion #2 ) 9 , 29 ., In the S1 File , we show that if repression by the novel quorum-sensing system is strong , the two autoinducers will regulate the response in an AND-like manner ., We further demonstrate that formation of an AND-like gate is sufficient but not mandatory to select for acquisition of a novel quorum-sensing system ., Likewise , an OR-like gate between autoinducers is sufficient but not required to select against the acquisition of a novel quorum-sensing system ., The AND-like and OR-like gate structures provide an intuitive , albeit simplified , explanation for selection ( AND ) or counterselection ( OR ) of an evolved strain ., If both autoinducers are necessary to activate the quorum-sensing response in the evolved strain ( AND gate ) , while the ancestral strain produces and responds to only one of the autoinducers , then the evolved strain will cease to cooperate when present as a small minority together with its ancestor ., In contrast , if either autoinducer is sufficient , then the evolved strain will continue to cooperate even when it is present as a minority ., We next examined whether our results also apply to another well-studied model organism in which multiple quorum-sensing systems exist and control a common output ., The bioluminescent marine bacterium V . harveyi 15 has a quorum-sensing network composed of three parallel systems that regulate expression of the quorum-sensing master transcription factor LuxR , which controls multiple traits including bioluminescence emission ( Fig 4A ) ., A similar architecture composed of four quorum-sensing systems exists in the related pathogen , V . cholerae 19 ., While the deletion of any of the receptors does not affect the quorum-sensing response 18 , deletion of any of the autoinducer synthase genes represses LuxR-activated genes , demonstrating the dominant repressive effect of each ligand-free receptor 17 ., In addition , two of the autoinducers have been shown to act multiplicatively in their regulation of LuxR 30 and to synergistically control bioluminescence 31 ., We used the abundant quantitative data on this organism to construct a model of the expected social behavior of the wild-type and an ancestral-like strain deleted for any one of the quorum-sensing systems ( S1 File , S7A Fig ) ., The model predicts that the wild-type will reduce its cooperative investment in the presence of such an ancestral strain , which will lead to facultative cheating under appropriate conditions ., To verify the model experimentally , we constructed a putative ancestral strain deleted for the luxMN autoinducer-receptor system ., We introduced a null mutation into the lux ( luciferase ) operon in the wild type and derived ancestral strains , and by mixing Lux+ and Lux− pairs ( WT/Lux+ mixed with luxMN/Lux− and WT/Lux− mixed with luxMN/Lux+ ) , we could measure the level of quorum-sensing response per cell of the Lux+ strain in each coculture ( Fig 4B and S7B Fig , methods ) ., As expected from the model , we found that light production by the luxMN mutant strain remained almost constant irrespective of its frequency ( the small decrease is most likely due to the effect of the lux locus , S7B Fig ) ., The wild-type showed a near 100-fold reduction in bioluminescence output compared to the luxMN mutant at low frequency ( p < 10−13 , T ( 32 ) = 12 . 7 , t test on linear regression for intersect at zero frequency ) , while approaching the same level of light production as the luxMN mutant at high frequency ., Our rationale therefore also applies to the parallel quorum-sensing network of V . harveyi ., In this work , we propose that bacteria possessing multiple quorum-sensing networks that control the identical response , which are commonly found in nature , are selected through a facultative cheating process ., Facultative cheating has been described in the past as a strategy by which microorganisms exploit nonkin but return to cooperation in the presence of kin 32 ., Such behavior has been described in fruiting body-forming amoeba and bacteria 29 , 33 , but the underlying molecular processes that lead to it are unknown 34; however , links to cell—cell signaling and facultative cheating have been suggested 35–37 ., We predict that accumulation of multiple quorum-sensing systems requires a specific set of network design criteria , the functioning of which we explored in two diverse but well studied organisms ., Specifically , the introduced novel receptor must repress the quorum-sensing response in the absence of the novel autoinducer , as occurs in the B . subtilis Rap-Phr and V . harveyi Lux quorum-sensing systems ., In contrast to these systems that depend on repression , other quorum-sensing systems exist that act positively , in that the receptor functions as an activator upon autoinducer binding ., Our model and experimental results explain why accumulation of parallel positively acting systems is selected against ., Indeed , we do not know of any bacterium that possesses multiple activation-based quorum-sensing systems that function in parallel ., Rather , activation-based systems are commonly organized in a hierarchy , in which one quorum-sensing system regulates the expression of a second system ., A hierarchical network design is not fully redundant because the two quorum-sensing systems can control different genes ., Further work will be required to define the benefits and possible evolutionary routes giving rise to quorum-sensing systems that function positively and are arranged as hierarchies ., Beyond facultative cheating , other possible adaptive functions for possessing multiple quorum-sensing systems have been suggested ., These include gains in information acquired about cell density , information about the frequency of phenotypes in the vicinal population , and access to information about physical flow conditions 38–41 ., Our social selection model does not contradict those alternatives and may promote them by driving the initial fixation of the redundant network design , which can , subsequently , be further modified for other adaptive advantages ., Several processes may limit the accumulation of quorum-sensing systems ., First , each system contributes a signaling cost 5 , 42 ., Second , the facultative return to cooperation may not be complete , leading to reduced benefit during exploitation in structured populations ., Third , social selection of facultative characters is weak and can lead to variable mutation selection balance 34 ., Finally , rareness of available systems and the need to integrate them appropriately into the existing network may limit the rate of accumulation ., Further work will be required to define the importance of each of these mechanisms ., Exploitation can also occur between species; not only between variants within species ., For example , cooperative secretion of antibiotic degrading enzymes has been shown to lead to coexistence of secreting and nonsecreting genotypes , at both the species and interspecific levels 43 , 44 ., Accumulation of additional quorum-sensing systems could also be used to exploit species that produce fewer signals ., This ecological factor may contribute to the continuous selection for maintenance of multiple systems ., More generally , our results point to the roles facultative cheating and kin recognition may have in the ecology of complex microbial communities ., Routine growth was performed in Luria—Bertani ( LB ) broth: 1% tryptone ( Difco ) , 0 . 5% yeast extract ( Difco ) , 0 . 5% NaCl ., Experiments with B . subtilis were done using Spizizen minimal medium ( SMM ) : 2 g L−1 ( NH4 ) 2SO4 , 14 g L−1K2HPO4 , 6 g L−1KH2PO4 , 1 g L−1disodium citrate , 0 . 2 g L−1MgSO4∙7H2O ., This was supplemented with trace elements ( 125 mg L−1MgCl2∙6H2O , 5 . 5 mg L−1CaCl2 , 13 . 5 mg L−1FeCl2∙6H2O , 1 mg L−1MnCl2∙4H2O , 1 . 7 mg L−1ZnCl2 , 0 . 43 mg L−1CuCl2∙4H2O , 0 . 6 mg L−1CoCl2∙6H2O , 0 . 6 mg L−1 Na2MoO4∙2H2O ) ., Unless otherwise noted , 0 . 5% glucose was used as carbon source ., Petri dishes for routine procedures were solidified using 1 . 5%agar ( Difco ) ., Antibiotic concentrations: Macrolides-lincosamides-streptogramin B ( MLS; 1 μg ml−1 erythromycin , 25 μg ml−1 lincomycin ) ; Spectinomycin ( Sp , 100 μg ml−1 ) ; Tetracycline ( Tet , 10 μg ml−1 for B . subtilis ) ; Kanamycin ( Km , 5 μg ml−1 ) ; Chloramphenicol ( Cm , 15 μg ml−1 ) ; Ampicillin ( Amp , 100 μg ml−1 ) ; Carbenicillin ( Carb , 300 μg ml−1 ) ., Isopropyl β-D-thiogalactopyranoside ( IPTG , Sigma ) was added to the medium at the indicated concentration when appropriate ., Premeasurement Bacillus growth protocol: Prior to all measurements , an overnight colony from an LB agar plate was inoculated in 1 mL SMM liquid medium and grown for 7 h until an OD600 of 0 . 1–0 . 3 was reached ., The cultures were diluted by a factor of 106 and grown overnight at 37°C ., Overnight cultures were centrifuged , resuspended in PBS , and diluted to an OD600 of 0 . 01 ., We find that this long incubation in minimal medium both reduced the effects of quorum sensing prior to growth and reduced the arbitrary difference in growth between two cocultured wild-type colonies ., For coculture experiments , cells of different strains were mixed in appropriate ratios after overnight growth in SMM , based on relative optical density ( OD ) ., The exact ratios were subsequently measured using flow cytometry ., Synthetic pentapeptides PhrF ( NH2 QRGMI COOH ) and PhrC ( NH2 ERGMT COOH ) were purchased from GL Biochem ( Shanghai , China ) at >98% purity ., 10 mM aliquots were prepared by resuspension of the lyophilized peptides in H2O and stored at −20°C ., Samples were analyzed on a Gallios flow cytometer ( Beckman-Coulter ) , equipped with four lasers ( 405 nm , 488 nm colinear with 561 nm , 638 nm ) ., The emission filters used were: BFP– 450/50 , YFP/GFP– 525/40 , mCherry– 620/30 ., Events were discriminated using the forward-scatter parameter ., For each run , discrimination enabled a single , well-defined population to appear in the forward-scatter ( FS ) by side-scatter plot ., Gating on the fluorescent populations and inspection of the nondiscriminated forward by side-scatter plot indicated that over 99 . 9% of the fluorescent cells are present in the discriminated population ., In all analyzed samples , only single cells were considered by gating on correlated time-of-flight and FS events ., Gating of the different fluorescent populations was performed by inspection of the log-log FLx by FLy plots ( where x & y represent the appropriate filter number for each fluorescent marker ) , where two distinct populations were clearly visible , resulting in type-I and type-II errors of less than 0 . 05% ., For each run , at least 100 , 000 cells were analyzed and the total events analyzed such that the minority population was never below 1 , 000 events ., Cells were grown as described in the premeasurement growth protocol ., Five microliters of diluted cultures were placed at the centers of 0 . 7% agar plates containing 25 mL of SMM medium supplemented with trace elements and 0 . 05% glucose ., The plates were prepared 1 h prior to inoculation , allowed to solidify in room temperature , and dried for 5 min in a laminar flow chamber ., The plates were incubated at 30°C for up to 72 h ., The swarms were collected after suspension in 5 ml of PBS , the OD was measured , and the final ratios or the gene expression was determined using flow cytometry as described above ., For spatial analysis of swarming , samples were taken from the centers of the plates , in addition to several samples 1 cm and 2 cm from the center ., We find that a glucose concentration of 0 . 05% compared to 0 . 5% reduced the residual swarming of the comA mutant , increasing the difference in growth between the mutant and the wild-type , likely because residual production of surfactin by the mutant colony was reduced ., In all experiments , YFP level was determined from the median level of the unimodal distribution of YFP expressing cells using flow cytometry ., YFP level was normalized by the autofluorescence of the wild-type ., For coculture experiments ( Fig 3A and 3B ) , samples were taken at several time points , and the OD600 and YFP levels were measured by flow cytometry ., The expected YFP level at OD600 = 1 was calculated by interpolation ., V . harveyi strains were grown at 30°C in Luria-marine ( LM ) medium with aeration ., Following overnight growth , samples were diluted to OD600 = 0 . 005 with varying ratios of dark and bright strains ., Following 6 . 5 h of growth , bioluminescence was measured on a Tri-Carb 2810 TR ( Perkin Elmer ) scintillation counter ., Dilutions of the cultures were made and plated on LM agar plates ., Plates were incubated at 30°C overnight to allow colony formation ., Images of the plates were taken using an ImageQuant LAS system that detects both bioluminescence and total colony forming units ( CFUs ) ., Colonies were counted using the ImageQuant TL and ImageJ programs ., Values shown are calculated as ( total bioluminescence ) / ( # bioluminescent CFUs ) ., The values were normalized to the bioluminescence per cell of the bright strain ., All strains are detailed in S1 Table , while respective primers are provided in S2 Table ., All of the mutations and constructs were transferred to PY79 by transformation 45 ., Integration of amyE integration plasmids into the zjd89::amyEΩ Cm Km 46 was done as previously described 26 ., Deletion of rapF-phrF , rapC-phrC , comA , and comQXP from the PY79 chromosome and their replacement with the MLS resistance cassette was performed through the long flanking homology PCR method 47 using the primers rapF-P1-P4 , rapC-P1-P4 , comA-P1-P4 , and comQXP-P1-P4 , respectively ( S2 Table ) ., The rapFphrF::Cm deletion was generated using the antibiotic switching vector ece76 ., rapFphrF::Cm was next used as a template to generate rapFphrF::Tet using the antibiotic switching vector ece75 ( S1 Table ) ., To generate inducible zjd89:: ( Phyperspank-rapF ) and amyE:: ( Phyperspank-rapC ) constructs , a PCR product containing the relevant open reading frame was amplified using the primer pairs hsRapF-F/hsRapF-R andhsRapC-F/hsRapC-R ., The PCR products were digested with the appropriate enzymes ( S2 Table ) and ligated downstream of the hyperspank promoter of the pDR111 vector containing Spec resistance 48 ., Construction of sacA:: ( comQXPRO-H-1 Cm ) was performed by PCR amplification of comQXP from strain B . mojavensis RO-H-1 using the comQXP-ROH1-F and comQXP-ROH1-R primer pair ., The PCR product was digested with restriction enzymes BamHI and EcoRI and ligated to the ece174 plasmid ( S1 Table ) ., The resulting vector was integrated into the sacA site on the chromosome using Cm resistance for selection ., Construction of sacA:: ( Psrf-3xyfp Cm ) was performed by PCR amplification of Psrf-3xyfp using AEC945 as a template and the Psrf-sacA-F/Psrf-sacA-R primer pair ., The PCR fragment was digested with the appropriate enzymes ( S2 Table ) and ligated to the ece174 plasmid ., The resulting vector was integrated into the sacA site on the chromosome using Cm resistance for selection ., The swrA+ mutation allele is a spontaneous revertant that was selected by plating swrA−sfp+ cells on 0 . 7% LB agar plates and selecting motile variants , as was done previously 28 ., The reconstituted swrA+ allele was verified by sequencing ., The sfp+ allele was amplified from the undomesticated strain B . subtil

Introduction, Results, Discussion, Materials and Methods

Quorum sensing is a process of chemical communication that bacteria use to monitor cell density and coordinate cooperative behaviors ., Quorum sensing relies on extracellular signal molecules and cognate receptor pairs ., While a single quorum-sensing system is sufficient to probe cell density , bacteria frequently use multiple quorum-sensing systems to regulate the same cooperative behaviors ., The potential benefits of these redundant network structures are not clear ., Here , we combine modeling and experimental analyses of the Bacillus subtilis and Vibrio harveyi quorum-sensing networks to show that accumulation of multiple quorum-sensing systems may be driven by a facultative cheating mechanism ., We demonstrate that a strain that has acquired an additional quorum-sensing system can exploit its ancestor that possesses one fewer system , but nonetheless , resume full cooperation with its kin when it is fixed in the population ., We identify the molecular network design criteria required for this advantage ., Our results suggest that increased complexity in bacterial social signaling circuits can evolve without providing an adaptive advantage in a clonal population .

Quorum sensing is a mechanism through which bacteria communicate by producing , releasing , and detecting signal molecules encoding information about cell population density ., Quorum sensing allows bacteria to synchronize their behaviors and act as collectives ., Often , quorum sensing controls cooperative behaviors that benefit the entire community , such as the production and secretion of costly metabolites ., Some bacteria release multiple signal molecules which , once detected , funnel information into the same cellular response ., Thus , the benefit of using multiple rather than a single signal is mysterious since the signals seem redundant ., Here , we combine modeling and experiments to show that the evolutionary accumulation of multiple quorum-sensing systems can be attributed to social exploitation and kin recognition ., When in low abundance , a strain that has acquired an additional quorum-sensing system can avoid cooperating and can exploit its ancestor strain , which contains one less quorum-sensing system ., The cheater containing the additional system returns to a cooperative behavior when it is abundant ., We also identify the molecular mechanisms necessary for the acquisition of an additional signaling system ., Our work demonstrates that increased complexity in bacterial social signaling circuits can evolve without providing an adaptive advantage in a clonal population .

medicine and health sciences, pathology and laboratory medicine, engineering and technology, pathogens, bacillus, microbiology, electromagnetic radiation, regression analysis, prokaryotic models, luminescent proteins, model organisms, mathematics, statistics (mathematics), yellow fluorescent protein, optical equipment, molecular biology techniques, bacteria, bacterial pathogens, research and analysis methods, microbial physiology, artificial gene amplification and extension, proteins, medical microbiology, mathematical and statistical techniques, gene expression, microbial pathogens, statistical methods, luminescence, molecular biology, physics, biochemistry, bioluminescence, polymerase chain reaction, equipment, linear regression analysis, bacillus subtilis, genetics, biology and life sciences, physical sciences, quorum sensing, organisms

The accumulation of multiple, seemingly redundant, bacterial quorum-sensing systems is promoted by facultative cheating behavior; the strain with multiple systems cheats its single quorum-sensing system ancestor as a minority but returns to cooperation when in the majority.

journal.ppat.1002225

Transmission Characteristics of the 2009 H1N1 Influenza Pandemic: Comparison of 8 Southern Hemisphere Countries

In late April 2009 , the first cases of the novel swine-derived H1N1pdm influenza A virus were detected in Mexico and the United States , prompting the World Health Organization ( WHO ) to raise the level of influenza pandemic alert to phase 5 1 ., By the end of 2009 , the H1N1pdm virus had spread to more than 208 countries , resulting in hundreds of thousands of cases and at least 18000 deaths 2 , 3 ., Following WHO and Centers for Disease Control and Prevention ( CDC ) recommendations , generalized media coverage and international mobilization , many countries initiated mitigation measures and enhanced surveillance of H1N1pdm virus infection in humans , providing an abundance of epidemiological data for this epidemic 3 , 4 ., As a result the H1N1pdm is one of the most documented pandemics with enhanced surveillance established in many regions of the globe , with the exception of Africa 5 , 6 ., The H1N1pdm virus was introduced into most northern and southern hemisphere countries during the spring and summer of 2009 ., This period is outside the typical influenza season in temperate countries in the Northern hemisphere , but in the typical winter season for influenza transmission for countries from temperate regions of the Southern Hemisphere ., In most Southern hemisphere temperate countries , a full epidemic of H1N1pdm influenza was observed and the pandemic strain quickly became the predominant circulating influenza virus , replacing seasonal strains in many countries 7 ., Influenza transmission in a given community may depend on several factors: e . g . climatic characteristics as temperature and humidity 4 , 8 , 9 , virus intrinsic transmissibility , acquired immunity in affected populations , contact patterns in the community , collective and individual measures limiting virus spread 10 ., The 2009 H1N1 pandemic was a unique opportunity for comparing the spread of a novel influenza virus in a community setting in different countries with different population structures and contact patterns ., In this context , countries from temperate regions of the Southern Hemisphere , which present different demographic patterns and experienced the virus during their usual winter season , present an opportunity to evaluate the impact of these characteristics on transmission ., Here we use mathematical modelling to assess the transmission characteristics of H1N1pdm virus using epidemiological data from Southern hemisphere countries in temperate regions ., We address the question of the origins of the observed differences between countries by investigating the role of seasonality ( with latitude used as a proxy ) , population density and population demography ( with proportion of children used as a proxy ) ., We then explore more precisely the contributions of demography in the spread of the disease by fitting different transmission models to the set of countries ., The epidemiological data analysed here were weekly case incidence of laboratory-confirmed H1N1pdm or influenza-like-illness ( ILI ) and the distribution of cumulative incidence by age-group over the study period for seven Southern hemisphere countries and one state ( Argentina , Australia -whole country and Victoria- , Bolivia , Brazil , Chile , New Zealand and South Africa ) ., The data were extracted from websites or public reports issued by the countries surveillance systems ., Country datasets and corresponding sources are described and listed in Table 1 . Neither daily case incidence nor age-stratified weekly case incidence data were available ., Depending on the country , weekly incidence data were either laboratory confirmed H1N1pdm cases ( H1N1CC ) ( Argentina , Australia , Bolivia , Brazil , Chile , New Zealand , South Africa ) or influenza-like-illness ( ILI ) ( Australia , Chile , New Zealand , Victoria ) ., All available datasets were used in the analysis , even when multiple datasets were available for a given country ., Cumulative distributions of cases by age were extracted from the same data sources ( Table 1 ) ., These were generally of H1N1pdm confirmed cases , except for Australia and New Zealand , where we used the age distribution of ILI cases ., Due to differences between countries in the age stratification of available H1N1pdm data , country-associated age-groups were broken down into the following: Argentina ( 0–5 , 5–19 , 20–49 , 50–59 , ≥60 years old ) ; Australia ( 0–5 , 5–19 , 20–49 , 50–64 , ≥65 years old ) ; Victoria ( 0–5 , 5–19 , 20–49 , 50–64 , ≥65 years old ) ; Bolivia ( 0–5 , 5–19 , 20–44 , 45–49 , ≥50 years old ) ; Brazil ( 0–5 , 5–14 , 15–49 , 50–59 , ≥60 years old ) ; Chile ( 0–5 , 5–14 , 15–54 , 55–64 , ≥65 years old ) ; New Zealand ( 0–5 , 5–19 , 20–49 , 50–59 , ≥60 years old ) ; South Africa ( 0–5 , 5–19 , 20–49 , 50–64 , ≥65 years old ) ., Demographic information was extracted from census data of the national statistics institute of the corresponding countries ( data are presented in details and electronic URL for sources are listed in Table S1 in Text S1 ) ., A deterministic model was constructed to describe the spread of the virus in a population structured by age-groups ., Model parameters and their values are summarized in Table 2 . Five age-groups were defined in the model ( NA\u200a=\u200a5 ) : young children , children , young adults , adults , older adults ( with breakdowns as defined above ) ., Population structure was described by the vector Ni , with Ni representing the number of individuals in age-group, i . Total population size was noted NP ., Individuals in the population were assumed to be either susceptible , infected or recovered ( classical SIR model ) ., Each age group of the population was initialized with y0 ( a fitted parameter ) infections at the beginning of the simulation ( ten weeks before the first week of observation ) ., The model incorporated heterogeneous mixing by age , with a variety of mixing patterns being explored ( more details are presented below and in section 1 of Text S1 ) ., The parameter β defined the transmission coefficient ., Susceptibility to infection was hypothesized to vary with age and given by the vector ρi ., To avoid confounding with the parameter β , the susceptibility of young children was fixed at 1 ( ρ1\u200a=\u200a1 ) and the susceptibility of other groups was estimated ., Therefore , for a given individual of age i , the risk of infection per contact with an infected individual is given by βρi ., The generation time was assumed to be Gamma distributed 11 , 12 with mean µ\u200a=\u200a2 . 6 days and standard deviation σ\u200a=\u200a1 . 3 days 13 ., Although some previous studies have suggested that children infected with influenza may be more infectious than adults , there was no evidence of any significant age-specific transmission risk of H1N1pdm 13 , 14 ., Consequently , no age-specific infectiousness was considered in the model ., We also assumed that only a proportion of infected individuals were effectively reported to the surveillance system , represented in the model by a reporting rate preport ( underreporting included here both unreported symptomatic cases and asymptomatic cases ) ., No incubation period or reporting delay was considered , since so long as the generation time distribution is captured accurately , ignoring these factors does not affect transmission parameter estimates ., We finally assumed that ILI surveillance data included a constant incidence of non-influenza related cases ( baseline ) , defined as BL ., Technical details of the model can be found in section 1 of Text S1 ., The basic reproduction number of the virus spread , R0 , was computed as the largest eigenvalue of the next generation matrix K of the model ., The next generation matrix defines the next generation of new infected from a previous generation of infected 15 with element Ki , j representing the expected number of new cases from age-group i generated by one infected individual of age-group, j . K was defined as:with β being the contact rate , ρ the susceptibilities and M the mixing matrix among age-groups , defined as the proportion of contacts an infected individual in age class j makes with individuals in age class, i . The infection attack rate pI was defined as the proportion of individuals in the population having been infected after the epidemic ends ., Parameters of the dynamic model were estimated in a likelihood-based Bayesian setting using Markov Chain Monte Carlo ( MCMC ) methods with a Metropolis Hastings sampler to explore the space of parameters ., The posterior median and 95% credible interval were reported for each parameter ., See Text S1 for more details ., Initially , parameters were estimated for each country independently ( country-specific fits ) ., In order to better understand the role of demography on H1N1pdm spread , estimation was also run for all the countries together ( global fits ) ., We defined three model variants which differed in the assumption made on mixing patterns between age-groups ., In the first two models , assortative mixing between age groups was assumed 16 ., For a given age group , individuals had a proportion of their contacts θ occurring in their own age-group , with the remaining 1-θ fraction of contacts occurring at random in the whole population ., Model variant one ( M1 ) involved a simple assortative mixing in which individuals mixed preferentially in their own age-group ( with fixed probability θ\u200a=\u200a0 . 25 ) and randomly with all age-groups with probability ( 1-θ ) ., Although higher values for assortative parameter were proposed in previous studies 16 , θ\u200a=\u200a0 . 25 was chosen as it was consistent with mixing patterns measured in the UK via diary studies 17 ., Model variant two ( M2 ) involved a more elaborate description of mixing ., Three different assortativity parameters were defined: θ1\u200a=\u200a0 . 15 for young children; θ2\u200a=\u200a0 . 4 for older children; and θ3\u200a=\u200a0 . 14 for adults ., The numerical values were estimated by fitting the mixing matrix to the mixing patterns measured in the UK 17 ., For M1 and M2 , the contact rate parameter ( β ) was assumed to be common to all age-groups ., Given that contact rates vary among age-groups 17 , this means that the estimates of age-dependent susceptibility obtained for these model variants also implicitly incorporate variation in contact rates as well as actual variation in susceptibility arising from pre-existing immunity ., Model variant three ( M3 ) differed from M1 and M2 as it used an empirical contact matrix ., The matrix was derived from the POLYMOD study data published for casual contacts in United Kingdom 17 ., In order to derive appropriate matrices for each of the studied countries , two assumptions were made ., First , we assumed that in a country in which a given age-group is more prevalent than in the UK , any individual will have a higher proportion of his contacts appearing in this age-group than individuals from the same age-group in the UK ., Second , we assumed that contact rates varied between age-groups but were constant across countries ( see supplementary material ) ., Model parameters and their values ( if these were not fitted ) are listed in table 2 . Firstly , we fitted model variant M1 to weekly case incidence data and to the cumulative age distribution of cases for each country independently , using a negative binomial likelihood with fitted variance parameter ( to allow for over-dispersion in the case data ) ., For each country , nine parameters were estimated: reporting rate ( preport ) , four age-related susceptibilities ( ρi ) i\u200a=\u200a2 . . 5 , dispersion parameter for the negative binomial likelihood , baseline for ILI incidence in the sample population ( BL ) , initial number of cases at the beginning of the simulation ( y0 ) and reproduction number ( R0 ) ., Secondly , to assess the extent to which a single model could explain the patterns seen in different countries epidemics , we fitted model variants M1 to M3 to all the countries simultaneously , keeping most parameters common to all countries ., For these global fits , susceptibilities by age and contact rate were assumed to be common to all the locations ( five global parameters ) whereas reporting rate ( preport ) , ILI incidence baseline ( BL ) , and the initial number of cases ( y0 ) were fitted on a country-specific basis ( four country-specific parameters ) ., Further details of the models and fitting procedures are given in the supplementary material ., MCMC methods were used to obtain parameter estimates ., For the country-specific fits , MCMC samples of 3×106 were generated for each country with the first 100000 iterations discarded to allow the chain to converge ., For the global fits equilibration of the MCMC chains was slower , so we generated samples of 6×106 and discarded the first 2×106 of these ., In order to assess which factors could influence the spread of the virus in the different countries , the R0 estimates were regressed on countries demographic age-distribution , latitude of the capital city ( except for South Africa where the biggest city was considered ) and densities of populations ( see supplementary material ) ., This analysis was conducted for two different set of R0 estimates: the R0 values estimated from the exponential growth of confirmed cases in the early weeks of the epidemic in each country , using the renewal equation 11 , 12 ( supplementary material ) and the median posterior estimates from the country-specific fits ., H1N1 confirmed cases were used for those countries where such data was available and ILI data was used for the one area ( Victoria ) where such data were not available ., With the exception of South Africa , the H1N1pdm epidemic started at the end of May ( epidemiological weeks EW 20–22 ) and finished by the end of September ( around EW 40 ) ., South Africa experienced a first wave of seasonal H3N2 influenza followed by H1N1pdm influenza peaking in early August 2009 6 ( Table 1 ) ( Figures 1 and 2 ) ., Cumulative age-specific incidence is summarized in Table S1 of Text S1 , as well as demographic data and sources ., Estimated empirical R0-values derived from the early exponential growth rate of the epidemic were positively correlated with the proportion of children in the population ( p\u200a=\u200a0 . 004 ) as illustrated in figure 3a ., No significant association was found with latitude and density ( supplementary material ) ., Estimates of R0 , attack rate and reporting rate are summarized in Table 3 ., For each country and dataset , Figure 1 compares the fits of the model ( grey lines ) with the H1N1pdm incidence data ., The match to the age distribution of cases is shown in Figure 2 , and estimates of R0 for the 8 countries are plotted in Figure 3B ., Estimated posterior median values of R0 ranged from 1 . 2 and 1 . 8 , with the highest values ( 1 . 5 and 1 . 8 respectively ) being obtained from for Argentina and Chile ( though for Chile , only the ILI data gave a high estimate ) ., We found estimated age-related susceptibilities to vary markedly by country ., With the exception of Bolivia and Brazil , a consistent pattern of decreasing susceptibility with age and higher susceptibility for children under 20 was found ( Figure 4 ) ., We obtained estimated posterior median infection attack rates of between 20% and 50% of the population ( Table 3 ) ., These values also varied markedly from one country to another: from 20% for Australia to 40% for Argentina and Brazil ., Common and country specific parameter estimates from the fits of the global model are summarized for model variants M1-M3 in Table 4 , while fit quality to the incidence time series is illustrated in Figures 1 and 2 ., Overall , the global fits reproduce temporal and age trends in the surveillance data well , albeit not as precisely as the fits of the country-specific model ( see section 6 of Text S1 for evaluation of model fitting ) ., Peak incidences were slightly underestimated for Argentina , Chile-ILI and New Zealand-H1N1CC and overestimated for Australia-ILI , Victoria , Chile-H1N1CC and South Africa ., Likelihood comparison did not allow one of the 3 model variants examined to be identified as superior ( section 6 of Text S1 ) ., The global fits well reproduced the age distribution of cases for Argentina , Australia , Victoria and New Zealand , although the contribution of adult cases were underestimated for Bolivia and Brazil , and overestimated for South Africa and Chile ( Figure 2 ) ., Resulting R0 estimates were similar for the three model variants , with still significant ( albeit much reduced compared with the country-specific model ) variation between countries: the highest values were obtained for South Africa and Bolivia and the lowest ones for New Zealand , Australia and Victoria ( Figure 3B ) ., Lastly , age-dependent susceptibilities to H1N1pdm were still found to decrease with age ( Figure 4B ) ., This effect was higher in model M1 and M2 suggesting that children had both higher susceptibility to the virus and higher numbers of contacts ., Estimates from model M3 also suggested that resulting differences in relative susceptibilities among adult age-groups might largely be due to variation in contacts rates between these age-groups ., Only two country-specific parameters were fitted for the global fits: the initial number of cases ( y0 ) and the reporting rate ( preport ) ., As y0 , and preport mainly influence epidemic timing and the scaling required to match surveillance incidence data , the variation in R0 seen between countries and the qualitatively good fits obtained support the idea that demographic differences between countries may have had a substantial impact on H1N1pdm transmission ., Our results suggest transmission of H1N1pdm in 2009 varied significantly between the eight countries/states included in our analysis ., Differences were found in transmissibility ( R0 median estimates ranged between 1 . 2 and 1 . 8 ) and in the size of the epidemic ( estimated median infection attack rates ranging 20–50% ) ., Estimates of R0 are relatively low compared with previous estimates from past pandemics , for which values in the range 1 . 7–2 . 2 have been more typical 18–24 , though it should be noted that some of the higher values of R0 obtained for previous pandemics assumed a longer mean generation time than we do here ., Our estimates are comparable to typical flu seasons ( R0∼1 . 3 ) 25 and consistent with other studies for H1N1pdm in 2009 obtained from other countries 26–30 ., Our results further reinforce existing evidence that children ( <20 years old ) were substantially more susceptible to infection with H1N1pdm than adults 31–33 , with adults having 30–80% the susceptibility of children , depending on the model variant examined ., The country-specific fits led to differences in susceptibility estimates among countries , maybe indicating that some over-specification exists in the country-specific model ., However , this might also suggest that levels of prior existing immunity differ among the studied populations , which has been documented in some countries 31 , 34 , 35 , playing a role in the variation in patterns of H1N1pdm spread observed ., If real , such differences in pre-existing population immunity may have contributed to the unexplained variance of the global fits relative to the country-specific fits ., It should be noted that models M1 and M2 assumed simple assortative mixing by age with no age-dependent variation in contact rates , so that estimates of age-dependent susceptibility may be confounded with variation in contact rates with age ., Model M3 used data from a diary survey of contact patterns 17 and thus incorporated higher contact rates in children , and the resulting estimated differences in susceptibility between adults and children were therefore lower for that model variant ., In addition , in a context of high media coverage and public concern , it is possible that cases in children might have been more likely to lead to health-care seeking behaviour , affecting estimates ., Nevertheless , our finding that susceptibility decreased with age is consistent with recent serological study results which demonstrated a significant proportion of immune adults prior to the start of the 2009 H1N1 epidemics 31 , 34 , 35 ., Age-dependent susceptibility might arise from the effect of immune system maturation or cross-reactive immunity due to prior infections with other ( non H1N1pdm ) influenza subtypes/strains ., In a completely naive population , the reproduction number would therefore be expected to be substantially larger ., The lack of serological data during the pandemic prevented explicit incorporation of pre-existing immunity in the model 36 , though age-dependent susceptibility implicitly represents its effects ., Sensitivity analyses in which we assumed pre-existing immunity at the beginning of the pandemic suggested including immunity would substantially affect our estimates of R0 ( given the estimates provided here are implicitly in the presence of substantial pre-existing immunity ) , but also of attack rate ., Although H1N1pdm was a new virus , our results further reinforce the evidence base that there was substantial pre-existing partial cross-immunity to the virus prior to the 2009 epidemic , particularly in adults ., Cross-immunity , an important feature of seasonal influenza epidemiology , was not expected to play such a key role in a pandemic situation ., Clearly the experience of H1N1 in 2009 has highlighted the need for more research – both experimental and theoretical - on heterosubtypic immunity ( and perhaps non-HA mediated immunity ) ., Pre-existing immunity impeded the estimation of the classic basic reproduction number ( R0 ) from the data examined here ., Our R0 estimates are really estimates for R0 , the reproduction number at the beginning of the epidemic ( at time 0 ) , rather than for the reproduction number in the absence of prior immunity ., However , for ease of notation ( and because one might argue that transmission may never occur in a truly immunologically naïve population ) , we have chosen still to refer to the reproduction number of the 2009 virus at the start of each countrys epidemic as R0 ., Each of the three tested mixing matrices was clearly a simplification of the true mixing patterns that might be observed in the studied countries ., M1 and M2 assumed a simple assortativity model ( moderate preference for mixing preferentially within ones own age group ) ., The value of 0 . 25 assumed for the assortativity parameter is broadly consistent with the levels of assortativity seen in the mixing matrices provided by the UK POLYMOD survey 17 ., However , in order to test whether this choice influenced the estimates , we undertook a sensitivity analysis and looked at values in the range 0–0 . 5 ., This indicated that neither reproduction numbers nor susceptibility estimates were strongly affected by varyingθ ., The models presented here were intentionally parsimonious ., Our aim was to compare in the simplest way possible the initial epidemic of a novel influenza in different countries ., The models developed here cannot generate multiple waves of transmission , and do not capture potentially important behavioural changes that may have affected transmission and disease surveillance during the pandemic 37–39 , such as early risk avoidance and higher rates of health-care seeking behaviour early in the pandemic ., In addition we did not allow for the potential impact of school holidays and seasonal climate variation on transmission 40–42 , which may have improved the models fits ., Lastly , only local transmission was considered here ., Imported cases were not considered in the model as one would expect importations to be a substantial proportion of cases only in the first weeks before the epidemic starts and that the transmission would thereafter be predominantly local ., However , by exploring multiple model variants we have demonstrated that estimates of R0 and attack rates are largely robust to uncertainty in the parameterisation of age-specific mixing patterns in the population ., The differences in pandemic surveillance 43 in the countries considered may be the most influential factor affecting the reliability of our estimates and the variation found between countries ., Surveillance to detect virologically confirmed cases of influenza was likely to have been highly non-systematic in several countries and variable throughout the pandemic , meaning the relationship between measured incidence and true incidence of infection may have been highly non-linear ., In particular , many countries which initially undertook highly intensive case finding in 2009 moved to less intensive surveillance once case numbers grew too large for routine virological testing to be undertaken ., Syndromic surveillance of ILI , by comparison , is typically more systematic but suffers from ILI being non-specific for influenza ., All surveillance systems were subject to the effects of changes in health-care seeking behaviour over time ., While we estimate the proportion of infections appearing in surveillance incidence data ( the reporting rate ) , we did not have the statistical power to do anything other than assume that reporting rates were constant over time ., Perhaps the most interesting aspect of our results is that demographic differences between countries may have contributed strongly to the differences in observed H1N1pdm spread ., In particular , we found countries with higher proportions of children ( under 20 ) had higher estimated R0 values and attack rates ., Fits for the global models with shared parameters between countries are clearly poorer than the country-specific fits , but nevertheless capture much of the country to country variation ., That said , fit quality for Argentina and for South Africa may indicate other factors playing a role in determining the observed patterns of transmission ( or alternatively may result from imperfections in surveillance ) ., Several other factors have been demonstrated to impact the Influenza virus transmission , notably seasonal climatic variations , such as absolute humidity and temperature 8 , 44 ., Although the countries examined here have substantial geographical differences between them ( e . g . capital city latitudes between 15°S and 41°S and mean population densities between 3 and 24/km2 ) , no significant association between estimated R0 and latitude or densities of populations were found ( Section 8 and Figure S8 in Supplementary material ) ., More generally , our estimates of reproduction numbers did not differ strongly from those obtained from analyses of the spring/summer wave in countries from the Northern Hemisphere ( US , Mexico and UK ) 16 , 27 , 45 , suggesting a limited impact of seasonal variation in H1N1pdm transmissibility ., Prior immunity could also explain differences between countries as pointed out by recent serological surveys showing that immunity to H1N1pdm varied by country of tested individuals 31 , 34 , 35 , 46–48 ., Results presented here suggest there may be country-to-country differences in epidemiology ( driven in part by demographic variation , but not entirely so ) , suggesting some need to allow for appropriate modification of control policies on a country by country basis ., In particular , targeting vaccination at children may be more optimal for countries with populations with a high proportion of school-age children ., They also support the importance of developing accurate age-structured models for the analysis of influenza epidemics and the potential benefit of extending real time data collection by age-group , on serology and/or reporting rate ., To conclude , this study is one of the first attempts to gain insight into the dynamics of disease transmission via inter-country comparison ., Our analysis has shown that , although differences in spread of H1N1pdm were observed during the Southern hemisphere winter wave , many features of transmission were shared between countries and could be explained with largely common parameters for all countries ., We showed that differences between countries could be partially explained by differences in population demography ., Our results confirm that susceptibility to the virus decreased with age but also that higher contact rates in children may have partly shaped the way H1N1pdm influenza spread in 2009 .

Introduction, Material and Methods, Results, Discussion

While in Northern hemisphere countries , the pandemic H1N1 virus ( H1N1pdm ) was introduced outside of the typical influenza season , Southern hemisphere countries experienced a single wave of transmission during their 2009 winter season ., This provides a unique opportunity to compare the spread of a single virus in different countries and study the factors influencing its transmission ., Here , we estimate and compare transmission characteristics of H1N1pdm for eight Southern hemisphere countries/states: Argentina , Australia , Bolivia , Brazil , Chile , New Zealand , South Africa and Victoria ( Australia ) ., Weekly incidence of cases and age-distribution of cumulative cases were extracted from public reports of countries surveillance systems ., Estimates of the reproduction numbers , R0 , empirically derived from the country-epidemics early exponential phase , were positively associated with the proportion of children in the populations ( p\u200a=\u200a0 . 004 ) ., To explore the role of demography in explaining differences in transmission intensity , we then fitted a dynamic age-structured model of influenza transmission to available incidence data for each country independently , and for all the countries simultaneously ., Posterior median estimates of R0 ranged 1 . 2–1 . 8 for the country-specific fits , and 1 . 29–1 . 47 for the global fits ., Corresponding estimates for overall attack-rate were in the range 20–50% ., All model fits indicated a significant decrease in susceptibility to infection with age ., These results confirm the transmissibility of the 2009 H1N1 pandemic virus was relatively low compared with past pandemics ., The pattern of age-dependent susceptibility found confirms that older populations had substantial – though partial - pre-existing immunity , presumably due to exposure to heterologous influenza strains ., Our analysis indicates that between-country-differences in transmission were at least partly due to differences in population demography .

Although relatively mild , the 2009 H1N1 pandemic reminded us once again of the on-going threat posed by novel respiratory viruses and the need for understanding better how such pathogens emerge and spread ., From April to September 2009 , countries in temperate regions of the Southern hemisphere experienced large epidemics of H1N1pdm during their winter season , with the new virus quickly becoming the predominant circulating influenza strain ., We use mathematical modelling to analyse H1N1pdm epidemiological data from 8 southern hemisphere countries ., We aim at understanding better the factors which may have influenced virus transmission in these countries ., We find that transmissibility of the virus was relatively low compared with previous influenza pandemics , largely because of strong pre-existing age-dependent susceptibility to the virus ( older people being less susceptible to infection , perhaps due to pre-existing immunity ) ., We suggest that population demography had a strong impact on the virus spread and that higher transmission rates occurred in countries having a younger population ., Our results highlight the requirement to use age-structured models for the analysis of influenza epidemics and support the need for country-specific analyses to inform the design of control policies for pandemic mitigation .

medicine, infectious diseases, public health and epidemiology, epidemiology, infectious disease epidemiology, global health, infectious disease control, epidemiological methods, infectious disease modeling

journal.pbio.1001360

Regulation of DNA Replication within the Immunoglobulin Heavy-Chain Locus During B Cell Commitment

During the S phase , mammalian chromosomes replicate in a precise temporal order , with the timing of replication typically changing gradually across hundreds of kilobases ., Cell differentiation induces regional changes in the order of replication which can affect 45% , or more , of the mouse genome 1 ., Various studies have examined how the temporal order of replication is established and modified at specific gene loci , but provided discordant explanations about the role played by DNA origins of replication ., For example , within a 340 kb portion of the Igh locus , changes in replication timing have been linked to modifications in the distribution of active origins and in their firing efficiency ( see definitions in Table 1 ) 2 ., In contrast , within the beta-globin locus , changes in replication timing can occur without significant changes in origin distribution , or firing efficiency , and have been ascribed to modifications in the timing of origin firing 3–5 ., Does this mean that the temporal order of replication is determined by multiple mechanisms ?, Are origin distribution , firing efficiency , and the timing of origin firing regulated independently ?, Which aspect of origin activation is controlled by cell differentiation ?, These are some of the questions addressed in this study ., Answering these questions requires a quantitative understanding of the dynamics of origin firing ., Based on measurements of average origin activity across entire genomes , various stochastic models of origin firing have been recently used to explain specific aspects of eukaryotic DNA replication , such as the duration of S phase 6–13 ., If origin firing can occur stochastically anywhere along the genome and at any time during S phase , origin distribution and the timing of origin firing cannot be responsible for establishing the temporal order of replication 14 ., Recent observations indicate that the profile of replication timing of the budding yeast genome can be explained by differences in the firing rate of individual origins and stochastic origin firing 15 ., However , yeast differs from metazoans in many aspect of DNA replication ( e . g . , S . cerevisiae has well-defined origins of replication , lacks the developmental control of the temporal order of replication , shows no correlation between gene expression and the temporal order of replication , has a short S phase , etc . ) ., In addition , previous studies have mostly relied on the measurement of individual parameters of DNA replication which can be modeled with limited detail to determine the dynamics of origin firing ( e . g . , the timing of replication ) and can produce misleading results if applied to complex genomes 16 ., Hence , testing this hypothesis rigorously in mammalian cells requires the measurement of multiple parameters of DNA replication ., Any large portion of a mammalian genome can be used to test the stochastic firing of origins provided enough information is available about DNA replication ., For this reason , we used the assay called single molecule analysis of replicated DNA ( SMARD ) 17 to collect unbiased information about all aspects of DNA replication initiation , progression , and termination across a 1 . 4 megabase region encompassing the mouse immunoglobulin heavy-chain ( Igh ) locus ( Figure 1A ) ., The experimental data sets collected by SMARD included the temporal order of replication , the steady-state distribution of replication forks , the time required to replicate the region , the average speed of replication forks , the distribution of initiation and termination events , the percentage of replicating molecules containing initiation and termination events , and the average number of events per molecule ., Using a novel mathematical procedure 18 , we established that the experimental data sets collected by SMARD are fully consistent with the stochastic firing of origins as defined in Table 1 ., We also show that many aspects of DNA replication ( including the temporal order of replication ) can be explained by variations in the rate of origin firing I ( x , t ) along the Igh locus ( see definition in Table 1 ) ., According to the nomenclature proposed by others 19 , this rate indicates the number of initiation events occurring per length of unreplicated DNA , over a given period of time , as mathematically defined in Materials and Methods ., Our results point to significant differences in the regulation of DNA replication between the mouse Igh locus and yeast chromosomes ., In S . cerevisiae , each origin of replication appears to be characterized by a specific firing rate , which differs for different origins 15 ., Within the Igh locus , initiation events lack focal points corresponding to individual origins ., Instead , the locus comprises large domains ( spanning tens or hundreds of kilobases ) where the firing rate of multiple origins is virtually uniform and similarly regulated ., Along the locus , changes in the firing rate occur abruptly at the border between different Igh domains , and the firing rate of each domain is not affected by deletions that span multiple origins ., It is the combined effect of different domain sizes and firing rates that determines the temporal order of replication ., This organization remains valid for cells at different stages of B cell development ( e . g . , bone marrow pro-B cells blocked at the uncommitted and committed stages of differentiation by homozygous mutations of the Pax5 and Rag2 genes; Figure 1B ) ., We also show that the changes in DNA replication that occur during B cell commitment can be quantitatively explained by substantial changes in the firing rate of origins within specific domains of the Igh locus ., Therefore , the rate of origin firing is the parameter that is being regulated across large sections of the locus during cell differentiation ., The role of the developmental regulator Pax5 in this process and its mechanism of action are also examined ., Bone marrow pro-B cells isolated from Pax5−/−Rag2−/− mice retain the ability to proliferate but are blocked at the uncommitted stage of differentiation and maintain the Igh locus in germline configuration 20 ., In order to perform SMARD , we sequentially labeled a population of exponentially growing cells with 5′-iodo-2′-deoxyuridine ( IdU ) and 5′-chloro-2′-deoxyuridine ( CldU ) , for 3–4 h ( see Table S1 , column b ) ., Under these conditions , each labeling period is long enough to allow the complete replication of large sections of the Igh locus , resulting in DNA molecules variably substituted with the halogenated nucleotide ( see example in Figure S1 ) 2 , 17 , 21 , 22 ., We then digested the genomic DNA with restriction enzymes that cut infrequently within the locus ( PmeI , or PacI , or SwaI ) and isolated four of the resulting restriction fragments by pulsed-field gel electrophoresis ( gray bars in Figure 1A ) ., These fragments were stretched on microscope slides , hybridized with specific DNA probes , and analyzed by fluorescence microscopy to detect the incorporation of the nucleotide analogs along individual DNA molecules ., The molecules fully substituted with halogenated nucleotides provide a vast amount of information about the process of DNA replication 2 , 17 , 21 , 22 ., In Pax5−/−Rag2−/− cells , the analysis of ∼4 , 000 hybridization signals yielded 1 , 304 fully substituted DNA molecules , 764 of which met the standards required to perform precise measurements ( Table S1 , columns c–f ) ., This population includes molecules that incorporated only one type of halogenated nucleotide ( single-labeled molecules ) , as well as molecules that incorporated both of them ( double-labeled DNA molecules ) ., As explained in previous publications 17 , 22 , the ratio between single- and double-labeled molecules is proportional to the time required to replicate each restriction fragment ( Table S1 ) , which is linked to the average number of replication forks participating in the replication of each restriction fragment and to their average speed ( Table S1 ) ., Thus , these values can be directly determined from the experimental data as described in the legend of Table S1 ., This allowed us to determine that , in uncommitted pro-B cells , the average speed of replication forks is similar within the four restriction fragments ( between 2 . 25 and 3 kb/min ) ., A similar value was also measured at the genomic level using a different assay ( ∼2 . 5 kb/min; IV and PN , unpublished observation ) ., This suggests that replication forks move at comparable speeds throughout the Igh locus ., In the double-labeled DNA molecules , IdU-CldU transitions mark the positions of replication forks at the time of the label switch ., Initiation events appear as IdU-labeled regions surrounded by CldU , while fork collisions display the complementary pattern ( e . g . , Figure 1C–D ) ., Figure 2A summarizes the location of each initiation event ( red bars ) and fork collision ( green bars ) detected in 158 individual restriction fragments ., The normalized frequency of the events across the four fragments is also indicated ( expressed as number of events scored per 100 double-labeled DNA molecules , per 100 kb ) ., The values before normalization are presented in Table S1 ., Note that initiation events and fork collisions spanning adjacent restriction fragments are not shown in the figure since these events are scored as simple IdU-CldU transitions ( a fact that is taken into consideration by our mathematical model and in Figure S2C ) ., Overall , these results indicate that initiation events are more frequent near the DH-JH and middle-VH gene families than in other parts of the locus ., In these regions , the presence of non-overlapping events implies the existence of clusters of active origins that are reminiscent of initiation zone 23 , 24 ., In contrast , fork collisions are more frequent between the origin clusters ( termination regions in Figure 2A ) ., It is important to point out that SMARD experiments are designed to detect only a fraction of the active origins ( the number of double-labeled DNA molecules required to reach mapping saturation increases with the number of potential origins ) ., Moreover , reaching mapping saturation is increasingly unlikely where origin efficiency is low , and near the end of the restriction fragments ., Thus , Figures 2A and S2 are expected to largely underestimate the actual density of origins along the locus ( see definition in Table 1 ) ., Even so , the distance between the midpoints of initiation event within the DH-JH region suggests an origin density higher than one potential origin per 10 kb ( Figure S2E , PacI#3 ) ., In Figure 2A , origin density appears lower near the middle-VH region genes ., However , within this section of the locus , the midpoints of initiation events overlap only rarely , indicating that that origin mapping is far from saturation ( Figure S2E , PmeI#5 ) ., Lack of saturation is even more pronounced across the proximal-VH region ( none of the detected initiation events overlap; Figure 2A , PmeI#4 ) ., We conclude that potential origins of replication are likely to be present at a relatively high density across most of the Igh locus ., Since cells grow asynchronously during DNA labeling , the IdU-CldU transitions depict all stages of DNA replication , reflecting the dynamics present at the steady-state of growth ( right panel in Figure S1 ) 2 ., This means that the population of double-labeled DNA molecules can be used to determine the steady-state distribution of replication forks across the Igh locus ( arrowheads in Figure 2B–C ) , and the average number of forks participating in the replication of each restriction fragment ( Table S1 , column l ) ., Similarly , the temporal order of replication can be obtained from the level of IdU substitution of these molecules , with peaks marking the regions that replicate first and valleys marking the regions that replicate last ( Figure 2D ) ., Overall , these results indicate that , in uncommitted pro-B cells , DNA replication tends to begin near the DH-JH and middle-VH gene families , as well as at origins located 3′ and 5′ of the Igh locus ., From there , replication forks proceed to replicate the locus until they collide with an oppositely moving fork ( predominantly within the termination regions ) ., The data presented above show that the replication of the Igh locus follows a precise temporal program ., However , this program represents a population average ., At the level of individual DNA molecules , initiation events do not seem to occur in any particular order , with different origins firing in different molecules ., Initiation events were also detected within the portions of the Igh locus that replicate last ( e . g . , the termination region within PmeI#4 and the region between Cδ and Cγ3; Figure 2A ) , taking place as the corresponding molecules were at various stages of their replication ., Such events are not easily explained by deterministic models of origin activation ., A domino activation of origin firing 13 also seems unlikely ., The initiation events detected in the population of double-labeled DNA molecules are rarely associated with externally generated replication forks , and when such forks are present their location is tens to hundreds of kilobases from the firing origins ( e . g . , 40 and 100 kb in the example shown in Figure S2A ) ., Hence , it is possible that Igh origins fire according to stochastic dynamics ., To determine whether the results obtained by SMARD can be quantitatively explained by stochastic origin firing ( see definition in Table 1 ) , we used a mathematical formalism and a simulation procedure that we recently developed for this purpose 18 ., As briefly described in Materials and Methods , this procedure allows us to fit many of the data collected by SMARD ( namely , the distribution of the replication forks , Figure 2B–C , the temporal order of replication , Figure 2D , and the replication time of the restriction fragments , Tr , Figure S2B ) to computer-generated curves calculated from a series of rate equations ., In the simplest possible scenario , the curves were calculated assuming that bidirectional origins of replication fire stochastically throughout the Igh locus and the S phase ( see Table1 for a precise definition ) , generating forks that move at a constant speed ., In these calculations , the rate of origin firing was the only parameter allowed to vary freely across the Igh locus , while remaining constant in time ( meaning that , within a genomic region , initiation events continue to occur at the same rate from the beginning of S phase until the region is replicated in the entire population of cells ) ., It is important to point out that the rigid constrains imposed by this scenario do not accurately reflect physiological conditions ., For example , it is known that modest changes in the speed of replication forks can indeed occur along the genome and during the S phase 25 ., However , by limiting the number of free variables , this scenario allows us test the stochastic firing hypothesis more stringently ., Strikingly , we found that this simple scenario is sufficient to reproduce all experimental data sets collected by SMARD ( Figures 2B–D and S2B–C ) ., This scenario can even reproduce data sets that were not used during the fitting procedure ( such as the location of initiation events and fork collisions , the number of molecules containing such events , the average number of events per molecules , and the average speed of replication forks ) ., Since the fit was performed simultaneously for all restriction fragments , the presence of local discrepancies is not particularly surprising ( e . g . , PmeI #4 in Figures 2D and S2B ) ., This is likely to reflect the constrains imposed by this scenario ( e . g . , within some portions of the locus fork speed may deviate from the average and origin density may also vary ) ., Nevertheless , the high quality of the fit ( reduced chi-square , 1 . 18 ) indicates a close match between calculated and experimental data sets ., Hence , the results collected by SMARD are fully compatible with the stochastic firing of origins throughout the Igh locus ., In addition to the changes in firing rate along the locus , scenarios involving a larger number of variables were also considered ( e . g . , allowing for a variable speed of replication forks or for changes in the rate of origin firing during the S phase ) ., In principle , changing these parameters could have a major effect on the replication dynamics of the Igh locus ., However , these scenarios improved the quality of the fit only marginally ( M . G . G . , J . B . , and P . N . , unpublished observation ) ., This means that variations in parameters other than the firing rate of origins along the genome , while possible , have a limited impact on the replication dynamics and the temporal order of replication of the Igh locus in uncommitted pro-B cells ., The high quality of the fit calculated above indicates that the computer-generated data set provides a good approximation of the firing rate of origins across the Igh locus ., Figures 2E and S2D show that the firing rate is very low throughout most of the locus ( 4 . 5×10−6 initiation events per kb per minute ) ., Two exceptions are the DH-JH region ( 55±13 . 2 kb in size ) and of the middle-VH region ( 281±43 . 1 kb in size ) , where the firing rate is up to 77-fold higher ., This variation implies that origins located in different parts of the Igh locus have a very different tendency to fire ., Another striking feature of these results is that the firing rate changes abruptly at a few specific locations ( transitions ) , while remaining virtually uniform across large sections of the locus ( plateaus ) ., Attempts at fitting bell-shaped curves ( which lack both plateaus and sharp transitions ) produced fits of lower quality compared to box-shaped curves of the kind shown in Figure 2E ( M . G . G . , J . B . , and P . N . , unpublished observation ) ., Therefore , the Igh locus appears to be organized in precisely defined domains where origins have a similar rate of firing ( suggesting a significant degree of coordination among the origins of individual domains ) ., We conclude that the temporal order of replication is a consequence of the combined effect of domain sizes and firing rates ., Equation 3 , in Materials and Methods , also allows us to determine the efficiency of origin firing for specific sections of the Igh locus , according to the definition provided in Table 1 ., The values obtained for the DH-JH and middle-VH regions are only 1 . 1 and 1 . 5 initiation events , per allele , per S phase ( IAS , gray dashed lines in Figure 2E ) ., These regions are known to contain multiple origins of replication ( Figure 2A ) ., Hence , most of the origins remain silent during each cycle of replication ( firing is inefficient ) ., Notably , the portion of Igh locus spanning the proximal-VH genes produces 0 . 4 IAS even if its firing rate is about 2 orders of magnitude lower than the DH-JH and middle-VH regions ., The nonlinear relationship between firing rate and origin efficiency can be explained by the fact that the proximal-VH genes occupy a very large genomic region and tend to replicate after adjacent portions of the locus ( Figure 2D ) ., This provides more time for the origins therein located to fire ., Hence , efficiency is not an intrinsic property of individual origins , but rather the result of the distribution of the firing rate throughout the Igh locus ., The computer-generated data set allows us to draw a few additional conclusions ., Although Igh origins fire inefficiently , they are responsible for the replication of 87% of the locus , with only minor contributions from external origins ( Figure S2F ) ., This means that inefficient origin firing extends beyond the margins of the locus into the surrounding regions ., In addition , various portions of the locus have a very high probability of being replicated by forks moving in one particular direction ( peaks and valleys in Figure S2G ) ., Since the molecules analyzed in this experiment originated from both Igh alleles , the strong bias in fork direction indicates that both alleles follow a similar replication program ., Finally , we calculate that it takes approximately 4 h to complete the replication of this portion of the genome in 100% of the cell population ( Figure 2F ) ., Hence , these experiments provide a complete description of DNA replication , within the Igh locus , over a broad portion of the S phase ., Bone marrow pro-B cells from Pax5−/− mice are blocked at the uncommitted stage of differentiation but , unlike Pax5−/−Rag2−/− cells , they can undergo DH-JH recombination ., The resulting loss of various sections of the DH-JH origin cluster ( the ∼55±13 . 2 kb region containing the origins with the highest firing rate; Figure 2E ) provides us with a tool to study how the firing rate is regulated ., Specifically , we can use ex vivo propagation to obtain clonal populations of cells carrying particular DH-JH deletions ., Here , we studied a heterozygous pro-B cell clone that carries a 65 kb deletion on its 129/Sv allele and a 25 kb deletion on the C57BL/6 allele ., The different sizes of the resulting restriction fragments allowed us to use SMARD to investigate the effect of each deletion on DNA replication ., Consistent with the loss of the entire DH-JH origin cluster , the 65 kb deletion reduced the number of initiation events occurring within fragment PacI#3 to an undetectable level ( left portion of Figure 3A and Figure S3B ) ., The direction of replication fork movement ( Figure 3B–C ) and the uniform slope of the temporal order of replication ( Figure 3D ) indicate the passive replication of the region by forks originating 3′ of the Igh locus ., This is similar to results previously obtained in non-B cells , where the Igh locus is part of a replication timing transition region 2 ., Besides for the loss of the DH-JH origins , the rate of origin firing in adjacent portions of the locus remains unchanged ( compare PacI#3 and SwaI#2 in Figures 2E and 3A ) ., This rate is very low compared to the origins located at 3′ of the Igh locus ., Hence , even if the 5′-end of PacI#3 now replicates 3 h later compared to unrearranged pro-B cells ( compare Figure 3F at 280 kb to the corresponding portion of Figure 2F ) , this delay is not sufficient to produce large numbers of initiation events within PacI#3 and SwaI#2 ( infrequent initiation events may still occur within this portion of the Igh locus but their visualization would require the analysis of a much larger sample of double-labeled DNA molecules ) ., In contrast , the 25 kb deletion removes less than half of the DH-JH origin cluster ., On this allele , initiation events and fork collisions continue to occur at the same locations described for Igh locus in germline configuration ( compare the right portion of Figure 3A with Figure 2A ) ., However , there is a strong right-to-left bias in the direction of fork movement ( Figure 3B–C ) , and the IdU content shows a steady decrease in the same direction with an inflection point at the site of the remaining DH-JH origins ( Figure 3D ) ., These factors indicate the passive replication of the region in a fraction of the cell population ., Accordingly , there is a 40% reduction in the number of initiation events occurring within this section of the locus during each S phase ( 0 . 7 IAS , Figure 3E ) , compared to the level detected in unrearranged pro-B cells ( 1 . 1 IAS , Figure 2E ) ., These results indicate that the decrease in initiation events is proportional to the reduction in size of the DH-JH origin cluster ., Hence , the firing rate of the remaining Igh origins and the location of the firing rate transitions are unaffected by the 25 kb deletion ( Figure 3E ) ., This is reminiscent of results obtained by studying deletions of the DHFR locus by 2D-gel electrophoresis ( although in that case conclusions were based on measurement of origin efficiency , which is context dependent , and not of firing rate ) 26 , 27 ., We conclude that firing rate of origins is regulated independently in different sections of the locus and that origin activity at one location is not affected by the presence , or absence , of neighboring origins ., In order to study how the replication of the Igh locus changes during cell differentiation , we isolated B cell progenitors from the bone marrow of a Rag2−/− mouse ( 129/Vs . ) ., These cells efficiently undergo B cell commitment but maintain the Igh locus in germline configuration , which prevents them from developing any further ., In these cells , initiation events and fork collisions are distributed across most of the Igh locus ( Figure 4A ) ., Only a few of these initiation events are centered at the same genomic location ( Figure S4 ) , indicating that the experiment is far from mapping saturation ( leading to an underestimate of origin density ) ., This suggests that potentially active origins are present at high density across most of the Igh locus , perhaps every 10–20 kb , although larger gaps may exist at a few locations ., Despite the widespread activation of origins , it is still possible to distinguish portions of the locus where replication forks have a preferred direction of movement ( e . g . , PacI#3; Figure 4B–C ) and regions that replicate first and last ( Figure 4D ) ., Thus , our results indicate that the level of origin activity is not uniform across the Igh locus and point to major differences in DNA replication between committed and uncommitted pro-B cells ., Once more , we found that the data obtained by SMARD are fully compatible with the stochastic firing of origins ( reduced chi-square , 1 . 03; Figure S4C ) ., The nearly perfect fit confirms that the temporal order of replication is mostly determined by the regulation of a single variable along the locus ( the firing rate of origins ) ., Computer calculations also show that the rate of origin firing is not uniform ( blue line in Figure 4E ) ., A comparison with the results obtained for uncommitted pro-B cells ( red dotted line ) reveals that B cell commitment is accompanied by changes in the rate of origin firing across most of the Igh locus ., However , the largest changes ( whether positive or negative ) occur in regions where a previous genomic screening identified the presence of Pax5 binding sites in committed pro-B cells 28 ., For example , the firing rate increases nearly 50-fold across the CH-3′RR region ( ∼216 kb in size and containing two Pax5 binding sites ) and decreases 10-fold throughout the DH-JH origin cluster ( ∼55 kb in size and containing one Pax5 binding site ) ., These results indicate that the rate of origin firing is the parameter that is being regulated during cell differentiation ., In addition , they suggest that Pax5 participates in regulating the firing rate of origins during B cell commitment ( although the number of known binding sites for this factor is far smaller than the number of potential origins affected by its expression ) ., As a result of B cell commitment , changes in origin efficiency were also observed throughout the Igh locus ( compare gray dotted lines in Figures 2E and 4E ) ., However , the overall efficiency of origins across the 1 . 4 Mb region increases only from 3 . 2 IAS in uncommitted pro-B cells ( Figure S2D ) to 7 . 4 IAS in uncommitted pro-Bs ( Figure S4C ) ., Thus , the firing rate can increase by 1–2 orders of magnitude across most of the Igh locus but produce variations of only 2 . 3-fold in the total number of initiation events ( and replication complexes ) involved in the replication of the locus ., Given the high density of potential origins throughout the Igh locus , these results also suggest that the firing efficiency of individual origins is below 10% ., Therefore , following B cell commitment , origins continue to fire inefficiently even if the firing rate increases across most of the Igh locus ., Reconstituting Pax5 expression in Pax5−/− pro-B cells induces B-cell commitment 20 , 29 and can be used to study the role of this protein in origin regulation ., For this purpose , we transduced bone marrow pro-B cells from a Pax5−/− mouse ( 129/Sv-C57BL/6 ) with a retroviral vector containing the expression cassette Pax5ER-IRES-GFP 30 , 31 ., We then sorted and expanded the GFP+ cells to obtain a polyclonal population expressing Pax5ER ( KO-Pax5ER pro-B cells; Figure S5A–B ) ., This protein is the fusion product of Pax5 and the hormone-binding domain of the estrogen receptor , which becomes biologically active in the presence of 4-hydroxy-tamoxifen ( 4-OHT ) ., In KO-Pax5ER pro-B cells , 4-OHT is able to induce commitment-specific changes in the methylation of the 3′RR DNA , indicating that Pax5ER is able to interact with at least one of the Pax5 binding sites of the CH-3′RR region 32 ., In KO-Pax5ER pro-B cells , the occurrence of B-cell commitment was monitored by the Pax5-dependent expression of the cell-surface-marker CD19 ., Before the addition of 4-OHT , we consistently found that only 4% of KO-Pax5ER pro-B cells are CD19+ ( Figures 5A and S5C ) ., However , 65%–90% of cells become CD19+ after the addition of 4-OHT ., Thus , in our inducible system , the activity of Pax5ER is modest in the absence of 4-OHT but increases dramatically after induction , leading to B cell commitment ., Can the induction of Pax5ER increase the firing rate of origins within the CH-3′RR region ?, To answer this question , we studied KO-Pax5ER pro-B cells before and after induction with 4-OHT for 28 h ( see labeling scheme in Figure 5B ) ., This induction time was chosen because it allows enough time for the synthesis and turnover of the gene products regulated by Pax5 and for cells to become fully committed to the B lineage 30 ., In the absence of 4-OHT , we detected a limited number of initiation events within the CH-3′RR region , which is consistent with the modest activation of Pax5ER described above ( first and third column in Figure 5C ) ., However , this portion of the locus continues to be passively replicated in the majority of cells , as indicated by the replication fork distribution and the temporal order of replication ( Figures 5D–F and S5H–I ) ., In contrast , 4-OHT induction profoundly alters all aspects of DNA replication within the CH-3′RR region , suggesting a strong increase in origin activity ( second and fourth column in Figures 5C–F and S5H–I ) ., Numerical calculations confirm that the firing rate of origins reaches the level detected in committed pro-B cells ( compare the 129/Sv results in Figures 5G and 4E ) ., Hence , inducing B cell commitment in vitro produces changes in the firing rate of Igh origins that are similar to those observed in pro-B cells isolated from mice ., Additional experiments confirmed that these effects are specific ., For example , the firing rate of CH-3′RR origins does not increase when 4-OHT is provided to uncommitted pro-B cells that do not express Pax5ER ( Table S1 ) ., In addition , the induction of Pax5ER in KO-Pax5ER pro-B cells does not affect origin activity broadly and non-specifically ., For example , we did not observe significant changes in cell proliferation and cell-cycle profile ( Figure S5L ) or the appearance of markers of DNA damage and DNA damage checkpoint activation ( Figure S5M ) ., We conclude that the change in origin activity observed in KO-Pax5ER pro-B cells results from the activation of Pax5ER and the induction of B cell commitment ., B cell commitment involves changes in the expression of hundreds of tissue-specific genes 33 , raising the question of whether Pax5 is directly responsible for regulating the firing rate of origins ., After the addition of 4-OHT , Pax5ER requires only a few minutes to translocate from the cytoplasm to the nucleus , where it changes the expression of its target genes over a period of several hours 30 ., We suspected that the firing rate would change very slowly if its reg

Introduction, Results, Discussion, Materials and Methods

The temporal order of replication of mammalian chromosomes appears to be linked to their functional organization , but the process that establishes and modifies this order during cell differentiation remains largely unknown ., Here , we studied how the replication of the Igh locus initiates , progresses , and terminates in bone marrow pro-B cells undergoing B cell commitment ., We show that many aspects of DNA replication can be quantitatively explained by a mechanism involving the stochastic firing of origins ( across the S phase and the Igh locus ) and extensive variations in their firing rate ( along the locus ) ., The firing rate of origins shows a high degree of coordination across Igh domains that span tens to hundreds of kilobases , a phenomenon not observed in simple eukaryotes ., Differences in domain sizes and firing rates determine the temporal order of replication ., During B cell commitment , the expression of the B-cell-specific factor Pax5 sharply alters the temporal order of replication by modifying the rate of origin firing within various Igh domains ( particularly those containing Pax5 binding sites ) ., We propose that , within the Igh CH-3′RR domain , Pax5 is responsible for both establishing and maintaining high rates of origin firing , mostly by controlling events downstream of the assembly of pre-replication complexes .

Each time a mammalian cell duplicates its genome in preparation for cell division it activates thousands of so called “DNA origins of replication . ”, The timely and complete duplication of the genome depends on careful orchestration of origin activation , which is modified when cells differentiate to perform a specific function ., We currently lack a universally accepted model of origin regulation that can explain the replication dynamics in complex eukaryotes ., Here , we studied the mouse immunoglobulin heavy-chain locus , one of the antibody-encoding portions of the genome , where origins change activity when antibody-producing B cells differentiate in the bone marrow ., We show that multiple aspects of DNA replication initiation , progression , and termination can be explained mathematically by the interplay between randomly firing origins and two independent variables: the speed of progression of replication forks and the firing rate of origins along the locus ., The rate of origin firing varies extensively along the locus during B cell differentiation and , thus , is a dominant factor in establishing the temporal order of replication ., A differentiation factor called Pax5 can alter the temporal order of replication by modifying the rate of origin firing across various parts of the locus .

developmental biology, theoretical biology, genetics, immunology, biology, computational biology, molecular cell biology

journal.pcbi.1000273

Model-Based Therapeutic Correction of Hypothalamic-Pituitary-Adrenal Axis Dysfunction

The hypothalamic-pituitary-adrenal ( HPA ) axis constitutes one of the major peripheral outflow systems of the brain , serving to maintain body homeostasis by adapting the organism to changes in the external and internal environments ., It does this by regulating the neuroendocrine and sympathetic nervous systems as well modulating immune function 1 ., Through regulation of these systems , the HPA axis initiates and coordinates responses to physical stressors; such as infection , hemorrhage , dehydration , thermal exposure and to neurogenic stressors; such as fear , anticipation and fight or flight ., Many aspects of the organization and function of the HPA axis have been characterized in clinical and laboratory studies revealing a number of component feedback and feed forward signaling processes ., Stress activates the release of corticotropin-releasing hormone ( CRH ) from the paraventricular nucleus ( PVN ) of the hypothalamus ., The release of CRH into the hypophysial-portal circulation in turn acts in conjunction with arginine vasopressin on CRH-R1 receptors of the anterior pituitary stimulating the rapid release of adrenocorticotropic hormone ( ACTH ) ., ACTH then is released into the peripheral circulation and stimulates the release of the glucocorticoid cortisol from the adrenal cortex by acting on the receptor MC2-R ( type 2 melanocortin receptor ) ., Cortisol enters the cell and binds to the glucocorticoid receptor present in the cytoplasm of every nucleated cell; hence the widespread effects of glucocorticoids on practically every system of the body including endocrine , nervous , cardiovascular and immune systems ., To keep HPA axis activity in check , glucocorticoids also exert negative feedback at the hypothalamus and pituitary glands to inhibit the synthesis and secretion of CRH and ACTH , respectively ., Moreover , glucocorticoid negative feedback causes a reduction in corticotroph receptor expression leading to a desensitization of the pituitary to the stimulatory effects of CRH on ACTH release ., This negative feedback is also felt in the hippocampus where it exerts a negative influence on the PVN ., A detailed review of the physiology and biochemistry of the HPA axis as well as its know interactions with the immune system may be found in work by Silverman et al . 2 ., A number of chronic diseases have been characterized by abnormalities in HPA axis regulation ., These include major depression and its subtypes , anxiety disorders such as post-traumatic stress disorder , panic disorder and cognitive disorders such as Alzheimers disease and minimal cognitive impairment of aging 3 ., Dysregulation of the HPA axis has also been linked to the pathophysiology of Gulf War illness 4 , post-infective fatigue 5 , and chronic fatigue syndrome ( CFS ) 6 , 7 ., It is not clear what causes this dysregulation , but it is manifested in many HPA axis disorders as a hypercortisol or hypocortisol state ., The existence of these separate and stable states is not surprising when one considers the multiple feedforward and feedback mechanisms that regulate the HPA axis ., Systems such as this often display complex dynamics that readily accommodate multiple stable steady states which are known as attractors because the system is naturally drawn back to these resting states after perturbation ., However , if the perturbation is of sufficient strength and duration , the system can be pushed away from a given resting state and into the basin of new attractor ., Though much is known about its components , one of the main difficulties in studying the behavior of the HPA axis has been in integrating the expansive body of published experimental information ., Numerical models provide an ideal framework for such integration ., Simple models of the HPA axis have been constructed using deterministic coupled ordinary differential equations 8 , 9 ., Though successful in reproducing some of the basic features of HPA axis dynamics these early models neglected to include feedback and feed-forward immune effector molecules and associated mechanisms ., Linear approximations of some components lead to unrealistic predictions beyond a very narrow region of concentrations ., In addition transport processes involved in the distribution of these chemical signals from the brain throughout the body were not modeled explicitly ., This level of abstraction made direct comparison of simulation results to actual HPA axis chemistry and physiology highly tenuous ., In a move towards increased fidelity Gupta et al . 10 introduced a more detailed description of glucocorticoid receptor dynamics enabling the latter to demonstrate bistability in HPA axis dynamics ., As mentioned previously this theoretical proof of the existence of a second stable steady state is highly compatible with clinical observations ., Moreover the abnormally low cortisol levels characterizing this stable resting state or basin of attraction are consistent with documented observations of hypocorticolism in patients with CFS 11 , Gulf War illness and other similar conditions 12–14 ., In this work we adopt the model proposed by Gupta et al . 10 as a recent and detailed representation of the HPA axis ., On the basis of this model we propose a framework for estimating robust corrective measures for displacing the HPA axis from a chronic hypocortisol state back to a healthy state ., Using model-based predictive control ( MPC ) methodology we demonstrate that it is possible to compute such treatment time courses while dealing with the inherently high level of uncertainty characteristic of biological systems ., While this uncertainty might lead to compromises in efficiency , interventions can be computed that predict a positive outcome ., Our analysis indicates that one such treatment could involve a pharmacologically induced reduction in cortisol forcing a build-up of ACTH ., Upon reaching a specific threshold concentration of ACTH , the intervention is discontinued and the HPA axis will return to a healthy steady state under its own volition as this is now the closest attractor for the system ., A model of the HPA axis which includes glucocorticoid receptor and the dynamics of glucocorticoid receptor-cortisol interactions have been proposed by Gupta et al . 10 ., This model is described by the following differential equations as System H ( Eq . 1 ) ., ( 1 ) The system states are given as x\u200a=\u200ax1; x2; x3; x4T and are described in Table, 1 . Note that the states in this model are scaled values , as described by Gupta et al . 10 ., The system parameters are given by the vector p\u200a=\u200aki1; kcd; kad; ki2; kcr; krd; kT ., Nominal values for the system parameters are listed in Table, 2 . The variable d in System H is the stress term which describes the effect of stress ( both physical and psychological ) on the hypothalamus ., This variable is seen as a disturbance that perturbs the System H from a steady-state value ., In this first analysis the HPA axis system is considered under idealized conditions where all parameters are assumed constant and precisely known ., In addition , the states x are assumed known as a function of time with no measurement error and the control action is implemented perfectly ., The approach taken for choosing an optimal control is based on the Model Predictive Control ( MPC ) framework 15–17 ., Under this framework , an objective function of the manipulated and measured variables is defined ., Typically the objective function is a mathematical expression which corresponds to engineering objectives or underlying system constraints ., The input computed under the MPC framework is the one in a class of permissible inputs that minimizes the chosen objective function ., In this work it is assumed that the variable to be manipulated for treatment is the rate of addition or removal of cortisol from circulation ., To model this control action , System H is augmented with a control term u in the equation for cortisol ( x4 ) ( Eq . 2 ) ., Note that System Hu is affine with respect to the control action and the disturbance ., ( 2 ) To avoid dangerous destabilization of the HPA axis by the application of control action u ( t ) we define the following penalty function to enforce minimal departure from normal ACTH ( x2 ) and cortisol ( x4 ) levels even though we purposely manipulate circulating cortisol to perturb the system Hu ., Where t0 and tf are the start and end time of the optimization horizon , λ is a tuning parameter taking values from zero to one and x2* and x4* are the healthy steady-state concentrations of ACTH and cortisol , respectively ., R is a penalty assigned to the input and Q is the penalty assigned to the state variables ., R was chosen as 0 because the cost for therapy was considered negligible compared to the cost of ongoing disability ., Q was chosen as follows because x2 and x4 are the only measured states ., The resulting cost function can be written as: ( 3 ) The parameter λ is used to penalize excessive imbalance of the other hormones ( x1 , x2 , x3 ) in response to the control action applied to cortisol ( x4 ) ., In this case , the objective of the controller is to bring the cortisol concentration to set point while minimizing the impact of the treatment on the other three states of the HPA axis ., Any change in CRH ( x1 ) or the glucocorticoid receptor ( GR; x3 ) will be reflected in the concentration of ACTH ( x2 ) by virtue of the coupled dynamics described by System Hu ., The tuning parameter λ can be selected to match the intensity of the desired treatment ., A λ value of near zero will lead to a more intense treatment while a value of λ near one will lead to very conservative treatment ., For proof of concept , a more direct treatment was favored in this work and a λ value of 0 . 01 was used throughout ., Note that x2* and x4* correspond to the stable steady state of the unperturbed system ( i . e . , when u\u200a=\u200a0 ) ., As a result , once the system has been brought to the healthy steady state it will stay at this steady state even if the external control action ( treatment ) is removed ., Typically , a treatment or control action is applied at discrete intervals ., As a result , the objective function in Equation 3 was optimized with respect to a piece-wise constant input signal x4 ( u ( t ) ) ., That is , the optimization procedure searched for an optimal input in the set Uc of all piecewise constant functions on defined such that the input level may be changed every 1/2 scaled time unit ., The optimization problem was therefore posed in Equation 4 as: ( 4 ) The initial condition , x ( t\u200a=\u200at0 ) is the steady state of the unperturbed system with d0\u200a=\u200a0 ., The optimal control , was computed using Matlabs built-in “fminsearch” function ., The steady-state solutions for HPA axis model described above as System H can be computed by setting and treating the right side of System H as a set of four algebraic equations in the four unknowns {x1; x2; x3; x4} ., Under this framework , the disturbance variable , d , is assumed to take on a constant value ., At steady state the system is therefore described by the following equations ( Eq . 5–8 ) ., ( 5 ) ( 6 ) ( 7 ) ( 8 ) The above is a set of polynomials in x , with real coefficients , and maximum total degree of five ., Equations 5 to 8 can be simplified using the theory of polynomial ideals 18 ., Specifically , the latter can be reduced to the following set of equations ( Eq . 9–12 ) ., ( 9 ) ( 10 ) ( 11 ) ( 12 ) Therein f3 is a polynomial in x3 of degree seven , and f1 , f2 and f4 are functions only of x3 and d0 ., The functions f1 to f4 can be computed using a symbolic algebra package such as Maple ., For the nominal parameter values proposed in Gupta et al . 10 there are at most three real-valued solutions for x3 and these correspond to the roots of f3 ., Each root is a steady-state value for x3 and can be used to generate the corresponding values of x1 , x2 and x4 given Equations 10 to 12 ., Note that at steady state x2\u200a=\u200ax4 ( Eq . 8 ) ., A plot of the steady-state values of x1 , x2 and x3 as a function of d0 is shown in Figure 1 ., In this model of HPA axis dynamics a chronically stressed individual would occupy the stable steady state associated with a depressed cortisol concentration ( ∼0 . 05 ) at rest or at d0\u200a=\u200a0 ., If a healthy person were subjected to extreme stress ( i . e . , d0>0 . 168 ) for an extended period of time their body would reach the only steady state available locally that is one corresponding to chronic stress ., In other words , for values of d0 greater than 0 . 168 , Equation ( 9 ) dictates that there is only one steady-state solution for free GR ( x3 ) concentration as opposed to the 3 solutions available for 0≤d0<0 . 168 ., By virtue of Equation ( 12 ) this results in only one steady-state solution being available for cortisol ( x4 ) for d0>0 . 168 ., When the stress is removed ( i . e . , d0\u200a=\u200a0 ) , the body will stay at this new depressed steady-state value of cortisol concentration ., This process is shown graphically in Figure 2 by the red dashed trajectory ., According to this model the inability of the body to return to the healthy steady state is due to the fact that once the body establishes a new equilibrium it inherently seeks to stay near this point ., In order to force the body to return to its original equilibrium its state must first be shifted to a point where the only stable condition in proximity is one corresponding to this original healthy state ., Once this is done , the internal regulatory mechanisms of the body will ensure that this healthy stable point is achieved and maintained ., This approach is illustrated in Figure 2 by the green dashed trajectory ., The design of such a shift is presented in the following section ., As one might expect the assumptions of ideal control do not correspond to a physically realizable system ., However , the analysis of the system under idealized conditions allows the study of possible treatments ., Any practical treatment would then be a suboptimal solution as compared to the treatment under idealized conditions ., This allows proposed treatments to be benchmarked and compared ., In addition , the solution obtained under idealized conditions can serve as a qualitative guideline for the creation of a practical , although suboptimal , treatment ., In engineering terms the objective of treatment is to succeed in bring the subject to the healthy steady-state target while exerting the smallest disturbance possible to the HPA axis ., For example , even though we intend to manipulate circulating cortisol concentration it should not be allowed to decrease excessively because of the important role cortisol plays in regulating a number of cellular and physiological functions ., To avoid such excess perturbations the concentration of ACTH has been included in the objective function of Equation 3 . This concentration is more readily measured than that of either CRH or GR making ACTH a good candidate for monitoring the progress of a treatment ., The optimal control solution that minimizes disruption of HPA axis function ( Eq ., 3 ) is shown in Figure 3 along with the systems overall trajectory ., Note that the optimal input does indeed bring the system to the healthy steady-state point ., This is done while maintaining a circulating cortisol concentration that is near the steady-state value with the exception of a rapid drop at the start of treatment ., The optimal control solution as computed under the MPC framework has several key features ., The cortisol concentration is rapidly dropped at the outset ., Once this drop in cortisol concentration is achieved , the system requires little additional control action to come to steady state ., This qualitative information can be used to formulate a suboptimal control strategy that will bring the system to the healthy steady state ., In this section a suboptimal control strategy is proposed for the HPA axis system ., The goal of this strategy is to mimic the qualitative results of the MPC solution while being realizable in a clinical setting ., The MPC solution suggests that manipulating cortisol concentration is a plausible strategy for redirecting the HPA axis to a healthy steady state ., The key difficulty in applying this approach is determining when the cortisol concentration has been sufficiently lowered with regard to the other state variables to allow the system to return to a healthy equilibrium ., That is , one must identify an observable event ( corresponding to a measurable variable ) which signals that the steady state of the system has shifted ., In a clinical setting only ACTH and cortisol concentrations , corresponding to x2 and x4 , respectively , can be readily measured ., The availability of cortisol analogues makes it possible to manipulate x4 directly ., Therefore as postulated previously ( Eq . 3–4 ) ACTH ( x2 ) can be used to determine when a change in available steady state or attractor has occurred ., Under the MPC framework , most of the control action is expended near the initial time ., In Figure 3 the external control action prescribed by MPC under ideal conditions and the response of ACTH ( x2 ) are both plotted as a function of time ., The value of x2 increases by about 30% as the system moves from the cusp of multiple candidate steady states to the basin of a single steady state ., The following treatment is therefore proposed: Treatment 1\u2003The cortisol concentration in the system should be slowly decreased until ACTH levels ( x2 ) have increased by more than 30% relative to the initial condition ., Once this signal is observed , the systems own natural feedback control action should restore cortisol levels to normal ., Simulation results for Treatment 1 are shown in Figure 4 ., As indicated the system is brought to the healthy steady state via the suboptimal but more realistic treatment course ., Furthermore , the drop in cortisol concentration is neither as severe nor as sharp as under naïve idealized MPC control ., A positive outcome may also be obtained by applying even less severe levels of cortisol suppression and extending the duration of the treatment ., Data presented in Figure 5 show that a combinations of treatment duration and cortisol suppression may be varied successfully over a large range ., Nonetheless there exists a minimum level of cortisol suppression below which the treatment fails regardless of how long conditions are maintained ., Conversely there also exists a minimal treatment duration below which even severe levels of cortisol suppression will prove unsuccessful in restoring normal hormone levels ., The results for Treatment 1 shown in Figure 4 are computed under nominal conditions ., For the proposed treatment to be clinically useful , it must be effective over a wide variety of conditions , and parameter values ., The robustness of the proposed approach to changes in the parameter values , initial conditions , and ambient stress level ( i . e . , value of d0 ) is examined in this section ., A direct computational evaluation of robustness of Treatment 1 is difficult to implement ., There are four initial conditions ( x1 ( 0 ) ; x2 ( 0 ) ; x3 ( 0 ) ; x4 ( 0 ) ) , seven parameters , and one disturbance variable ( d0 ) ., A simulation study where each variable ( initial condition , parameter and disturbance ) is evaluated at a nominal , high and low values , would require , at a minimum 312\u200a=\u200a531 , 441 simulations ., Even if these simulations were completed , the choice of high , low and nominal value for each variable would be difficult to justify using available data ., An alternative approach analyzing robustness analysis is to study the asymptotic behavior of System Hu ., Let the concentration of cortisol ( x4 ) be manipulated so that the product of cortisiol and GR concentrations ( x3x4 ) is constant ., Under these conditions , the asymptotic value of glucocorticoid receptor concentration GR ( x3 ) is obtained from Eq ., 7 as: ( 13 ) The asymptotic value or GR concentration x3∞ has a minimum as a function of cortisol concentration ( x4 ) at x4\u200a=\u200a0 ., That is , if one were to lower the cortisol concentration to zero one would obtain the lowest possible steady-state value for GR and this value would be: ( 14 ) At the steady-state point given by x4∞\u200a=\u200a0 and x3∞\u200a=\u200akcr/krd the unique asymptotic solution for CRH ( x1 ) and ACTH ( x2 ) is given by ( 15 ) It should be noted that the values for GR , CRH and ACTH identified in Equations 14–15 represent an asymptotic minimum for the externally controlled system ( System Hu ) ., For the closed loop HPA axis system it represents the minimum achievable cortisol concentration ., Note that this equilibrium point is only achievable under external input ., This result is independent of the trajectory of the input u ( t ) and is a property of the HPA axis ., Moreover the solution in Equation 15 is unique indicating that only a single steady state exists at the minimum for x4→0 and that this state corresponds to a stable set of non-zero real-valued concentrations of CRH , ACTH and GR ., This result confirms that reducing the cortisol concentration to a small enough positive value can indeed take the system to a single stable condition ., This is regardless of the value of d0 , parameters , or initial conditions ., This condition will correspond to a healthy equilibrium value when treatment is administered in the absence of elevated levels of external stress d0 ., At high levels of the stressor d0 the success of the treatment would be short lived as we would simply be immediately re-administering the same insult originally responsible for the illness state ., This is true regardless of whether the idealized or the suboptimal treatment approach is used ., In conclusion we have demonstrated in this work the use of model-based predictive control methodology in the estimation of robust treatment courses for displacing the HPA axis from an abnormal hypocortisol steady state back to a normal function ., Using this approach on a numerical model of the HPA axis proposed by Gupta et al . 10 a candidate treatment that displays robust properties in the face of significant biological variability and measurement uncertainty requires that cortisol be suppressed for a short period until ACTH levels exceed 30% of baseline ., At this point the treatment may be discontinued and the HPA axis will progress to a stable attractor defined by normal hormone profiles ., The concentration of biologically available cortisol could in principle be altered by binding proteins or metabolizing enzymes to inhibit negative feedback to the HPA axis without affecting the synthesis and accumulation of ACTH ., Our analysis shows that this treatment strategy is robust and that a positive outcome can be obtained reliably for a wide range of treatment efficiencies .

Introduction, Methods, Results, Discussion

The hypothalamic-pituitary-adrenal ( HPA ) axis is a major system maintaining body homeostasis by regulating the neuroendocrine and sympathetic nervous systems as well modulating immune function ., Recent work has shown that the complex dynamics of this system accommodate several stable steady states , one of which corresponds to the hypocortisol state observed in patients with chronic fatigue syndrome ( CFS ) ., At present these dynamics are not formally considered in the development of treatment strategies ., Here we use model-based predictive control ( MPC ) methodology to estimate robust treatment courses for displacing the HPA axis from an abnormal hypocortisol steady state back to a healthy cortisol level ., This approach was applied to a recent model of HPA axis dynamics incorporating glucocorticoid receptor kinetics ., A candidate treatment that displays robust properties in the face of significant biological variability and measurement uncertainty requires that cortisol be further suppressed for a short period until adrenocorticotropic hormone levels exceed 30% of baseline ., Treatment may then be discontinued , and the HPA axis will naturally progress to a stable attractor defined by normal hormone levels ., Suppression of biologically available cortisol may be achieved through the use of binding proteins such as CBG and certain metabolizing enzymes , thus offering possible avenues for deployment in a clinical setting ., Treatment strategies can therefore be designed that maximally exploit system dynamics to provide a robust response to treatment and ensure a positive outcome over a wide range of conditions ., Perhaps most importantly , a treatment course involving further reduction in cortisol , even transient , is quite counterintuitive and challenges the conventional strategy of supplementing cortisol levels , an approach based on steady-state reasoning .

The hypothalamic-pituitary-adrenal ( HPA ) axis is one of the bodys major control systems helping to regulate functions ranging from digestion to immune response to metabolism ., Dysregulation of the HPA axis is associated with a number of neuroimmune disorders including chronic fatigue syndrome ( CFS ) , depression , Gulf War illness ( GWI ) , and posttraumatic stress disorder ( PTSD ) ., Objective diagnosis and targeted treatments of these disorders have proven challenging because they present no obvious lesion ., However , the bodys various components do not work in isolation , and it is important to consider exactly how their interactions might be altered by disease ., Using a relatively simple mathematical description of the HPA axis , we show how the complex dynamical behavior of this system will readily accommodate multiple stable resting states , some of which may correspond to chronic loss of function ., We propose that a well-directed push given at the right moment may encourage the axis to reset under its own volition ., We use model-based predictive control theory to compute such a push ., The result is counterintuitive and challenges the conventional time-invariant approach to disease and therapy ., Indeed we demonstrate that in some cases it might be possible to exploit the natural dynamics of these physiological systems to stimulate recovery .

diabetes and endocrinology/neuroendocrinology and pituitary, mathematics, neurological disorders/neuroendocrinology and pituitary, computational biology/signaling networks, computational biology/systems biology

journal.pcbi.1002767

The Role of Exposure History on HIV Acquisition: Insights from Repeated Low-dose Challenge Studies

Before being tested clinically , vaccines or preventive treatment strategies against human-immunodeficiency virus ( HIV ) are assessed in non-human primates ., The paradigm of how vaccine candidates are assessed preclinically has been shifting in recent years ., A decade ago , vaccine protection was not quantified directly by determining how much the vaccine candidate reduced the susceptibility of the animal hosts to infection ., Instead , indirect measures were used , such as the level of virus-specific immune responses , or as the reduction of the viral set-point induced by vaccination ., Unfortunately , there are many uncertainties about how these immunological and virological measures correlate with protection ., As a consequence , vaccines have recently been tested using repeated low-dose challenge experiments 1–13 ., In such experiments animals are challenged repeatedly with doses of the virus , which do not give rise to infection with certainty ., This protocol not just more realistically reflects the repeated exposure to HIV that human hosts face in the epidemic , but also allows the experimenter to directly measure the reduction of the hosts susceptibility induced by the vaccine ., Thus , repeated low dose challenge experiments are conceptually closer to clinical studies 14 ., The challenge data generated in such trials are usually analyzed to infer the efficacy of a vaccine candidate , or post-exposure prophylaxis ., In addition to information on treatment efficacy , however , these data contain information on other , very relevant aspects of the transmission of HIV ., In the present study , I analyzed challenge data that had been published previously ., I focused on the challenge data from control animals that did not receive any vaccine or treatment ., My first main question was if hosts are immunized by repeated challenges ., In SIV challenge experiments , potential immunization is usually studied by measuring systemic or localized immune responses in an animal after a non-infecting challenge ., This approach relied on the strong assumption that these immune measures are causatively linked to protection ., To my knowledge , however , no such link has been systematically ascertained for SIV infection to date ., In this paper , I adopted an alternative approach to assess if immunization has occurred after challenge: I essentially compared the susceptibility of animals before and after challenge ., Therefore , throughout this paper , immunization denotes a reduction of susceptibility that is brought about by non-infecting challenges ., This definition of immunization does not require to know and measure the immune effector conferring protection , and directly quantifies what matters epidemiologically ., The second main question I posed was if there are differences in susceptibilities between animals ., Again my approach focused on differences of animals in terms of their susceptibility to infection , and did not involve the quantification of target cells or their susceptibility in relevant anatomical sites ., As is common in mathematical epidemiology , I conceptualized infection as a stochastic event that occurs with a given probability ., More precisely , I described infection as a Bernoulli process ., While the infection probability can be thought of as a trait of an individual animal , its estimation requires data of more than one animal ., Repeated low-dose challenge data allow us to estimate the infection probability ., They also allow to test if animals are immunized by non-infecting challenges , as immunization leads to a smaller and smaller fraction of animals becoming infected in the course of the challenge experiment ., The same is true if there are differences in susceptibility between animals ., As I show below , one typically finds evidence for both , immunization and susceptibility differences , and further analysis is needed to disentangle the two effects ., Formally , I used simple stochastic models and maximum likelihood estimation to estimate the infection probability , and to study how it varies across challenge repeats and animals ., I found that there is no evidence for immunization in any of the studies ., There is also no evidence for variation in susceptibility , except for one recent experiment conducted with the strain SIVsmE660 13 ., In that study , genetic differences in susceptibility to SIVsmE660 had been previously established ., Taken together , these results show that one of the central assumptions of the repeated low-dose challenge approach is not violated: there is no evidence that challenge history affects the probability of infection ., The findings also have implications for our understanding of the role of repeated exposures in HIV acquisition ., First , I assessed if there is evidence in the repeated low-dose challenge data for immunization in the sense that challenges , which do not give rise to infection , reduce suceptibility of the host ., To this end , I first fit a stochastic model ( the geometric infection model ) , which assumes that each animal has the same susceptibility to infection , , and that this susceptibility does not change from challenge to challenge ( see see Figure 2A and Materials and Methods ) ., This model served as a null model against which the more complex models are compared ., The fit of the geometric infection model is summarized in Table 2 , and a plot of the likelihoods as a function of can be found in Figure 3 ., The estimated infection probability , , of each individual study ranges from 0 . 16 to 0 . 25 ( see Table 2 ) , except for the SIVmac251 challenge data in Letvin et al 2011 , for which the estimate of ., In a second step , I fit the immunization model to the data ., This model assumed that the susceptibility to the challenge decreased with challenge repeats ( see Figure 2B ) ., I considered various ways in which such a decrease could occur ., The susceptibility could drop after the first challenge from a value , , to a lower value , ., Alternatively , it could drop to this lower level after the th challenge , rather than the first ., Lastly , the susceptibility may start at a value , and change incrementally by a fixed amount ( see Materials and Methods ) ., While these models certainly do not comprise every conceivable immunizing effect they serve as a good compromise between what is conceivable immunologically and the mathematical simplicity of their description ., Without this simplicity one would lose statistical power ., Irrespective of the way I implemented immunization , the immune priming model fails to outperform the geometric infection model statistically , except for the SIVsmE660 challenge data of Letvin et al 2011 ( in which susceptibility difference between the monkeys have been established ) , and Wilson et al 2006 ., Table 3 shows the maximum log-likelihoods , , for the immune priming model in its various formulations , along with maximum likelihood estimator for the parameters and , or and for the incremental immune priming model variant ., The likelihoods of this model are smooth and have clearly defined global maxima ( see Figure 4 ) ., Table 3 also shows the -value of a likelihood ratio test against the geometric infection model ., In most cases , none of the immune priming model variants explain the data better than the geometric infection model ., A notable exception are the data by Wilson et al , J Virol 2006 ., 6 ., It is important to note , however , that the effect in these data is the opposite of immunization: the susceptibility increases with challenges ., This result is due to the fact that none of the animals in the experiments by Wilson et al , J Virol 2006 6 became infected at the first or second challenge ., Therefore , the susceptibility at first and second challenge is estimated as for the models with , and the incremental variant of the immune priming model is estimated to be positive ., I also found a significant improvement of the fit of the immune priming models over the geometric infection model for the SIVsmE660 challenge data of Letvin et al 2011 ., In particular , the immune priming model with an approximately two-fold drop in susceptibility after the second or third challenge ( or ) improved the fit significantly over the geometric infection model ( ) ., However , this can be attributed to the susceptibility differences between the monkeys in that dataset ., Repeating the analysis on the subgroups carrying permissive and restrictive TRIM5 alleles separately , yielded no evidence for immunization ( see Table 3 ) ., In summary , there is no evidence for immunization by viral challenges in these datasets ., To assess if there is any evidence for differences in susceptibility between animals , I followed the same statistical approach as in the previous subsection: I compared the fit of the geometric infection model to that of a model , in which the susceptibilities are allowed to vary from animal to animal ( the heterogeneous susceptibility model ) ., The heterogeneous susceptibility model is mathematically defined in the Materials and Methods section and diagrammatically shown in Figure 2C ., This model has two parameters , one for the mean infection probability , , and another measuring the variance in susceptibilities across animals , ., The likelihoods as a function of the two parameters of this model are shown in Figure 5 for each of the datasets analyzed ., Table 4 shows the maximum log-likelihood , , the maximum likelihood estimators , and , and the -values for a likelihood ratio test against the geometric infection model fit ( see Table 2 ) for the heterogeneous susceptibility model ., For the dataset for which susceptibility difference have been established ( Letvin11stm . SIVsmE660 ) , I found significant levels of inter-animal susceptibility differences ( ) ., This shows that the statistical approach I adopted works ., It further shows that susceptibility differences can be established without having to know their molecular or genetic basis ., ( My analysis did not use information on the TRIM5 alleles that the animals carried . ), For none of the other datasets does the heterogeneous susceptibility model fit better than the geometric infection model , and hence there is no evidence for variability of susceptibilities between animals ., The absence of evidence must not be confused with the evidence of absence ., The non-significant results in the previous two subsections could simply be due to low sample sizes ., To address this possibility , I conducted a power analysis ., I simulated experiments using the same number of animals and challenges as in the experimental data , assuming immunization effects or inter-animal susceptibility differences of various sizes ., I then analyzed these simulated data to test for immunization or heterogeneous susceptibility ( see Materials and Methods for more detail ) ., For the immunization model , I defined the effect size as the relative reduction of susceptibility after the first challenge , ., Figure 6 shows the result of this analysis ., The least powerful experiment is that by Wilson et al , J Virol 2006 6 because it involves only eight control animals and at most eight challenges ., The most powerful experiments are those conducted with SIVmac251 by 13 and those by 15 ., The first experiment involved 20 animals challenged at most 12 times , the second 28 animals challenged at most 25 times ., For these experiments , the probability not to uncover a significant immunization effect is less than 5% for an effect size of , i . e . if the susceptibility was reduced by a factor of approximately 4 . 5 after the first unsuccessful challenge ., A four-fold reduction still constitutes a large immunization effect ., How likely is it that I missed a significant effect in all of the studies simultaneously ?, A power analysis , in which I simulated each study repeatedly assuming study-specific model parameters ( see Materials and Methods ) , yielded that the probability to miss an immunization effect in all these studies of size is less than 5% ., This analysis is valid only if an immunization effect is present and equal across all the early studies ., For the heterogeneous susceptibility model , the effect size was defined as the variance of the susceptibility distribution ., Depending on the variance , the shape of the susceptibility distribution can be hump-shaped , monotonously falling ( or rising ) , or U-shaped ., The maximum variance depends on the mean of the distribution ., For example , for a mean infection probability of 0 . 2 — the most common estimate obtained by fitting the geometric infection model to the various datasets — this maximum is 0 . 16 ., For Letvin11stm . SIVmac251 , however , the mean infection probability is 0 . 5 , and the maximum possible variance is 0 . 25 ., Figure 7A shows the result of a power analysis for various levels of heterogeneity in animal susceptibility ., The power to establish susceptbility differences between animals differs for each individual study ., The dataset by Hansen et al 2011 has the highest power , and can be used to exclude a level of heterogeneity ., The shape of this critical susceptibility distribution with is shown in Figure 7B ., The critical describes a monotonously falling susceptibility distribution with large differences in heterogeneity between animals ., For this dataset , one can exclude only the largest conceivable heterogeneity: a U-shaped susceptibility distribution describing a scenario according to which approximately one fifth of the animals are almost completely susceptible and the remaining animals are alomost completely resistant to infection ., Again , one can ask how probable it is that we missed a significant effect in all of the studies simultaneously ., The probability not to detect susceptbility differences in any of the early studies is lower than 5% for susceptibility differences larger than ., This analysis assumed equal levels of heterogeneity across the different studies ., Using challenge data that were generated in the context of preclinical HIV vaccine studies in non-human primates , I investigated if low-dose challenges immunize the animal hosts ., Potential immunization has been raised as an argument against the repeated low-dose challenge approach , which could impair its statistical power advantage ., I also studied if there is evidence for susceptibility differences between animals ., Formally , the analysis involved fitting simple stochastic models to the challenge data ., To establish immunization or heterogeneity in susceptibility , the fits of models that accounted for such effects were compared statistically to fits of a model that ignored them ., For none of the datasets , I found evidence for immunization ., There is also no evidence for differences in susceptibilities , except in the SIVsmE660 challenge data presented in 13 , in which susceptibility differences have been previously identified ., For SIVsmE660 it had been established that an animals susceptibility depends on the TRIM5 alleles it carries ., TRIM5 encodes for the restriction factor Trim5 , which is thought to interact with the capsid of the virus after it has infected a cell ., Letvin et al show that monkeys that carry exclusively TRIM5 alleles classified as “restrictive” have a reduced susceptibility to infection as compared to monkeys that carry “permissive” alleles ., Comparing the fit of the geometric infection model to that of the heterogeneous susceptibility models , I found significant levels of heterogeneity in these challenge data ., It is important to emphasize that the evidence for susceptibility differences in this dataset is not based on information of the TRIM5 alleles the animals carry ., The inference only relies on the distribution of the number of challenges across animals ., Hence , the method I am presenting allows the identification of heterogeneity in susceptibility from the challenge data alone and does not rely on measuring traits that modulate susceptibility ., Such a factor ignorant method is important tool as considerable uncertainties about the determinants of susceptibility remain ., The SIVsmE660 challenge data presented in 13 are also consistent with an immunization effect , although not for every type of immune priming I considered ., This finding , however , is very likely an artefact of the statistical approach known as one of “hetergeneitys ruses” 16 ., Both effects — immunization and susceptibility differences — manifest themselves by an overdispersion of the challenge data as compared to a geometric distribution ., Immunization decreases the susceptibility to late challenges due to the immune responses elicited by early challenges ., A similar pattern arises if susceptibilities among the animals differ , but for a different reason ., In this case , early challenges will more likely lead to infection of animals with higher susceptibility ., This results in more animals with lower susceptibility late in the challenge schedule as these animals are more likely to remain uninfected ., An immunization effect could therefore be incorrectly inferred from challenge data that arise from hosts that vary in their susceptibility ., In general it is difficult to disentangle the two effects ., The difference between immunization and host heterogeneity is too subtle to be detected with the sample sizes of the challenge data I analyzed , and depends sensitively on the quantitative details of immunization effects and heterogeneity ., However , as the SIVsmE660-challenged animals of the study by Letvin et al had been classified with respect to their susceptibility , I could test for immunization within these subgroups ., As I did not find any evidence for immunization in each susceptibility class , I concluded that the immunization effect in the pooled data is misidentified ., The lack of evidence for immunization does , of course , not prove that there is no such effect ., It may simply result from the low sample sizes in these studies ., To go beyond this plain caveat , I conducted a power analysis that quantifies the probability that an effect was missed ., The study by Wilson et al ( 2009 ) may provide a likely case of too low power ., This study used the same challenge strain ( SIVsmE660 ) as the data by Letvin et al 2011 , in which susceptibility differences between animals have been established ., The animals involved in study by Wilson et al ( 2009 ) were , to my knowledge , not monitored for their TRIM5 alleles , but it is conceivable that some animals differed in their susceptibility for this reason ., The power of this study , however , was the lowest among all the studies ., To detect the same level of heterogeneity as I found in Letvin11stm . SIVsmE660 ( ) the power of the study by Wilson is below 10% ., While sample sizes were clearly an issue in the study by Wilson et al ( 2009 ) , especially the later studies 13 , 15 involved substantially larger numbers of animals ., According to the power analysis , these studies allow the detection of immunization effect , albeit only large ones ., Additionally , it is important to note that my power estimates are optimistic as the simulated data for the power analysis were generated with the same model as was used for the statistical analysis ., The model describing the true immunization effect is likely to be different from the immune priming model , and this model misspecification will generally lead to lower power ., The lack of evidence for immunization by non-infecting challenges in the majority of the studies constitutes a crucial validation of the repeated low-dose challenge approach ., Only if challenges do not immunize , one can safely assume that infection probabilities are independent ., According to my analysis , there is no evidence against the assumption of independence ., While we cannot exclude immunization effects of small size , the analysis presented in this paper provides evidence against at least very strong immunization effects ., This suggests that the repeated low-dose challenge approach increases statistical power as we and others have previously predicted 14 , 17 ., Independently from the findings I present in this paper , the statistical power is further corroborated by the increasing number of studies that have used this approach successfully ., For example , Ellenberger et al , Virol 2006 5 established an efficacy of 64% of a DNA/MVA vaccine with 30 animals , and Garcia-Lerma et al , PLoS Med 2008 8 could establish a treatment efficacy of 74–87% of pre-exposure prophylaxis with antivirals with 42 animals ( see Table 1 ) ., To harvest the full power of the repeated low-dose approach it is further necessary to keep the time between challenges large enough to allow the identification of the challenge which gave rise to infection ., Identification of the infecting challenge may have been the problem in the study by Wilson et al , J Virol 2006 6 in which no animal was infected before the third challenge ., The power analysis also suggests that the susceptibility distribution among the experimental animals is , with high probability , not U-shaped ., This is also very relevant to how vaccine efficacies are estimated statistically , and how many animals have to be involved in a preclinical study ., If the susceptibility distribution were U-shaped , the animal population would essentially fall into two classes: almost completely susceptible and almost completely resistant ., Any effect of a vaccine would be confined to the susceptible subpopulation , thus effectively decreasing the sample size ., But even in the case in which the susceptibility distribution is not U-shaped , yet susceptibilities still vary from animal to animal , some of the standard assumptions made when estimating vaccine efficacies from repeated low-dose challenge experiments are violated ., While some studies consider animal-to-animal variation in the effect of the vaccine 18 , the susceptibility of unvaccinated animals is most commonly assumed not to differ across animals ., In future repeated low-dose challenge trials , I suggest to first check if there is evidence for susceptibility differences using the statistical approach presented in this paper ., If this turned out to be the case frailty approaches should be adopted along the lines of 19 to estimate vaccine efficacies ., Beyond the context of assessing HIV vaccines or prophylaxis , repeated low-dose challenge data provide insights into the natural transmission of HIV ., It is extra-ordinarily challenging to assess how the rate of HIV acquisition depends on the exposure history ., The reason for this difficulty is that , on logistic grounds , individuals cannot be monitored frequently enough to generate exposure and acquisition data with the level of detail required to establish the role of exposure history ., For example , the Rakai cohort 20 , 21 , which provides some of the best data to estimate HIV transmission rates , involves approximately 200 HIV discordant couples ., Individuals in this cohort are monitored every 10 months , and report on average 10 sex acts per month ., Thus , individuals are tested for infection every 100 exposures on average , which does not allow to estimate reductions in susceptibility that are likely to be most pronounced during the first exposures to the virus ., For these reasons there is no quantitative understanding of the dependence of the rate of HIV acquisition on exposure history to date ., Consequently , most mathematical models that forecast the epidemiological spread of HIV neglect exposure history and assume that hosts retain no memory of previous exposures ., The findings in this paper provide limited support for this assumption ., The support is only limited because of issues relating to statistical power mentioned above , but also because the doses used in repeated low-dose challenge experiments are still much higher than those transmitted naturally ., To definitively rule out any impact of exposure history it will be necessary to conduct experiments in which the challenge dose is further reduced and the frequency is systematically varied from more often than daily to less often than weekly ., Some immunization effect may not be detectable if hosts are exposed weekly , as was done in most of the studies I analyzed in the present paper ., There is a group of HIV exposed individuals — sex workers from Kenya and Uganda — who remain uninfected despite frequent exposure to the virus ., These highly exposed seronegative ( HESN ) individuals are hypothesized to be immunized by frequent exposures to the virus 22 ., It has been realized fifteen years ago that non-human primate models may provide a way to test for potential resistance due to exposure to the virus ., However , early studies of this issue remained equivocal 23 , 24 ., These early studies on the role of exposure history also employed very high doses to challenge the monkeys ., At these high challenge doses the experiments may not have been sensitive enough to detect resistance mechanisms that protect against naturally-occurring low-dose exposure ., The repeated low-dose challenge of monkeys much better reflects the frequent exposure of the HESN individuals , although the doses used in repeated low-dose challenge experiments are still high when compared to the doses to which humans are exposed ., ( They are termed “low” to distinguish them from the very high doses normally used in non-human primate challenge studies . ), Therefore , if frequent exposure by itself were sufficient to lead to resistance , at least partial immunization should be observed in the challenge experiments ., The fact that I failed to find any immunization effect suggests that there is more to the resistance of HESN individuals than high and frequent exposure ., It is conceivable that the exposure frequency or dose is required to start at a low level and increase over time ., The exposure route may also be relevant: in the studies I analyzed the challenge was performed rectally , while HESN individuals are exposed vaginally ., A last possibility is that the frequency of challenges in the most of the experiments of one week is too low to kick off the immunizing mechanism , which render HESN individuals resistant ., In any case , the hypotheses about resistance in HESN individuals will have to be refined by specifying the routes of infection as well as the ranges of exposure dose and frequency that can lead to resistance ., An important conclusion from the analysis presented in this paper is that the challenge data in every study — with the exception of the one by Letvin et al using SIVsmE660 as a challenge strain — are consistent with the geometric infection model ., This means there is no evidence that animals differ in their susceptibilities ., As a consequence , there is no justification to divide the animals in these studies into those that become infected early versus those become infected late in the challenge schedule ., Neither is there any justification to compare these two groups immunologically , virologically or genetically ., Approaches , such as the statistical comparison between the fit of geometric infection model with a fit of the heterogeneous susceptibility model , are required to establish susceptibility differences and to provide a solid statistical foundation for comparisons between animals with low and high susceptibility ., I selected repeated low-dose challenge data from seven previously published studies ., These studies are: Ellenberger et al , Virol 2006; 5 , Wilson et al , J Virol 2006; 6 , Wilson et al , J Virol 2009; 10 , Garcia-Lerma et al , PLoS Med 2008; 8 , Hansen et al , Nat Med 2009; 9 , and Hansen et al , Nature 2011; 15 , and Letvin et al , Sci Transl Med 2011 13 ., The criteria for this selection were a sufficiently high number of monkeys involved , more than five challenge repeats applied , and the regularity of the challenge schedule ., Table 1 summarizes the most important aspects of the data ., The monkey hosts were rhesus or pigtailed macaques , and the challenge virus was either the standard challenge strains SIVmac239 or SIVmac251 the sooty mangabey virus SIVsmE660 , or SHIV-SF162P3 , a chimera between SIV and HIV featuring a CCR5 tropic envelope protein 25 ., The number of animals involved in the studies ranged from 16 to 86 ., The maximum number of challenges ranges from 8 to 26 ., Challenges were given rectally with a frequency of one week ., Rectal challenges are the preferred route in such experiments as they can be performed on male animals and are relevant for human transmission ., The involvement of female animals in preclinical studies is rare as they are required to maintain the colonies ., I analyzed only challenge data of the control animals involved in the studies listed in Table 1 ., In these animals the susceptibility was not manipulated , and is thus most relevant to study the effects of exposure history ., In some studies the challenge dose was increased after a certain number of challenges ., I ignored the data generated with increased doses ., The reason for this is that the challenge with increased doses pertained to only few animals , and would therefore be only marginally informative ., Moreover , incorporating these data would have forced me to introduce an additional susceptibility parameter into my models , which — due to the low sample size — could not be reliably estimated ., The dataset by Letvin et al ( 2011 ) involving challenges with the viral strain SIVsmE660 will serve as a control for our approach to establishing susceptibility differences ., For this strain , genetic correlates of susceptibility have been identified ( see Results and Discussion ) ., The challenge data consist of two pieces of information for each animal ., The first is the number of challenges , , the th animal received , and the second is the infection status of this animal after the challenges ., Hereby , means that the animal remained uninfected , and means that the animal was infected after receiving challenges ., Note that in these experiments an animal that is found infected is not given any challenges anymore ., I constructed stochastic models and used them in combination with the challenge data to infer parameters characterizing the probability of animals becoming infected upon challenge with the virus ., In the next subsection , I describe these models ., To test for immunization by repeated challenges or for differences in the susceptibilities of animals to infection , I first fit the geometric infection model , and then the immune priming and heterogeneous susceptibility models ., The model fits were then compared by a likelihood ratio test ., I applied a significance level of 0 . 05 ., To determine the statistical power of the model fitting and comparison , I simulated data that conform to the immune priming or heterogeneous susceptibility models ., In these simulations , I chose numbers of animals and maximum numbers of challenge repeats consistent with each experimental study ., In the case of the immune priming model , I set an animals susceptibility at the first challenge is for all studies except for Letvin11stm . SIVmac251 for which I set ., This probability was reduced after the first challenge by a factor ranging from 1 ( =\u200ano effect ) to 12 ., In the simulation according to the heterogeneous susceptibility model , I set the mean probability of infection of for all studies except for Letvin11stm . SIVmac251 for which I defined ., I further set the variance parameter to values ranging from 0 ( =\u200ano effect ) to 0 . 155 ., The value 0 . 16 is the maximum variance possible for a Beta distribution with mean 0 . 2 ., The simulated data were then analyzed and significance was assessed ., Power was determined as the fraction of simulated experiments , in which a significant immunization effect or heterogeneous susceptibilities could be established ., In accordance with the comparison of the model fits to the experim

Introduction, Results, Discussion, Materials and Methods

To assess the efficacy of HIV vaccine candidates or preventive treatment , many research groups have started to challenge monkeys repeatedly with low doses of the virus ., Such challenge data provide a unique opportunity to assess the importance of exposure history for the acquisition of the infection ., I developed stochastic models to analyze previously published challenge data ., In the mathematical models , I allowed for variation of the animals susceptibility to infection across challenge repeats , or across animals ., In none of the studies I analyzed , I found evidence for an immunizing effect of non-infecting challenges , and in most studies , there is no evidence for variation in the susceptibilities to the challenges across animals ., A notable exception was a challenge experiment by Letvin et al . Sci Translat Med ( 2011 ) conducted with the strain SIVsmE660 ., The challenge data of this experiment showed significant susceptibility variation from animal-to-animal , which is consistent with previously established genetic differences between the involved animals ., For the studies which did not show significant immunizing effects and susceptibility differences , I conducted a power analysis and could thus exclude a very strong immunization effect for some of the studies ., These findings validate the assumption that non-infecting challenges do not immunize an animal — an assumption that is central in the argument that repeated low-dose challenge experiments increase the statistical power of preclinical HIV vaccine trials ., They are also relevant for our understanding of the role of exposure history for HIV acquisition and forecasting the epidemiological spread of HIV .

Individuals are exposed to Human Immunodeficiency Virus ( HIV ) many times before they contract the virus ., It is not known what an instance of exposure , which does not result in infection , does to the host ., Frequent exposures to the virus are hypothesized to immunize an individual , and result in resistance to infection with HIV ., This hypothesis may explain the resistance observed in some individuals despite frequent exposure to the virus ., Since it is very difficult to monitor the HIV exposure and infection status of humans , this question is easier to address in animal models ., I took data from previously published infection experiments of monkeys with Simian Immunodeficiency Virus ( SIV ) and analyzed them with newly developed mathematical models ., I found that there is no evidence that challenging monkeys with the virus reduces their susceptibility to infection ., These findings have important repercussions for the testing of HIV vaccines in monkeys , and also for our understanding of the role of exposure history for the acquisition of HIV .

medicine, infectious diseases, immunology, biology, computational biology

journal.pcbi.1003375

Detecting Genetic Association of Common Human Facial Morphological Variation Using High Density 3D Image Registration

The human face plays an essential role in everyday life ., It hosts the most important sensory organs and acts as the central interface for expression , appearance , communication and mutual identification ., Inheritance of facial appearance from parents to their offspring is a constantly intriguing question to the public and scientific community ., Indeed , human facial morphology is highly heritable ., Twin studies have shown that heritability of facial features is as high as 80% 1 , 2 ., On the other hand , non-genetic factors also play important roles in shaping the human face , such as age and climate 2–6 ., High heritability suggests that ones facial characters might be predicted to a certain extent , as long as the genetic determinants are identified and their effects fully understood ., Face prediction based on genetic profiling , if feasible , may revolutionize forensics 7 and strongly benefit medical diagnosis 8 ., However , the influences of common genetic variants on facial morphogenesis are largely unknown ., The current understanding of facial morphogenesis has mainly arisen from developmental biology studies in model organisms ., Facial morphogenesis involves a coordinated growth of facial prominences in a precise temporal and spatial sequence , which is tightly regulated by many signaling pathways , including the BMP , SHH , FGF , GHR and Wnt/β-catenin pathways 9–16 ., In humans , knowledge of the effects of genetic variation on facial morphology has mainly arisen from studies of congenital craniofacial abnormalities ., Non-syndromic cleft lip with or without cleft palate ( NSCL/P ) is the most common congenital craniofacial defect 3 , 16 , 17 ., Great efforts have been made towards identifying the genetic factors that predispose carriers to NSCL/P , and a large number of candidate risk genes have been proposed 17–19 ., Among these , the IRF6 gene has shown the most convincing and consistent signals for association across many studies 17 , 20–24 ., Many other craniofacial abnormalities can also result from rare genetic disorders , such as Down syndrome , Rubinstein-Taybi syndrome , Sotos syndrome , Bardet-Biedl syndrome and Noonan syndrome 25–29 ., Nevertheless , these studies have mainly focused on pathological facial morphological changes ., Relatively few studies have attempted to associate genetic polymorphisms to common facial morphological variations ., Several non-synonymous changes in the growth hormone receptor ( GHR ) were suggested to affect mandible shape in Japanese and Chinese populations 30–32 ., Ermakov et al . found that a SNP in ENPP1 , a gene essential in bone physiology , was significantly associated with upper facial height in Chuvashians 33 ., In the FGFR1 gene , a genetic marker was found to be associated with the cephalic index in multiple populations 34 ., Interestingly , a recent study examined several high frequency SNPs associated with differential risks of NSCL/P in a few healthy cohorts , and found that two were associated with normal facial shape variation 6 ., This suggests that disease risk alleles may also modulate the phenotypes of unaffected carriers , although within a range of normal variation ., Subtle shape alteration patterns induced by disease risk alleles , if properly defined , may help to screen carriers of disease alleles , and therefore facilitate disease prevention ., In addition to these candidate gene studies , two genome wide association studies ( GWAS ) have also recently been carried out in Europeans , to search for genetic loci that influence common facial shape variation , and five loci were found to significantly modulate several nose related features 2 , 35 ., Anthropometric phenotypes , especially facial features , are highly complex and diverse ., Traditional phenotype collection involves the manual measurement of specific distances and angles directly on the specimen or subjects , which is infamously tedious and error prone ., In recent years , new imaging technologies , have been developed to allow fast and accurate acquisition of three dimensional facial landscapes without direct physical contact with the subject ., Such imaging technologies have greatly facilitated human evolutionary analyses of craniofacial phenotypes 4 , 5 , 35 , 36 , as well as genetic association studies of human facial morphological variations 2 , 6 , 35 ., However , the analysis post image acquisition still generally involves manual annotation of landmarks on digital images 4 , 5 , 35 , 36 ., More importantly , these inter-landmark distances were the most widely used phenotype measurements in the recent genetic studies of human facial morphology 2 , 6 , 33–35 ., Inter-landmark based approaches have several problems ., First , when pairwise distances are used as phenotypes , the number of phenotypes increases exponentially with that of landmarks , which often results in over conservative p values after multiple-testing correction ., Second , the information on shape changes that is conveyed by inter-landmark distances is usually obscure ., For example , an extended distance between the nasion and nose tip could signal either more pointed or overall bigger nose ., Third , the facial shape cannot be fully reconstructed based on pairwise distances and it is , therefore , hard to perceive the biological meaning of the variation in distances ., Thus , methods that directly examine the geometrical configuration of shapes are more desirable for general shape analyses ., Such methods involve superimposing sample shapes according to their landmarks , followed by multivariate analyses/tests based on landmark coordinates 37 ., More recently , new methods have also emerged to better use high resolution geometrical information ., Instead of using only the limited number of traditional landmarks , these methods establish high density correspondence for thousands of mathematical landmarks 83839 ., Based on such methods , rare genetic diseases could be precisely identified and the syndrome effects could be extracted , predicted and visualized in great detail 40–42 ., In this study , we first applied the method of high resolution 3D image registration to test the potential genetic associations of the complex normal facial variations , and to infer the detailed effects of genetic variants on face ., In brief , we applied high density face registration ( HDFR ) to capture the comprehensive facial variation information of ∼30 , 000 3D points ( referred to as marker points hereafter ) 39 ., Based on HDFR , three different schemes of phenotype representation were systematically compared for the detection of genetic associations with 10 candidate SNPs ., The first scheme used traditional inter-landmark distances; the second represented the face geometrical shapes based on 15 major landmarks; the third is the high density geometric approach that we first proposed in such kind of studies ., It uses the complete geometric data of over 30 , 000 marker points ., The high density geometric data was then further used to examine the detailed phenotype changes associated with candidate SNPs ., We reviewed the literature for candidate SNPs that may be involved in the morphogenesis of the human face . 10, SNPs from 4 genes , ENPP1 , GHR , FGFR1 and IRF6 were identified and their functional relevance was listed ( Table 1 ) ., The ectonucleotide pyrophosphatase/phosphodiesterase 1 ( ENPP1 ) gene is a key regulator of bone mineralization ., Ermakov et al found that the upstream promoter and 3′ un-translated regions in this gene harbor genetic variants associated with the upper facial height and other phenotypes 33 ., Four SNPs , rs7773292 , rs6925433 , rs6569759 , rs7754561 that carry the strongest association signals were added to our candidate list ., GHR is the growth hormone receptor , which plays essential role in the development ., Mutations in this gene induce idiopathic short stature and Laron syndrome , marked by a characteristic facial appearance 31 ., Several non-synonymous SNPs , including Pro561Thr ( rs6184 ) , I526L ( rs6180 ) and C422F ( rs6182 ) were suggested to contribute to mandibular measures in East Asian populations 15 , 31 , 32 ., In this study , we included rs6180 and rs6184 in our study , as they were validated in Han Chinese 32 ., FGFR1 , the fibroblast growth factor receptor 1 plays an important role in facial morphogenesis , and mutations in this gene lead to syndomes associated with facial abnormality , such as the type 1 Pfeiffer syndrome ( MIM 101600 ) and Kallmann syndrome 2 ( KAL2 ) ( MIM 147950 ) 34 ., A tagging SNP of this gene , rs4647905 showed moderate signals of association with cephalic index in multiple ethnic groups 43 ., We added another tagging SNP rs3213849 to span the full length of FGFR1 ., The Interferon regulatory factor 6 ( IRF6 ) plays a critical role in keratinocyte development ., Genetic variants of IRF6 , especially SNP rs642961 , were found consistently associated to NSCL/P throughout many candidate gene and GWAS studies 17 , 23 , 44 , 45 As the genetic risk factors of NSCL/P may also contribute to normal facial variation in healthy carriers 16 , we enrolled rs642961 into our study ., We further included the SNP rs2236907 of IRF6 , which seems to have a particularly strong signal in Han Chinese 23 , 46 ., The genetic effects of these SNPs were examined in a Han Chinese population from Taizhou , Jiangsu province on the east coast of China ., The complete work flow is summarized in figure 1 ., In total 1001 self-reported Han Chinese individuals were enrolled in the analyses ( 604 females and 397 males ) , with an age range of 17∼25 years ., Saliva was collected to obtain DNA ., For the phenotype data , we collected high resolution 3D facial images for each individual ., Facial images were automatically annotated with 15 salient landmarks ( see Fig . 2 for the full list of the landmarks ) , using a novel landmark recognition method ( see Methods ) 39 ., This was followed by HDFR that resulted in 32 , 251 mathematically derived marker points , which were corresponded one to one across all individuals ( see Methods ) 39 ., The facial shape phenotypes were represented with three different schemes ., In the first scheme , the Euclidean distances between pairs of the landmarks were taken as phenotypes , and hereafter collectively referred to as the landmark-distance ( LMD ) data ., In the second scheme , the 15 landmarks of different individuals were first superimposed into a common coordinate system , by partial general procrustes analysis ( PGPA ) ( see methods ) 37 ., PGPA removes the differences in location and rotation , while keeping the size and shape information ., The coordinates of the aligned landmarks were thus used as the second type of phenotypes , hereafter referred to as landmark-geometric data ( LMG ) ., In the third scheme , all the 32 , 251 marker points were used to describe the phenotypes ., The marker points were similarly superimposed onto a common 3D space using PGPA , and the coordinate vectors specified a phenotype data space of 32 , 251×3\u200a=\u200a96 , 753 dimensions ., This data is hereafter referred to as dense-geometric ( DG ) data ., As sampling was carried out during winter time , many 3D images were affected by the participants collared clothing , especially around the upper neck and lower jaw area ., Furthermore , heavy facial hair in males caused defects and holes in some surface meshes ., During quality control , the images with obvious caveats were removed from further analysis ( 105 individuals ) ., 40 individuals were further removed due to the poor DNA quality ( uv light absorption A260/280<1 . 6 or total DNA quantity lower than 300 ng ) ., In the end , 856 individuals were successfully processed for their 3D images and have corresponding DNA ., We carried out the genetic association study in two stages ., The individuals of the original cohort were randomly assigned to 2 panels: panel I included 376 individuals ( 168 males and 208 females ) , and panel II included 480 individuals ( 174 males , 306 females ) ., Tests were carried out separately for different genders ., In stage I , all 10 candidate SNPs were genotyped for panel I . Then in stage II , the markers that showed preliminary evidence of correlation were validated using panel II ., For stage I analysis , individuals were assigned into 3 possible genotype groups for each SNP ., None of these SNPs deviated significantly from the Hardy-Weinberg equilibrium ., For the LMD data , the landmark-distances were tested for association with SNP genotypes using the Tukeys honestly significant difference test ( Tukeys HSD test ) ., Of the total 105 possible pairwise distances , 6 departure from normal distribution according to Shapiro-Wilk normality test ., As normality is required in Tukeys HSD test , these phenotypes were removed from further analysis ., For the remaining 99 phenotypes , the raw p values were calculated and corrected for multiple-testing with 10 , 000 permutations ( see Methods ) ., Table 2 shows the summary of the preliminary association signals ., Several SNPs demonstrated some preliminary association signals in terms of nominal Tukey test p value ( p value<0 . 01 ) ( Table 2 ) ., In particular SNPs rs642961and rs6184 , showed enriched signals ( Table 2 ) ., For SNP rs642961 , many signals appeared in females between the mutant ( TT ) and the other two groups CC and CT ., Interestingly , the strongest signals seemed to all point to the area around mouth and lower nose area ., The distances between the right/left lip corners and the right/left alare ( RLipCn – RAla and LLipCn – LAla ) had nominal Tukey test p values between 0 . 002∼0 . 004 in both the CC/TT and CT/TT comparisons ( Table 2 ) ., The distance between the upper lip point and lower lip point ( ULipP-LLipP ) also suggested potential shape difference between the CC and TT groups ( nominal p value\u200a=\u200a0 . 0023 , Table 2 ) ., The suggestive involvement of this SNP with mouth shape is consistent with the known role the host gene IRF6 plays in NSCL/P 17 , 23 , 44 , 45 ., SNP rs6180 and rs6184 both showed some signals in males , which seemed to mainly involve the two lip corners and their relative positions to the middle line landmarks such as Pronasale , Nasion , Subnasale , lower lip point and chin ( Table 2 ) ., These phenotypes may suggest size differences in the lower face among different genotype groups , but the overall trend is not clear ., However , after the permutation correction of the multiple testing , none of these phenotypes stood significant at the individual SNP level , before accounting for multiple SNPs and different genders ( Table 2 ) ., For the LMG and DG data , we did the test for the whole geometric shapes , in a similar way to that previously described 37 ., Briefly , the mean shapes were computed for each genotype group ( see Methods ) , and the mutual distances between any two mean shapes were calculated ., The mutual distance was calculated as the point-wise Procrustes distances ( PPD ) , which is the Procrustes distance normalized by the number of landmarks/marker-points ( see Methods ) ., PPD distance can be directly compared between the LMG and DG data ., The observed PPD distances were compared to 5000 random permutations to calculate empirical p values ( see Methods ) ., This procedure resulted in a single empirical p value for each comparison ., The geometric permutation tests indicated that several SNPs had a nominal significance of association in stage I , and these signals were highly consistent between the LMG and DG data ( Table 3 , Table S1 ) ., To facilitate the visualization of the detailed differences , we also calculated the point-wise Euclidean distances between the mean shapes of the DG data , plotted as color gradients across the whole face ( see Methods , Fig . S1 ) ., In gene IRF6 , two SNPs rs2236907 and rs642961 exhibited moderate evidence of association ., rs2236907 showed preliminary signals in both males and females ( Table 3 ) ., However , a stronger association was found for rs642961 in females , where the CC and CT types both differ substantially from the TT type ., The LMG tests had lower p values ( nominal p\u200a=\u200a0 . 005 and 0 . 006 for the CC/TT and CT/TT comparisons ) than the DG tests ( nominal p\u200a=\u200a0 . 04 and 0 . 02 for the CC/TT and CT/TT comparisons ) in both comparisons ., Color gradient plots reveal that the major changes occur around the lips ( Fig . S1A ) ., The GHR SNP rs6184 showed some potential association between CC and AA in males ( Table 3 , Fig . S1 J ) ., Two SNPs in the ENPP1 gene , rs6925433 and rs7773292 that were previously found to be associated with vertical upper face measurements in the Chuvashian population 33 , also showed potential association signals in our data ( Table 3 ) ., The preliminary signals were in males , although rs7773292 may be involved in forehead shape ( Fig . S1B ) , whereas SNP rs6925433 may be related to the chin area ( Fig . S1D ) ., SNP rs7773292 had the second strongest association signal among all the 10 markers , with the corresponding nominal p values scoring 0 . 015 and 0 . 034 in LMG and DG data respectively ( Table 3 ) ., The highly consistent pattern of p values between LMG and DG suggests that the 15 landmarks for the LMG data captured the total facial shape variation well ., It is also worth noting that signals based on LMD data ( rs642961 , rs2236907 and rs6184 ) overlapped substantially with those from LMG and DG data , suggesting a general compatibility among the three different schemes ., The signals from geometric tests ( LMG , DG ) were stronger than those of LMD , as their p values stood nominally significant at individual SNP level , whereas none of the LMD tests passed the single SNP significance level after permutation correction ., Globally , none of the LMG/DG proved significant after Bonfferoni correction assuming 60 independent tests ( 3 genotypes and 2 genders ×10 SNPs ) ., Since the geometric tests gave obviously stronger association signals than the LMD tests , we chose the candidate SNPs based on the LMG/DG results for further re-validation ., The two SNPs rs642961 and rs7773292 , from genes IRF6 and ENPP1 respectively exhibited the most prominent signals in stage I tests , and were selected to be revalidated in sample panel II ., The same tests as in stage I were carried out either solely with panel II or with the combined panel of I and II together ., The LMD data showed strong associations between rs642961 and several distances involving mouth landmarks , e . g . LLipP , ULipP and Stm ( Table S2 ) ., In particular in panel II , six pairwise distances , RAla-Stm , RAla- LLipP , LAla-LLipP , Stm-Sbn , ULipP-Sbn and LLipP-Sbn remained significant or marginally significant for the CC/TT and TT/CT comparisons ( corrected significance level 0 . 01 , Table S2 ) ., For the combined panel , the distance between LAla and LLipP gave corrected p values of 2 . 0×10−4 and 3 . 0×10−4 respectively for the CC/TT and CT/TT comparisons ( Table S2 ) ., Association signals in rs642961 were much more significant when the tests were carried out using the geometric data ( Table 4 ) ., In panel II alone , the females remained significant in the CC/TT comparison ( corrected p values 0 . 022 and 0 . 011 ) , and marginally significant in the CT/TT comparison ( corrected p values 0 . 089 and 0 . 054 ) after correcting for 12 tests ( 2 SNPs×6 comparisons ) ., The same 4 comparisons were more significant in the combined panel ( corrected p values 0 . 001∼0 . 065 ) after correcting for all 60 possible tests with 10 SNPs ( Table 4 ) ., The color gradient plots based on the dense geometric data in combined panel revealed substantial facial morphological differences between rs642961 TT and the other two genotypes ( Fig . 3 A , C , E ) , which were also highly consistent with the patterns revealed in panel I ( Fig . S1 A ) ., These plots clearly show that the strongest changes occur around the mouth region ., The comparison of the face profile lines revealed that the TT carriers on average had a slightly elevated forehead , as well as thicker and more protrusive ( 2–3 mm outwards ) lips , than the other two genotypes ( Fig . 3 B , D , F ) ., However , the signals from rs7773292 completely disappeared in all stage II tests ( Table S3 ) , suggesting a possible false positive signal ., To investigate the mouth shape changes associated with SNP rs642961 in more details , we extracted the mouth DG data from the whole face by retrieving a defined set of marker points for the mouth ., The 5 mouth landmarks ( LLipCn , RLipCn , ULipP , Stm , LLipP ) were also extracted to compose the mouth LMG data ., The landmark-distance analyses were not repeated as they remained the same despite the extraction of the mouth data ., Geometric permutation tests were conducted as before for the mouth LMG and DG data ., In general , the results seemed to be much more significant than the corresponding whole face comparisons ( Table 4 ) ., In panel II , the extreme nominal p value of 7×10−5 ( corrected p\u200a=\u200a0 . 00084 ) occurred between CC and TT in females in the LMG data ., In the combined panel , the CC/TT comparison in females had the minimum p value of 1×10−5 ( corrected p\u200a=\u200a0 . 00012 ) in both the LMG and DG data ., It should be noted that these p values for mouth region do not indicate any formal statistical significance as they were conditional on the prior information of the genetic association in mouth shape ., Nonetheless , the extreme p values suggested there are substantial impacts of genetic variants on normal mouth shape variation ., One potential problem that may affect the mouth shape analysis is the stomion point ., Stomion is the central point between the upper and lower lips ., None-neutral expressions or open mouth may induce altered distances between stomion and other mouth landmarks , therefore confound the association signals ., Our image dataset has been carefully screened for such cases ., In order to formally test the impacts , we removed stomion from the landmark set , and re-ran the image registration procedure and the LMG/DG analyses for SNP rs642961 ., As can be seen in Table S4 , the results remained largely unchanged , indicating that our results were not confounded by stomion variation ., Another potential confounding factor is age , as facial appearance changes during the time course of aging ., We carried out formal tests to examine whether there were non-negligible age effects in our sample ., As age 18 and 21 seemed to define the tails of the sample age distributions ( Fig . S2 ) , we grouped the individuals of 18 years or younger , and of 21 years or older , from the combined panel ., The average shape difference was tested on the DG data using permutation ( see methods ) ., Neither test was significant ( p value\u200a=\u200a0 . 267 for female; and 0 . 576 for male ) ., The same test was performed between other age groups , and also did not reveal any significant age/face interactions ., This suggests that age has little impact to the overall analyses in this study ., The mouth shape changes among different genotypes seem to involve complex shape changes , thus we performed further high-dimensional data analyses to describe such changes ., In the following analyses , we used the combined female panel unless otherwise specified ., We first carried out principle component analysis ( PCA , see Methods ) on both the LMG and DG data ., In the DG data , the first PC mode best distinguished the TT and CC genotypes ( t-test nominal p\u200a=\u200a1 . 3×10−6 ) , and the TT/CT comparison was also highly significant ( t-test nominal p\u200a=\u200a1 . 8×10−6 ) on this PC ., The first PC from the LMG data revealed similarly strong differences in the TT/CC ( t-test nominal p\u200a=\u200a2 . 7×10−6 ) and TT/CT ( t-test nominal p\u200a=\u200a2 . 2×10−6 ) comparison ., The large differences between TT and the other two genotypes and the little difference between CC and CT suggested that this locus may follow a dominant model ., To formally test this , we constructed an additive model and a dominant model based on the standard linear model ( see Methods ) ., The additive model did not suggest any statistical significance , whereas the dominant model was highly significant both with the LMG ( nominal p\u200a=\u200a1×10−6 ) and the DG data ( nominal p\u200a=\u200a6 . 8×10−6 ) ., Based on the dominant model , the genotypes of rs642961 explained a substantial proportion of the total variance ( 5 . 24% in the LMG data; 4 . 46% in the DG data ) in PC1 ., Interestingly , when we tested these two models in a combined panel that included both males and females , the additive model remained insignificant , and the dominant model also became only marginally significant ( nominal p\u200a=\u200a0 . 003 in the LMG data; nominal p\u200a=\u200a0 . 0159 in the DG data ) ., This suggests that the effect of TT is female specific ., To extract the facial pattern that best distinguishes TT from the other genotypes , we further carried out a simple linear discriminant analysis ( LDA ) ., As a hyperline that transects the mean points of TT and CC groups would best separate these two groups , this line was defined as a new data axis onto which individual data points were projected to generate hyperline ( HL ) scores ., The HL scores were plotted against the PC1 scores to visualize data distribution ( Fig . 4 ) ., As can be seen from Fig . 4 , the distributions on PC and HL are highly correlated ( r2\u200a=\u200a0 . 97 ) ., The TT distribution differed substantially from that of CC and CT ., Specifically , the average PC1 score of 0 found 18 of the 19 TT individuals at the minus side; similarly , the average HL score of 0 . 444 had 18 out of 19 TT individuals at the minus side ., To visualize the mouth shape changes , we transformed the mean shape ( Fig . 4B ) by adding or subtracting 3 standard deviations along either dimension as: st\u200a=\u200asm±3σvv , where st was the transformed shape , sm the average shape , v the Eigen vector of the dimension and σv the standard deviation ., The resulting shapes were defined as PC1+ , PC1− , HL+ and HL− respectively in Fig . 4 ., The PC1− shape ( Fig . 4A ) , which represents the trend for TT , has more protrusive and thicker lips compared to the finer and thinner lips in the PC1+ shape ( Fig . 4E ) ., The whole mouth region of PC1− is also more prominent and bigger than that of PC1+ ., Consistent with the high correlation between HL and PC1 , the face models along the HL dimension reveal similar shape changes ., ( Fig . 4 ) ., To the authors knowledge , this study is the first to use high resolution face image registration to test the genetic association for common facial variation ., Human face is a highly complex geometric surface ., The simple inter-landmark distances used in previous studies may have over-simplified the common variation of human faces ., As the high throughput acquisition of high content 3D image data becomes easier , methods based on shape geometric information , especially of high definition , become increasingly necessary to enable comprehensive and fully quantitative analyses of the complex facial traits ., Based on high density 3D face registration , we compared three different schemes of phenotype during tests of genetic association , including LMD , LMG and the high resolution geometric data DG ., We found that , in general , the three schemes produced consistent signals for the candidate SNPs ., In the stage I test , the LMD method had only moderate association signals , mainly due to the large number of tests ., The 15 landmarks gave rise to 105 possible tests in each genotype comparison ( Table 2 ) ., One strategy to reduce the number of tests is to use only the essential distances , e . g . the conventional craniometrical measures that correspond to obvious anatomical structures ., However , this risks missing the strongest signals ., The other major problem with distance data is the difficulty in perceiving the underlying shape changes ., For example in stage I , SNP rs642961 did not show a clear involvement with mouth shape changes in the LMD tests ( Table 2 ) ., However , such an involvement was already quite clear on the DG comparison in stage I ( Fig . S1 ) ., The LMD method seemed to improve both in the test power as well as the inference of shape changes ( most significant landmark-distances involved the mouth landmarks ) when larger sample sizes were used in stage II tests ., The two geometric schemes were generally found to give stronger association signals , implying better statistical power for the geometric methods ., This may be due partially to the fact that the geometric tests were carried out in one step , which avoided a complex test structure ., Interestingly , the LMG data of only 15 landmarks showed highly consistent test signals with that based on DG data ., This suggests that these 15 landmarks capture the majority of the normal facial morphological variation ., When only shape difference is to be tested , the LMG method seems to provide better efficiency ( given the smaller data involved ) and potentially higher test power ., However , the strong consistency between LMG and DG in the association signals attributed to rs642961 may be partially accounted for by the high landmark density around the mouth area ( 5 out of 15 chosen landmarks ) ., Features with fewer landmarks would confer lower power in the LMG tests ., On the other hand , the DG data has other unique advantages for shape change inference and modeling ., We also show here that the point-wise distance distribution between the mean faces can highlight the areas of shape changes in high definition ( Fig . 3 ) , which can guide future in depth exploration ., Furthermore , the effects of potential genetic factors may also be modeled visually as realistic 3D face images ( Fig . 4 ) ., This may have hugely beneficial applications to forensic studies ., Variants in the IRF6 gene have been found to predispose to the risk of NSCL/P 21–23 , 47 ., Nevertheless , a link between the IRF6 gene and common facial variation has not been established ., This is the first study that provides strong evidence that rs642961 also affects normal facial shape variation ., In particular , TT individuals may have more protrusive and thicker lips ( Fig . 4 ) ., Interestingly , such an effect is very likely female specific as the tests in males did not yield significant signals ., Combination of both sexes in the dominant model test also suggested that males did not contribute to the association signals ., This is not uncommon ., For example , various types of NSCL/P have been found to have sex specific spectra , suggesting sex is an important epistatic factor in mouth morphogenesis 16 , 48 ., In females , the TT individuals showed a highly specific distribution on the plane defined by PC1 and hyperline ( Fig . 4 ) ., This could be used during diagnosis to pre-screen the risk allele carriers by interpreting 3D pictures , therefore facilitating early prevention of NSCL/P ., We have also detected preliminary associations for other SNPs ., Failure to validate these association signals does not exclude them from the candidate list of loci related to normal facial shape variation ., Extended sample sizes as well as inclusion of samples from other populations will be needed to further increase our understanding of the genetics of human facial morphology ., Sample collection in this study was carried out with the approval of the ethics committee of the Shanghai Institutes for Biological Science and in accordance with the standards of the Declaration of Helsinki ., Written informed consent was obtained from every participant ., In total 1001 combined individuals ( 604 females , 397 males ) from self-reported Han Chinese were sampled from Taizhou , Jiangsu province ., Age

Introduction, Results, Discussion, Materials and Methods

Human facial morphology is a combination of many complex traits ., Little is known about the genetic basis of common facial morphological variation ., Existing association studies have largely used simple landmark-distances as surrogates for the complex morphological phenotypes of the face ., However , this can result in decreased statistical power and unclear inference of shape changes ., In this study , we applied a new image registration approach that automatically identified the salient landmarks and aligned the sample faces using high density pixel points ., Based on this high density registration , three different phenotype data schemes were used to test the association between the common facial morphological variation and 10 candidate SNPs , and their performances were compared ., The first scheme used traditional landmark-distances; the second relied on the geometric analysis of 15 landmarks and the third used geometric analysis of a dense registration of ∼30 , 000 3D points ., We found that the two geometric approaches were highly consistent in their detection of morphological changes ., The geometric method using dense registration further demonstrated superiority in the fine inference of shape changes and 3D face modeling ., Several candidate SNPs showed potential associations with different facial features ., In particular , one SNP , a known risk factor of non-syndromic cleft lips/palates , rs642961 in the IRF6 gene , was validated to strongly predict normal lip shape variation in female Han Chinese ., This study further demonstrated that dense face registration may substantially improve the detection and characterization of genetic association in common facial variation .

Heritability of human facial appearance is an intriguing question to the general public and researchers ., Although it is known that some facial features are highly heritable , the exact genetic basis is unknown ., Previous studies used simple linear measurements such as landmark distances , to evaluate the facial shape variation ., Such approaches , although easy to carry out , may lack statistical power and miss complex morphological changes ., In this study , we utilized a new 3D face registration method that enables subtle differences to be detected at high resolution 3D images ., Based on this , we tried to test and characterize the associations of 10 candidate genetic variants to common facial morphological variations ., Different types of phenotype data were extracted and compared in the association tests ., Our results show that geometry based data performed better than simple distance based data ., Furthermore , high density geometric data outstood the others in capturing small shape changes and modeling the 3D face visualization ., Interestingly , a genetic variant from IRF6 gene , which is also a well-known risk factor of non-syndrome cleft lip , was found to strongly predispose the mouth shape in Han Chinese females .

journal.pcbi.0030084

Enhancer Responses to Similarly Distributed Antagonistic Gradients in Development

With the availability of complete genome sequences and quantitative gene expression data , it becomes possible to explore the relationships between sequence features of regulatory DNAs and the transcriptional responses of their associated genes 1–7 ., Developmental genes regulated by multiple enhancer regions and their spatio–temporal dynamics of expression are of particular interest 8–11 ., The enhancers of developmental genes , such as gap and pair-rule genes , interpret maternally deposited information and participate in the formation of progressively more complex expression patterns , thus increasing the overall spatial complexity of the embryo ., In part , the information required to generate these downstream patterns ( e . g . , gap and pair-rule ) is present in the enhancer sequences ., Much attention has been paid to the investigation of transcription factor binding motifs and motif combinations , and to interpreting their role in the formation of spatial gene expression patterns ., 5 , 12 , 13 ., However , some early enhancers of Drosophila contain virtually identical sets of binding motifs , yet they produce distinct expression patterns 6 , 14 ., It has been argued extensively that binding site quality ( affinity ) and site arrangement within enhancers ( grammar ) contributes to the levels and precision of enhancer responses 6 , 15–21 ., In fact , some experimental studies of differentially arranged binding sites confirm the dependence of enhancer response on distances between binding sites and on binding site orientation 6 , 16 , 22–24 , and some structural enhancer features such as motif spacing preferences and characteristic binding site linkages ., “Composite elements” and other syntactical features were identified in many model organisms using computational analyses of binding site distributions throughout entire genomes 5 , 25 , 26 ., Recent studies involving in vivo selection of optimal binding-site combinations in yeast also revealed a number of preferred motif combinations and structural features 27 ., Nevertheless , some phylogenetic studies indicate significant flexibility in the regulatory code 28–31 ., The analysis of unrelated , structurally divergent , but functionally similar enhancers aids in defining the balance between the stringency of the functional cis-regulatory “code” and its flexibility as demonstrated by changes in primary enhancer sequence over the course of evolution ., 6 , 18 , 32 ., Requirements for multiple cofactors that influence transcription via protein–protein interaction complicate computational predictions and studies of enhancers ., While known binding motifs are easy to find , most protein–protein interactions leave no clear footprints in the DNA sequence of enhancers—some developmental coregulators such as CtBP ( C-terminal binding protein ) and Groucho influence the transcriptional response through interactions with sequence-specific transcription factors ( e . g . , 33 ) ., Finally , regulatory signals from enhancers must be transmitted to the basal transcriptional machinery; this involves enhancer–promoter communication of some sort , as well as the recruitment of mediator complexes 2 , 21 , 34–36 ., Both aspects further complicate the in silico prediction and analysis of enhancer activity ., Until recently , most models explaining enhancer responses in development were largely qualitative 37 , 38 ., Davidsons group 2 , 39 and Hwas group 21 undertook quantitative modeling of enhancer–promoter interactions and investigated the responses of architecturally complex regulatory units ., The elaborate nature of developmental enhancers in Drosophila was described in quantitative models introduced by Reinitzs group 1 , 7 ., Here , we summarize some basic structural considerations and investigate mechanisms of enhancer regulation to demonstrate how such features may affect the transcriptional responses ., Our quantitative analyses involve models based on the fractional occupancy of transcription factor binding sites present within enhancers 2 , 21 , 40 , 41 ., On the one hand , the described models are similar to those developed by Hwas group 21 as they consider structural enhancer details ., On the other hand , the models include biological assumptions for developmental enhancers ( i . e . , quenching ) , similar to those introduced by Reinitzs group 7 ., Technically , our models use a homotypic array ( a unit containing a number of identical sites ) of binding sites as an elementary unit for modeling ., Based on quantitative analysis of transcriptional responses , we analyze some models for developmental pattern formation ., In particular , we explore the outcome of the interplay between two antagonistic transcription factors , an activator and a repressor ., We demonstrate that a pair of antagonistic gradients with similar or even identical spatial distributions is sufficient to initiate stripes of expression of a downstream gene ., Given that the antagonistic gradients may be deposited by the same localized or terminal signal ( e . g . , in the fly embryo ) 42 or by a focal signal ( e . g . , in the case of a butterfly eyespot ) 43 , the models explain how initiation from a single point in space can lead to efficient gains in spatial complexity ., The transcriptional state of enhancers of developmental genes is among major factors in developmental pattern formation 6–8 , 10 ., If a transcription factor is present in a concentration gradient , the probability of that factor occupying a binding site in a target enhancer at a given position along the gradient depends on the factors concentration at that position ( coordinate ) ., This logic suggests that in the case of activator and repressor gradients , calculating the probability of activator , but not repressor , binding ( i . e . , the successful transcriptional state resulting in transcription ) may serve well to model the spatial expression patterns of the early developmental genes ., Let us consider an elementary enhancer , which contains two binding sites: one for an activator and one for a repressor ., Let us assume that binding of the activator A in the absence of the repressor R brings the elementary two-site regulatory unit i ( the enhancer; see Figure 1A ) into a successful transcriptional state ., The equilibrium probability of the successful state pi depends on the binding probabilities of A ( pA ) and R ( pR ) , which depend on the concentrations of the regulators ( A and R ) and on the binding constants ( KA and KR ) of the binding sites for the corresponding transcription factors ( see Equations S1–S5 in Protocol S1 ) :, Extending this formula to multiple different activators or repressors may be easily obtained with the same logic ( see Equation S6 in Protocol S1 ) ., Bintu and coworkers recently introduced a number of similar models , describing DNA–protein and protein–protein interactions on proximal promoters 21 , where the authors used an “effective dissociation constant , ” which is the inverse of the binding constant ( K ) used in this study ., Developmental enhancers usually contain homotypic or heterotypic binding site arrays for multiple activators and repressors 44 ., The probability of achieving a successful transcriptional state for the binding site array ( enhancer ) i , containing M identical , noninteracting activator sites and N identical , noninteracting repressor sites , is equal to ( see Equation S7 in Protocol S1 ) :, Here , Ψ is the sum of the statistical weights of molecular microstates for a homotypic site array and the denominator ΨAMΨRN is the sum of the statistical weights for all microstates of the system ( i . e . , the partition function; see Protocol S1 , “Binding site arrays” ) ., In such site arrays , bound transcription factors may cooperate or compete for binding ., Let us consider a cooperative array as an element of enhancer architecture ( Figure 1B ) ., Assuming presence of lateral diffusion 41 , 45 , equal binding affinities for all sites in the array and expressing cooperativity C as the ratio between the second and the first binding constants , one can approximate the sum of statistical weights Ψ of all possible molecular microstates for a cooperative array as follows ( see Equations S8 and S9 in Protocol S1 ) :, Binding sites for an activator and a repressor may overlap , and the corresponding proteins compete for binding ., Well-known examples in Drosophila development include Bicoid and Krüppel 46 , Caudal and Hunchback 44 , and Twist and Snail 6 ., The classic example outside Drosophila is the competition between CI and Cro in the phage lambda switch 47 ., The sum of microstates for a competitive site array , containing M overlapping A/R binding sites ( Figure 1C; also see Figure S1 and Equations S8–S12 in Protocol S1 ) , can be approximated by:, In addition to competitive interactions , this model also includes homotypic cooperative interactions between the regulators ( see Equations S10–S12 in Protocol S1 ) ., Structural elements within an enhancer ( single sites or entire site arrays ) may be distributed over extended genomic regions ( thousands of bases , e . g . , the Drosophila sna enhancer ) 48 , 49 ., In these cases , the distant regulatory elements within the enhancer may represent relatively independent units—modules 15 , 26 ( see Figure 1D ) ., Each independent module may include a single binding site or a binding site array ., Redundancy of the enhancer elements ( binding sites and modules ) is a well-known biological phenomenon 44 ., If the modules within an enhancer are independent from one another , bringing any one module into a successful transcriptional state may be sufficient for bringing the entire enhancer into a successful state , even if another module ( s ) is repressed ., Given the probabilities pi of successful states of all i independent modules or enhancers ( Equations 1–4 ) , the probability PEnc of the multimodule enhancer being in a successful state is equal to:, This is the reverse probability of the enhancer being in an inactive state , which is the product of the probabilities of each independent module being in an inactive state ( 1 − pi ) ; Reinitzs group 1 , 7 implemented similar expressions for the quenching mechanism ., While distinct modules may provide simultaneous responses to different inputs , multiple equivalent modules may allow for the boosting of an enhancers overall response to a single input 50 ( see Figure S1E and S1F ) ., In practice , however , the modules may not be completely independent from each other ., Short-range repression and other factors ( discussed below ) may be involved in distance-dependent module responses 22–24 , 48 ., Let us consider an enhancer containing two modules , a and b ., Module a contains an activator site and a repressor site; module b contains an activator site only ( see Figure 1E ) ., Potentially successful enhancer states include all combinations in which at least one activator molecule is bound ., However , the mixed state KaAAKaRR is always inactive as the repressor , and the activator sites in the module a are “close” ., If module b is not “too far” from module a , short-range repression from a may reach the activator site in b ., We can account for this possibility ( and for its extent ) by introducing a multiplier δ , depending on distance between the modules a and b ( see also Equations S14–S16 in Protocol S1 ) :, In this formula , Ψab is the sum of weights for all microstates , and Ψaboff is the sum of weights for the microstates that are always inactive ( see Protocol S1 , Equation S14 ) ., If modules a and b are “far , ” δ = 1; if they are “close , ” δ = 0 . If the distance between a and b is somewhere in between , so that a repressor bound in a partially affects the activator bound in b , we could introduce a distance function δ = f ( x ) ( 0 ≤ δ ≤ 1 ) , where δ depends on the distance x between a and b ( and perhaps other variables , such as the repressor type ) ., However , all we currently know about the distance function is that short-range repression is effective at distances less than 150–200 bases , and long-range repression may spread through entire gene loci ( i . e . , 10–15 kb 23 , 24 , 48 ) ., Without exact knowledge about the distance function , the module concept ( Equation 5 ) allows modeling of distance-dependent responses , but only in a binary close/far ( yes/no ) fashion ., Most of the enhancer response models ( Equations 1–6 ) consider inputs from two antagonistic gradients , but enhancers may be under the control of a larger number of regulators ( see Figure 1D ) ., However , gradients of some of these regulators may either have similar spatial distributions ( e . g . , Dorsal and Twist ) 51 , or non-overlapping spatial expression domains ( e . g . , Krüppel and Giant ) 37 ., Therefore , in many cases the combination of all inputs may be parsed down to one or more pairs of antagonistic interactions ., Based on the described quantitative models approximating enhancer responses ( see above ) , we analyzed possible spatial solutions produced by gradients of two antagonistic regulators ., The examples in Figure 2A–2C demonstrate that the spectrum of possible enhancer responses is quite rich ., One surprising result of these simulations is that even identically distributed antagonistic gradients can yield distinct spatial expression patterns such as stripes ( Figure 2B ) ., We identified conditions for the “stripe” solutions using differential analysis of the site occupancy function shown in Equation, 1 . For example , if both regulators are distributed as identical gradients and if their concentrations and binding constants are equal ( KA = KR; A = R ) , then it is sufficient to identify conditions for the maximum of a site-occupancy function y ( x ) depending on the spatial coordinate x:, In this variant of Equation 1 , k is the product of absolute concentration of the regulators Abs and the binding constant KA ( k = KAAbs ) ., The function f ( x ) is the distribution of the relative concentration ( 0 ≤ f ( x ) ≤ 1 ) of the transcription factors along the spatial coordinate x ( i . e . , the embryo axis ) ., The functions maximum y′ ( x ) = 0; x > 0 is f ( x ) = 1/k ., In the Gaussian , logistic , and exponential decay forms of the function f ( x ) ( see details in 52 ) , the maximum 1/k exists only if KAAbs > 1 ( i . e . , if binding constants and/or the absolute concentrations are high ) ( see also Figure S2 ) ., In the simple case ( Equation 7 ) , the absolute value of the fractional occupancy at the maximum is not very high ( 0 . 25 ) ; adding more sites or modules ( see Figure S1 ) allows for the functions values to approach 1 ( see Figure 2B ) ., However , if the antagonistic gradients are not identical ( e . g . , if the activator gradient is “wider” than the repressor gradient ) , the solutions for the stripe expression are more robust ( Figure 2A ) ., Shifting the peak of the activator gradient relative to the repressor gradient produces even more robust stripe patterns , as in the case of classical qualitative models 37 , where a repressor “splits” or “carves out” the expression of a target gene ( Figure 2C ) ., The formation of distinct gene expression domains ( e . g . , stripes ) in response to similarly or even identically distributed gradients is of interest because this mechanism can lead to the very efficient gain of spatial complexity in just a single step: based on primary sequence , enhancers of target genes can translate two similarly distributed gradients into distinct gene expression domains or stripes ., Such similarly distributed antagonistic gradients may come about by induction due to a single maternal gradient or due to a terminal ( focal ) signal emanating from a discrete point or embryo pole ., The general pattern formation mechanism in the case described can be represented as follows: ( 1 ) maternal/terminal signal initiates two antagonistic gradients; and ( 2 ) interactions between the two gradients produce multiple stripe patterns ., In an extreme case ( e . g . , Figure 2B ) , the described “antagonistic” mechanism could use only a single gradient/polar signal to produce multiple stripes of target gene expression ., The interaction between two antagonistic gradients is an example of a feed-forward loop ., Due to a cascade organization of the developmental transcriptional networks , feed-forward loops are among the most common network elements ( network motifs ) ; a detailed analysis of the feed-forward networks and potential solutions can be found in a recent work by Ishihara et al 53 ., To explore the interplay of antagonistic gradients in detail , we considered particular examples , such as the regulation of rhomboid ( rho ) by gradients of Twist and Snail and the regulation of knirps by the maternal gradients of Hunchback and Bicoid 54 ., The enhancer associated with rho directs localized expression in ventral regions of the neurogenic ectoderm ( vNEs ) 51 ., The rho vNE enhancer , as well as enhancers of other vNE genes such as ventral nervous system defective ( vnd ) , is activated by the combination of Dorsal and Twist , but is repressed by Snail in the ventral mesoderm 13 , 51 ., Both Twist and Snail are targets of the nuclear Dorsal gradient , which is established by the graded activation of the Toll receptor in response to maternal determinants 55 ., The Twist and Snail expression patterns occupy presumptive mesodermal domains in the embryo , yielding slightly distinct protein distributions ., Our recent quantitative analysis indicates that the boundaries of rho and vnd expression are defined largely by the interplay of the two antagonistic Twist and Snail gradients ( see Figure 2D and 2F ) 6 , and the expression patterns of rho and vnd resemble the predicted solutions shown in Figure 2A ., The patterning mechanism in this case can be represented as follows: ( 1 ) a terminal signal ( Toll/Dorsal gradient ) initiates two similar antagonistic gradients , Twi and Sna; and ( 2 ) Twi and Sna gradients produce multiple ( distinct ) stripe patterns ( rho , other vNE genes ) ., Another example of the interplay between an activator and a repressor gradient is the early expression of the gap gene knirps in response to maternal gradients of Bicoid and Hunchback ., Bicoid and Hunchback are deposited maternally and have similar , but distinct distributions—high in the anterior and low in more posterior regions of the embryo ( see Figure 2E ) ., The graded drop-off of the knirps repressor Hunchback at 50%–60% egg length is steeper than that of the knirps activator Bicoid ., This is similar to the theoretical case shown in Figure 2C , where a narrow repressor “splits” a wider activator expression domain , thus producing two peaks of expression of the downstream gene ., Known enhancer elements of knirps drive kni expression in the anterior and the posterior embryo domains and contain binding sites for Bicoid , Hunchback , Caudal , Tailless , and Giant 44 , 56–58 ., However , tailless , caudal , and giant are downstream of Bicoid; it is likely that these and some other genes participate in the later maintenance of kni expression ., It has been extensively argued that gap genes ( and hunchback ) stabilize their patterns along the anterior–posterior axis by mechanism of mutual repression 49 ., At later stages ( after cycle 14 ) , the inputs from Bicoid and Hunchback into knirps regulation may stabilize fluctuations in knirps expression and fluctuations in the entire gap gene network due to mutual repression ., Dynamic models from Reinitzs group based on slightly different logistic response functions support the sufficiency of Bicoid and Hunchback in the establishment of the early knirps expression 59 ., To explore the role of Bicoid and Hunchback interplay in the early expression of knirps , quantitative expression data for Bicoid , Hunchback , and Knirps were downloaded from the FlyEx database 60 , and models simulating the knirps enhancer response were generated based on Equations 1–4 ., One model assumed that Bicoid and Hunchback bind independently from each other; another model assumed that there is an interference ( possibly competition ) between the Bicoid and the Hunchback sites ( Equation 7: competitive binding ) ., Fitting the available quantitative data with the models ( see parameter values in Table 1 ) shows that both models are sufficient to explain the posterior expression of knirps ., However , the competitive model ( Figure 2G ) also predicts the anterior expression of knirps ., This result was especially striking , as the anterior knirps expression data were not included in some of the fitting tests ., Bicoid and Hunchback motifs are quite different , so it is unlikely that this is a case of direct competition for overlapping binding sites ., Other mechanisms may account for the negative interaction between the two regulators; for instance , binding of Bicoid may prevent Hunchback dimerization 61 and/or efficient binding ., Shifting the knirps expression data by more than 5% along the anterior–posterior axis ( see Materials and Methods ) results in reduction of the data-to-model fit quality for the posterior kni expression domain ( see Table 1 for exact parameter values ) ., The robustness of knirps regulation was emphasized earlier 59 , 62 , and the present analysis using site occupancy confirms that the interplay of the two antagonistic gradients , Bicoid and Hunchback , is sufficient to explain the initial formation of both the anterior and the posterior strips of knirps expression ., To test the models describing gene response to antagonistic gradients , we introduced mutations in the rho enhancer and compared the expression patterns produced by the reporter gene in vivo with the simulated expression patterns simulated in silico ( Equations 1–6 ) ., Specifically , the models for rho and vnd expression predicted the following 6: ( 1 ) The position of the dorsal expression border of rho is highly sensitive to Twist and/or its cooperativity with Dorsal ., Reducing Dl–Twi cooperativity or Twist–Twist cooperativity shifts the dorsal border ventrally ., ( 2 ) The number of independent elements ( groups of closely spaced Dorsal-Twist-Snail sites , or “DTS” elements ) contributes to the expression pattern of rho and vnd according to Equation 5 ( boost ) : a higher number of DTS elements in vnd is responsible for the shift of the ventral vnd expression border relatively to rho 6 ., These two specific predictions , based on the model analysis and simulations , were tested by modifying the structure of the minimal rho enhancer ., First , the distance between the Dorsal and the Twist sites in the DTS element was increased ( see Figure 3 ) ., The increased distance between the two sites reduced the cooperative potential between the Dorsal and Twist sites ., Indeed , the observed effect in vivo is consistent with the effect of the same mutation simulated in silico , causing a ventral shift of the dorsal border of the reporter gene expression ( compare Figure 3E with 3A ) ., An additional mutation eliminating the weaker Twist site from the DTS element affects Twist–Twist cooperativity in the enhancer and shifts the dorsal rho–lacZ expression border ., In fact , the combined effect produced by these two mutations in vivo ( Figure 3G; compare with 3C ) and the deletion of the weak Twist site alone ( Figure 3F; compare with 3B ) demonstrate shifts of the dorsal expression border of the rho-lacZ transgene in concordance with the models ., Last , a second DTS module was introduced into the rho enhancer in the context of the previous two mutations ., The predicted in silico effect is a “boost” in expression , resulting in the shift of both ventral and dorsal expression borders ., Again , the predicted changes in the expression pattern were observed in vivo—not only were the positions of the ventral and the dorsal border shifted ( Figure 3H; compare with 3D ) , but the overall level of expression of this transgenic construct appears higher ( unpublished data ) ., The described in vivo tests of the in silico predictions using site-directed mutagenesis of the rho enhancer have demonstrated that though the quantitative models based on fractional site occupancy are approximations , they can produce reasonable predictions for the response of complex regulatory units ( such as fly enhancers ) to gradients of transcriptional regulators ., Using transcriptional response models and quantitative expression data , we demonstrated how two similar terminal gradients can determine stripes of expression of downstream genes ., Related examples are quite frequent in development ., For instance , the posterior stripe of hunchback is the result of activation by Tailless and repression by Huckebein 63 , 64 ., As in the case with Twist and Snail , the posterior gradient of Tailless is slightly broader than the gradient of Huckebein ., Therefore , the mechanisms of posterior hunchback expression may be similar to the mechanisms shown in Figure 2A , 2B , 2D , and 2F ., However , while the examples above involve direct transcriptional regulation in the embryonic syncytial blastoderm , extracellular morphogen gradients may produce similar outcomes if the cellular response is transcriptional in nature ., Formation of eyespot patterns in butterfly wings is an elegant example of axial ( here focal ) patterning in a cellular environment ( see Figure 4A ) ., The interplay between Notch and Distalless specifies the position of focal spots and intervein midline patterns in the butterfly wing 65 ., Subsequent Hedgehog signaling from the focal spots is believed to induce the formation of concentric rings of gene expression and the pigmentation of the eyespots in the adult butterfly wing 66 ., Known targets of the Hedgehog gradient are the butterfly homologs of engrailed and spalt 67 ., Initially , both genes are expressed around the focal spot , but at later stages an external ring of engrailed expression appears around the spalt expression pattern ( see Figure 4B and 4C ) ., In the case of engrailed pattern formation , a simplified mechanism 67 may include elements of the following feed-forward network: ( 1 ) focal signal ( focal spot/Hedgehog signaling ) initiates two antagonistic gradients , the activator Engrailed and the repressor Spalt; and ( 2 ) subsequent interactions between Engrailed and Spalt produce multiple ring patterns ., An extension of the model in Equation 1 , ( k is the rate of synthesis and c is the rate of decay; dR/dt = 0 ) reproduces the dynamic changes in the engrailed pattern ( Figure 4A , 4D–4E ) :, Examples of axial or focal patterning using a single source of signaling or a combination of similar antagonistic gradients are common ., The interplay between maternal hunchback and maternal nanos during development of the short germ-band insect Schistocerca is an example of axial patterning similar to the interplay between Bicoid and Hunchback 68 ., Specification of segments during insect limb development is comparable to the mechanisms of Twist/Snail interplay and the butterfly eyespot formation 69 ., Nature uses many combinations of signals and gradients in pattern formation , but the most effective mechanism/combination may be one that allows maximal informational gain in a minimal number of steps ., From this perspective , the interplay between similar or identical gradients is of significant interest ., Quantitative distribution data for Dorsal , Twist , and Snail were published previously 6 ., Quantitative expression data for mRNA levels of mutated rho enhancers were generated by in situ hybridization ( the data are available at the DVEx database: http://www . dvex . org ) ., Multiplex in situ hybridization probes were used for colocalization studies , including co-stainings for the endogenous mRNAs and lacZ reporter gene expression as described previously , and confocal microscopy and image acquisition were performed as described 6 ., In short , signal intensity profiles of sum projections along the dorso–ventral axis of mid-nuclear cleavage cycle of 14 embryos were acquired using the ImageJ analysis tool ( National Institutes of Health , http://rsb . info . nih . gov/ij ) ., Background signals were approximated by parabolic functions and subtracted according to existing methods 70 ., Online programs for the automated background subtraction and data alignment are available from the University of California Berkeley Web resource ( http://webfiles . berkeley . edu/∼dap5 ) ., After background subtraction , the data were resampled and aligned according to the position of Snail gradient and the distribution of endogenous rho message ., Expression datasets for anterior–posterior genes were downloaded from the FlyEx database ( with options: integrated , without background ) 60 ., In all cases , signal amplitude was normalized to the 0–1 range , and the data was resampled to 1 , 000 datapoints along the coordinate of the corresponding axis ., In all models , we used the relative concentration multiplied by a maximal absolute concentration ., This absolute concentration is an independent unknown parameter ( range , 10−8–10−9 M ) equal for all reaction components ., The minimal rho enhancer 6 was mutated via site-directed mutagenesis in pGem T-Easy ( Promega; http://www . promega . com ) using the following primers: Dl-Twi distance , RZ65mut: 5′-GTTGAGCACATGTTTACCCCGATTGGGGAAATTCCCGG-3′; deletion of Twist site , RZ66mut: 5′-GGCACTCGCATAGATTGAGCACATG-3′; creation of a second DTS , RZ67mut: 5′-GCAACTTGCGGAAGGGAAATCCCGCTGCAACAAAAAG-3′; and RZ68mut: 5′-CACACATCGCGACACATGTGGCGCAACTTGC-3′ ., Mutated enhancers were cloned into the insulated P-element injection vector E2G as described previously 13: constructs were introduced into the D . melanogaster germline by microinjection as described previously 71 ., Between three and six independent transgenic lines were obtained and tested for each construct; results were consistent across lines ., To fit our models with actual quantitative data , we maximized the agreement r ( Pearson association coefficient ) between the model output predictions and the observed ( measured ) expression patterns:, The best set of parameters X* from the parameter space I is defined by the binding constants , cooperativity values , and the number of binding sites ., We used a standard hill-climbing algorithm ( full neighborhood search ) for the main parameter space ( e . g . , 72 ) ., For each identified maximum , we measured the value of the site occupancy function and discarded maxima that produce site saturation values below selected thresholds , as well as such that are located beyond selected realistic parameter ranges for binding constants and cooperativity values ., All maxima producing the highest data-to-model agreement were found multiple times , suggesting that exhaustive mapping of the parameter space was achieved ., Fitting “shifted data” ( wrong data ) for Knirps was performed by exploring exactly the same parameter space and exactly the same number of seed points for each shift value ., Quantitative gene expression data for dorso–ventral genes are available at http://www . dvex . org; the analysis tool “E-response , ” fitting utilities , and online data-treatment programs are available at the University of California Berkeley Web resource http://webfiles . berkeley . edu/∼dap5 .

Introduction, Results/Discussion, Materials and Methods

Formation of spatial gene expression patterns in development depends on transcriptional responses mediated by gene control regions , enhancers ., Here , we explore possible responses of enhancers to overlapping gradients of antagonistic transcriptional regulators in the Drosophila embryo ., Using quantitative models based on enhancer structure , we demonstrate how a pair of antagonistic transcription factor gradients with similar or even identical spatial distributions can lead to the formation of distinct gene expression domains along the embryo axes ., The described mechanisms are sufficient to explain the formation of the anterior and the posterior knirps expression , the posterior hunchback expression domain , and the lateral stripes of rhomboid expression and of other ventral neurogenic ectodermal genes ., The considered principles of interaction between antagonistic gradients at the enhancer level can also be applied to diverse developmental processes , such as domain specification in imaginal discs , or even eyespot pattern formation in the butterfly wing .

The early development of the fruit fly embryo depends on an intricate but well-studied gene regulatory network ., In fly eggs , maternally deposited gene products—morphogenes—form spatial concentration gradients ., The graded distribution of the maternal morphogenes initiates a cascade of gene interactions leading to embryo development ., Gradients of activators and repressors regulating common target genes may produce different outcomes depending on molecular mechanisms , mediating their function ., Here , we describe quantitative mathematical models for the interplay between gradients of positive and negative transcriptional regulators—proteins , activating or repressing their target genes through binding the genes regulatory DNA sequences ., We predict possible spatial outcomes of the transcriptional antagonistic interactions in fly development and consider examples where the predicted cases may take place .

drosophila, developmental biology, computational biology

journal.pcbi.1002212

Automatic Morphological Subtyping Reveals New Roles of Caspases in Mitochondrial Dynamics

Recent studies have shown that the fusion-fission dynamics of mitochondria are essential to many cellular processes , including ATP-level maintenance , redox signaling , oxidative stress generation , and cell death 1–4 ., Meanwhile , it is also known that dysfunctional mitochondrial dynamics ushers the aging process and neuronal degenerative diseases 5–11 ., Since mitochondrial morphology reveals physiological and pathological status , tracking mitochondrial morphological differences becomes important ., Previous studies roughly classified mitochondrial morphology into various subtypes , such as , fragmented globules , tubular threads , networks , clumps or swollen granules , and usually the classification was performed by human inspection 2 , 8 , 9 , 12 , which inevitably introduces biases and inconsistency and lowers replicability of the results ., Previous attempts of automatic quantification include measuring length , width , area and other primitive parameters of mitochondrial objects 13 and also skeleton length 14 , but these measures are insufficient to fully distinguish the morphological diversity of mitochondria ., They 13 , 14 investigated only the average of these feature values within each cell , while in this paper , we present a computational approach that allows us to identify representative morphological subtypes and quantify the morphological diversity of mitochondria ., Our approach consists of successive steps of image segmentation , consensus clustering and classifier learning designed to identify subtypes as well as construct a subtype classifier ., A large set of fluorescent microscopic images of Chinese Hamster Ovary ( CHO ) cells were used as the sample to derive the subtypes ., A subset of these CHO cells was treated with squamocin , a compound known to induce apoptosis 15–18 ., Squamocin treatment results in mitochondrial fragmentation , which can then be suppressed by inhibitors of Caspases 8 and 9 ( z-IETD and z-LEHD , respectively ) , but cells are still killed by squamocin even with the presence of these inhibitors 19 ., One possible reason is that Caspase inhibitors may not have fully restored mitochondrial structures ., With the developed computational approach , we were able to quantify the difference of morphological changes of mitochondria in cells under different treatments ., We used a previously developed image segmentation method 20 to accurately extract each individual mitochondrial object from cell micrographs ., This method applies adaptive local normalization 20 and Otsus image thresholding method 21 to deal with noisy background and variant object intensity that are constantly present in fluorescent micrographs , a challenging issue for existing image binarization methods ., We then applied this method to extract a large number of mitochondrial objects ., Each object was then represented by a set of image features , including morphological , skeletal , and binary Haralick texture features ., According to result of consensus clustering of the image features , we merged clusters that are functionally similar to define six morphological subtypes of mitochondria ., This new subtyping covers mitochondrial morphology reported in the literature and provides an unambiguous observable indicator of the status of a mitochondrion in a cell ., We then applied supervised learning algorithms to train an automatic classification system to classify mitochondrial objects in images into one of the six subtypes ., The resulting system combined with object extraction was named “MicroP , ” which allows us to accurately measure mitochondrial subtype compositions in individual cells to profile the outcome of different drug treatments ., We used MicroP to study the effect of squamocin , a compound known to induce cell apoptosis by triggering mitochondrial fragmentation ., Inhibitors of Caspases 8 and 9 can partially rescue the fragmentation but the mechanism was not previously characterized ., Analyzing subtype composition of cells treated by squamocin and Caspase inhibitors reveals a more comprehensive picture of how Caspases 8 and 9 interact with mitochondrial fusion-fission regulatory proteins to influence morphology and functions of mitochondria ., We extracted 225 , 556 mitochondrial objects from the whole dataset of 1422 cells , and a total of 19 distinct morphologies were identified by consensus clustering with manual validation as shown in Figure 1 ., Inspecting these morphologies reveals that these automatically derived morphological clusters indeed have distinctive shapes ., To develop representative morphological subtypes for automatic classification , we further regrouped these 19 clusters according to morphological similarity and molecular functional characteristics , supported by evidence reported in the literature ., The result is six subtypes: small globules , swollen globules , straight tubules , twisted tubules , branched tubules and loops ., These subtypes provide useful indicators of specific cell conditions , and will be used as a basis for generating manually labeled mitochondria as training sets ., Small globules ( Figure 1A ) comprise the largest clusters in the clustering results ., Swollen globules ( Figure 1B ) are in fact found as outliers in the clusters of small globules , since their number is too small to be distinguished by unsupervised clustering ., But they are an important morphological subtype of mitochondria , suggested to represent dysfunctional mitochondria undergoing mitophagy by previous investigations 9 ., Therefore , we separated swollen globules manually from the clusters of small globules using the mean plus 3 standard deviation of the area as a reference threshold to form another subtype ., Straight , twisted , and branched tubules were separated by unsupervised clustering into clusters of distinct lengths ., However , we assumed that the difference in length only reflects different frequencies of fusion events but not different biological mechanisms of formation ., Also , to the best of our knowledge , there is no report in the literature that explicitly distinguishes mitochondrial morphology by length ., Thus length was excluded as a criterion to differentiate morphological subtypes ., We merged five straight tubule clusters of distinct lengths ( Figure 1D ) into a single subtype , and similarly merged 3 clusters of twisted tubules ( Figure 1E ) and 5 clusters of branched tubules ( Figure 1F ) ., Twisted tubules have only three instead of five distinct lengths probably because longer twisted tubules have a higher chance to form branched tubules , and that tubules need to reach a certain length before they can become “twisted . ”, Mitochondria become twisted when they are not fully associated with microtubules , and portions of mitochondria not associated with microtubule motors will collide with water molecules and become twisted ., It was observed from our time-lapse images ( data not shown ) that these twisted mitochondria are constantly moving ., We consider only those mitochondria that exhibit these characteristics as “twisted tubules” but not those mildly curved ones ., Finally , we grouped “horseshoe” mitochondria with “donuts” into a single “loops” subtype ( Figure 1C ) , since both of them maintain a much higher degree of stable curvature than twisted tubules ., We observed that “horseshoe” mitochondria stably maintain a high degree of curvature , which would require extra force , e . g . , inter-mitochondrial end-to-end fusion , to stabilize ., In contrast , twisted mitochondria dynamically vibrate their ends due to collision with other molecules , e . g . , water , which makes them appear twisted ., We trained an ensemble of three classifiers on a training set manually labeled with the six morphological subtypes defined in the last section ., Performance of the three classifiers were individually assessed both by holdout testing accuracy and visual inspection of results for unlabeled mitochondrial micrographs ., Table 1 reports the performance assessment ., The figures are averages of 20 runs of holdout testing , where each run used a random partition of training and holdout sets ., Table 2 shows the aggregate confusion matrix over all holdout testing runs for all classifiers ., Performance for most subtypes are above 80% in accuracy , except for twisted tubules ( ) , which are often confused with straight tubules by the classifiers ., This is expected because it is also difficult for a human to judge whether a short mitochondrial tubule is twisted or straight ., Next , the entire set of labeled mitochondria samples were used as the training examples for the three classifiers ., Table 3 shows the cross validation accuracy results ., We also report the optimal parameters used for each classifier ., Visual inspection of classification results of unlabeled mitochondria in micrographs shows that all three classifiers performed adequately , with complementary types of errors for each morphology subtype ., Hence we used an ensemble classifier , which combines the decisions of the three classifiers by majority vote , as the final classifier for unlabeled cells ., The ensemble classifier outperforms individual classifiers in holdout testing ., Figures S5–7 in Text S1 show the histograms of basic morphological features of mitochondria for different subtypes ., These features are strongly correlated with the subjective criteria used by human inspectors in manual labeling of the training samples ., Morphological subtyping of mitochondria can help quantify how mitochondrial morphology are affected by drug treatments ., Following up on our previous biochemical studies 19 , 22 , we investigate whether squamocin-induced reduction of mitochondrial biomass and mitochondrial fragmentation are fully restored by z-IETD and z-LEHD ., Identifying morphological subtypes may provide new clues for the design of further experiments to determine how these features are correlated with apoptosis ., Figure 2 shows the effects of different treatments on the total number and area of mitochondria in cells ., Squamocin induces large numbers of mitochondria but reduces total mitochondria area compared to control ., z-IETD and z-LEHD fully restore the number of mitochondria to the control level but only partially restore the area ., Among them , z-LEHD shows slightly better restoration ability than z-IETD ., These results suggest that squamocin increases the mitochondria number by inducing mitochondrial fission and possibly also reduces the sizes of individual mitochondria ., Squamocin may also reduce biogenesis or enhance degradation of mitochondria , resulting in reduction of total mitochondrial area in cells ., In some images of squamocin-treated cells , the intensity of a few mitochondria was lower than the threshold of the segmentation algorithm and thus those mitochondria were omitted and may contribute to reduction of the total area ., However , their number is too small to affect the main conclusion here ., Figure 3 shows the average ratio of mitochondrial subtypes in cells treated by different drugs ., Small globules and straight tubules are predominant in all treatments , followed by branched tubules ., The other subtypes are relatively rare ., As expected , squamocin induces more small globules and notably decreases the number of branched , twisted tubules and loops ., Straight tubules , however , are not affected as much ., z-IETD and z-LEHD restore squamocin-reduced straight and twisted tubules completely , but only partially restore branched tubules ., Comparing the two Caspase inhibitors , it can be seen that z-LEHD restores more branched tubules than z-IETD , but the difference is not significant ., Long error bars ( representing standard deviation ) in the figure implies that responses of cells to these treatments were quite heterogeneous ., To determine whether the shape of each subtype is affected by different drug treatments , differences in the distributions of mitochondrial morphological features for each subtype was investigated ., No significant difference was found for average morphological feature values for cells given different treatments , either over all subtypes or within individual subtypes ( data not shown ) ., However , it was observed that tubules were slightly shorter in squamocin-treated cells and that loops were slightly shorter in z-IETD cells ., In summary , drug treatments affect the distribution of subtypes in cells but not the shape of the subtypes , although this is expected partly due to the use of the same classifier for all treatment populations ., The correlations between the ratios of mitochondrial subtypes in each cell may hint at the biological mechanism of their formation ., Positively correlated subtypes may be formed by the same mechanism while negative correlation of subtypes implies transition between subtypes ., Figure 4 shows the correlation heat map of different subtypes in cells treated by different drugs ., The patterns for DMSO ( control ) and squamocin are quite different whereas those for control and z-LEHD are similar ., The correlation pattern of z-LEHD is closer to control than z-IETD , while z-IETD is closer to control than squamocin ., This provides further evidence of better restoration capability of z-LEHD ., Notable pairwise correlations include: One of the challenges in cell-based analysis is that cells may be at different phases of the cell cycle and have different timing to respond to treatments ., The target of this study , mitochondrial morphology , is especially sensitive to bioenergetics , cell cycle , aging , regulatory proteins and various stresses 9 ., Therefore , although general trends of squamocin and the restorative effects of Caspase inhibitors can be verified visually , it is challenging to rigorously validate the differences by statistical analysis ., For cell response profiling , we used a set of cell features designed to characterize the composition of mitochondrial subtypes in cells ., The difference between cells in terms of their mitochondrial morphology can be easily measured as the Euclidean distance in cell feature space ., Figure 5A shows the result of multidimensional scaling ( MDS ) that maps the cells from multidimensional space of cell features to a two dimensional space so that we can visualize their difference ., Each point in the figure is a cell that is colored differently , representing the treatment received ., The MDS plot shows that cells receiving the same treatment respond heterogeneously and those treated differently overlap ., Diamonds in the plot mark the mean responses for each treatment ., The distance between the mean responses of Squamocin and DMSO ( control ) is the largest , implying that responses to these treatments are the most differentiated ., Between Squamocin and DMSO lies z-IETD and z-LEHD successively ., This result suggests a trend of differing restoration ability of Caspase inhibitors , similar to the results in Figures 2 and, 3 . Since both MDS and subtype ratio results show that the responses of cells to the same treatment are heterogeneous and many cells in different treatments have similar responses in terms of mitochondrial morphology , these overlapped responses may be considered as transitional responses or no response ., Here , we used the silhouette coefficient analysis 23 to identify cells with representative responses for each treatment ., By concentrating on representative cells , a more reliable profiling of these treatments can be established ., We removed cells with silhouette coefficient from the analysis , since they are on average closer to cells in another treatment population than to cells in its own treatment population ., In the end , the percentages of cells remaining with silhouette coefficient are 82 . 6% , 45 . 4% , 46 . 7% and 15 . 0% for treatment populations DMSO , Squamocin , z-IETD and z-LEHD , respectively ., Figure 5B shows the MDS plot of only these representative cells with the mean responses re-computed ., The plot shows that the relative distances between the mean responses stays the same as in Figure 5A , but the relative positions are changed ., The subtyping results of representative cells from each population ( Figure 6 ) suggest that in Figure 5B , the x-axis is highly correlated with the degree of fragmentation while the y-axis is highly correlated with the prevalence of twisted tubules and loops ., The equivalents of Figures 2 , 3 , and 4 calculated using only representative cells are shown in Figures S11–13 in Text S1 , respectively ., These additional results strengthen the support of our main conclusions ., Figure 6 shows subtyping results of select representative cells in different treatment populations ., The distributions of the cell features of representative cells ( those with ) reveal unique characteristics of the effect of the different drugs on the composition of morphological subtypes of mitochondria: For cells with silhouette coefficient , we also keep track of which other treatment population it is closest to , and the percentages of cells in each treatment that are closer to every other treatment populations is shown in Table, 4 . It can be seen that except for the z-LEHD population , the majority of cells in each treatment is closer to its own treatment population ., The six representative subtypes of mitochondrial morphology: small globules , swollen globules , straight tubules , twisted tubules , branched tubules and loops , are identified based on the unsupervised consensus clustering results and evidence of functional similarity reported in the literature ., Based on previous literature and hints from subtype ratio correlations ( Figure 4 ) , we propose a model for the formation and transition of the six subtypes ( Figure 7 ) ., Previous studies identified three types of mitochondrial morphologies , fragmented , tubular , and network-like mitochondria , and showed that regulatory proteins of mitochondrial dynamics determine which type of mitochondria is predominant 1 ., It is also known that bioenergetic states and oxidative stress will affect the distribution of mitochondrial morphology 24 ., In our subtyping , fragmented mitochondria are represented by “small globules , ” and tubular mitochondria are subdivided into four different subtypes: straight , twisted , branched tubules and loops ., Formation of straight or twisted mitochondria are mainly dependent on the assembly of microtubules , but independent of other factors affecting mitochondrial morphology 25–27 ., Branched tubules result from inter-mitochondrial end-to-end and side-to-side fusion , and has a unique molecular mechanism of formation and distribution specific to physiological conditions 28–30 ., Donut-shaped mitochondria are induced by moderate mitochondrial membrane potential and low cellular respiration 2 , 7 , 31 ., Network-like mitochondria were not observed in our dataset ., Finally , swollen granules are a new unique structure characterized recently ., Mitochondria can become swollen due to dysfunction or various conditions ( e . g . loss of / exchange activity ) , and are subsequently targeted for “mitophagy . ”, Another formation mechanism involves fragmented mitochondria that are packed into autophagosomes 9 ., Recently , Yoshii et al . 32 also showed that the outer membrane of mitochondria can be ruptured by mitophagy , resulting in swollen mitochondria ., The quantitative analysis in this study demonstrates for the first time that compositions of mitochondrial morphological subtypes may be heterogeneous within a treatment population , with representative morphologies specific to drug treatments ., We consider here the biological significance of these representative mitochondrial morphologies for each treatment condition ., The control ( DMSO ) cells are characterized by tubular mitochondria , especially branched ones ( Figure 6A ) , while small globules are the representative mitochondrial morphology of squamocin-treated cells that are responsive to squamocin ( Figure 6B ) ., Table 4 shows that a majority of control cells ( 83% ) are closer to the control population ., Hence these cells can be considered to be representative mitochondrial morphology in the control condition ., For squamocin-treated cells , 45% are closer to their own population , while 30% are closer to the control population ., This indicates that the CHO cells in this research responded heterogeneously when treated by squamocin ., Our unpublished data show that longer treatment duration will not change the percentage of cells affected by squamocin and that the effect of squamocin examined in this study is its terminal effect ., Cells treated with z-IETD after squamocin are characterized by straight tubules rescued from squamocin-induced mitochondrial fission , as can be observed from the representative cells in the z-IETD population in Figure 6C ., These cells comprise about 47% of the z-IETD population , much larger than the percentage of cells that are closer to other populations ., The morphological composition appears to be restored partially to that of the control ., In contrast , treatment with z-LEHD after squamocin is more effective for rescuing tubular mitochondria from squamocin-induced fission , as 54% of the z-LEHD population are closer to the control population , while only about 15% are representative ., Therefore , the relatively large number of twisted tubules and loops observed in representative cells in the z-LEHD population ( shown in Figure 6D ) may represent only a transitional morphology during tubule restoration by z-LEHD ., In summary , the mitochondrial morphologies of representative cells in the control , squamocin and z-IETD populations are all unique and specific to their respective treatment conditions ., While in the z-LEHD population , a majority of cells are restored by z-LEHD with the mitochondrial morphology similar to control condition ., Representative cells in z-LEHD population are the minority and may be special cases of intermediate responses ., Squamocin is a potent inhibitor which blocks mitochondrial functions and causes oxidative stress 15– ., Our results in general show that cells treated with squamocin contain more small globules or fragmented mitochondria and total mitochondrial area is reduced ., The squamocin-induced mitochondrial morphological change can be partially rescued by inhibiting Caspases 8 and 9 with z-IETD and z-LEHD treatments , respectively , with z-LEHD being more effective ., These results provide the first evidence that Caspases 8 and 9 are directly involved in mitochondrial dynamics ., Another interesting finding is that after cells are treated by squamocin , applying inhibitors of Caspases 8 and 9 can rescue most of the tubular mitochondria with the exception of branched ones ., According to our analysis and previous investigations , we proposed a hypothetical model of squamocin-induced mitochondrial morphological changes , as shown in Figure 8 ., We discuss these results in more detail in the following sections ., In this study , we developed MicroP for automatic classification and quantification of mitochondrial morphology in cell micrographs , which helped us confirm the number and range of representative morphological subtypes , the effects of treatments on the ratio of subtypes in cells , and make sense of the subpopulations within heterogeneous cell responses to different drug treatments ., The main contributions of our automated system in this study are summarized as follows: First , our computational method allows objective subtyping and automatic quantification of mitochondrial morphology in cell micrographs , which enables profiling of cell responses to drug treatments ., Second , using multidimensional scaling and silhouette coefficients to characterize cell response profiles , we discovered that Caspases suppress elongation fusion of mitochondria but not branching ., Our quantification analysis also differentiates the effects of Caspases 8 and 9 inhibitors on squamocin-treated cells ., For example , Caspase 9 inhibitor ( z-LEHD ) rescues longer mitochondria and results in larger numbers of twisted and looped mitochondria than Caspase 8 inhibitor ( z-IETD ) ., Given the heterogeneity of cell responses to drug treatments , it would be challenging to reach these findings if not for the subtyping and quantitative analysis ., Finally , our correlation analysis of subtype ratios within individual cells reveals unexpected trends which provide directions for further investigations ., For example , the stronger negative correlations of branched tubule mitochondria with small globules than those of other tubule subtypes suggests a higher fission rate of branched mitochondria than other subtypes of tubular mitochondria ., Our future work is to simultaneously measure cell viability and morphological subtypes of mitochondria to reveal their correlation and to validate the proposed pathway model with further biochemical and cell biological experiments ., We will utilize 3D time-lapsed imaging to provide solid evidence to study the transition of mitochondrial subtypes during different treatments , and to verify our interpretations of the correlations of mitochondrial morphological subtypes ., Our long-term goal is to investigate whether morphological features of mitochondria are specific to neurodegenerative diseases or aging and evaluate the use of mitochondrial morphological features as “high content” biomarkers ., CHO-K1 cells were obtained from the Food Industry Research and Development Institute ( Hsinchu , Taiwan ) ., CHO-K1 cells were cultured in McCoy 5A containing 10% fetal bovine serum and incubated in an incubator containing 5% at ., For cell imaging , CHO-K1 cells expressing DsRed-mito ( cells ) were seeded on a 24mm round coverslip ( thickness: 0 . 17mm ) ., After 24h culture , cells were pre-treated with z-IETD ( Caspase 8 inhibitor; Sigma , U . S . A . ) , z-LEHD ( Caspase 9 inhibitor; Sigma , U . S . A . ) , or a control medium without any drugs for 2h ., Cells pre-treated with control medium were further incubated with either 0 . 05% DMSO or squamocin for 24h ., Cells pre-treated with Caspase inhibitors were further treated with squamocin for 24h ., The following labels are used to refer to the different treatments in subsequent sections: DMSO ( DMSO + control ) , Squamocin ( squamocin + control ) , z-IETD ( Squamocin + z-IETD ) and z-LEHD ( squamocin + z-LEHD ) ., The coverslip attached with cells was put on a chamber and examined with a fluorescence microscope ( IX-71 , Olympus ) with a objective ( PlanApo , NA1 . 45 , Olympus ) ., Monochromator equipped with Xe-lamp ( polychrome II , Till-photonics , Gräfelfing , Germany ) was driven by TILLVision 4 . 0 ( Till-Photonic , Gräfelfing , Germany ) to excite DsRed-mito ( 550 nm ) ., Fluorescence was filtered by a filter cube ( Mitotracker orange: 565DCLP ( BS ) , D605/55m ( Em ) ; Chroma , Rockingham , Vermont , USA ) , and 2D fluorescent cell images were acquired by a CCD camera ( IMAGO , Till-Photonics Germany; exposure: 500ms; resolution: 12-bit , pixels , pixel size: ) ., The images are scaled down to 8-bit for image analysis , and pixel width is about 165nm ., We have compared 2D ( epi-fluorescence microscopy ) and 3D ( confocal fluorescence microscopy ) CHO micrographs , and found that CHO cells are flat and most of the mitochondria in 2D micrographs are in focus and clear enough for high content image analysis ( see Section S2 . 1 in Text S1 ) ., Therefore , 2D micrographs are the major data source used in this study ., Segmentation of individual cells within each micrograph field was done semi-automatically ., First the centroid of each cell nucleus was specified manually , which indicated the position of each individual cell object ., Then the Delaunay triangulation was calculated using all cell centroids and their dual Voronoi diagram 44 was used as the final cell segmentation ., The resulting single-cell images were manually validated and grouped by drug treatments ., In the resulting set , the DMSO group contains 178 cell images , the Squamocin group 357 cell images , the z-IETD group 454 cell images , and z-LEHD group 433 cell images , for a total of 1422 single cell images ., Mitochondria extraction involved segmenting each single-cell image into mitochondria and background ., Cell micrographs of mitochondria exhibit varying background brightness and contrast levels , hence proper segmentation needs to take into account the statistical properties associated with each locality in the image ., Here , we used our segmentation algorithm described in 20 ., In this algorithm , adaptive local normalization is used to preprocess the images ( with parameter value ) and then Otsus thresholding 21 was applied to the normalized image to obtain the final segmentation ., Adaptive local normalization applies dynamic window sizes determined by the intensity structure of each pixel region to effectively deal with local contrast and background variation , and at the same time enhance detailed subcellular structures ., This segmentation algorithm was quantitatively validated using manually generated gold standard segmentations from the same dataset used in this work with approximately equal amount of images from each treatment 20 ., All the CHO cells from the four treatment conditions were quite flat and easily focused on a 2D focal plane ., Seriously out-of-focus and blurred mitochondria generally only constituted a small fraction of all mitochondria in individual cells ., Cells treated by squamocin were a little bit rounder , but their mitochondria were still focused well on a single plane ( as shown in Figures S14-16 in Text S1 ) ., Visual examination showed that cells from four treatment conditions were roughly equally focused on average , and there were no seriously out-of-focus cells in this dataset ., To further test the robustness of our segmentation method , we analyzed representative subsets of cells from each treatment condition with perturbed threshold values ( data not shown ) ., For most cell features , such as total mitochondrial area and relative ratios of morphological subtypes , the results are consistent across different threshold values ., After segmentation , binary object images representing mitochondria were extracted by standard object labeling with 4-neighbor connectivity ., Postprocessing was performed to remove objects with low intensity ( both in the original or normalized images ) and small area , as well as any objects touching the image boundary ., Figure S17a-c in Text S1 illustrate the results of these processing steps on an example micrograph ., A total of 225 , 556 objects/mitochondria were obtained from segmenting the single-cell images , of which 27 , 752 are in the DMSO group , 66 , 438 in the Squamocin group , 67 , 288 in the z-IETD group , and 64 , 078 in z-LEHD group ., From each segmented binary mitochondrion object , a set of image features was extracted to represent its morphology ., These image features can be divided into three categories: morphological features based on the object binary mask , skeleton features based on one-pixel wide homotopic skeleton , and binary texture features based on the object bounding convex hull ., Table S1 in Text S1 contains a summary of these features and their notations , for more details please refer to Section S2 . 3 in Text S1 ., These mitochondrial features are used for both consensus clustering and classification , described in subsequent sections ., To identify meaningful morphological subtypes of mitochondria , the Gaussian mixture model ( GMM ) clustering was applied to cluster mitochondrial objects with the optimal number of clusters determined by Bayesian information criterion ( BIC ) ., We used the GMM implementation in the MATLAB Statistics Toolbox with default values for all parameters ., We performed multiple clustering runs on sampling-with-replacement random subsets of objects to ensure the robustness of the clustering ., Processing small subsets also

Introduction, Results, Discussion, Materials and Methods

Morphological dynamics of mitochondria is associated with key cellular processes related to aging and neuronal degenerative diseases , but the lack of standard quantification of mitochondrial morphology impedes systematic investigation ., This paper presents an automated system for the quantification and classification of mitochondrial morphology ., We discovered six morphological subtypes of mitochondria for objective quantification of mitochondrial morphology ., These six subtypes are small globules , swollen globules , straight tubules , twisted tubules , branched tubules and loops ., The subtyping was derived by applying consensus clustering to a huge collection of more than 200 thousand mitochondrial images extracted from 1422 micrographs of Chinese hamster ovary ( CHO ) cells treated with different drugs , and was validated by evidence of functional similarity reported in the literature ., Quantitative statistics of subtype compositions in cells is useful for correlating drug response and mitochondrial dynamics ., Combining the quantitative results with our biochemical studies about the effects of squamocin on CHO cells reveals new roles of Caspases in the regulatory mechanisms of mitochondrial dynamics ., This system is not only of value to the mitochondrial field , but also applicable to the investigation of other subcellular organelle morphology .

Mitochondria are “cellular power plants” that synthesize adenosine triphosphate ( ATP ) from degradation of nutrients , providing chemical energy for cellular activities ., In addition , mitochondria are involved in a range of other cellular processes , such as signaling , cell differentiation , cell death , cell cycle and cell growth ., Dysfunctional mitochondrial dynamics have been linked to several neurodegenerative diseases , and may play a role in the aging process ., Previous studies on the correlation between mitochondrial morphological changes and pathological processes involve mostly manual or semi-automated classification and quantification of morphological features , which introduces biases and inconsistency , and are labor intensive ., In this work we have developed an automated quantification system for mitochondrial morphology , which is able to extract and distinguish six representative morphological subtypes within cells ., Using this system , we have analyzed 1422 cells and extracted more than 200 thousand individual mitochondrion , and calculated morphological statistics for each cell ., From the numerical results we were able to derive new biological conclusions about mitochondrial morphological dynamics ., With this new system , investigations of mitochondrial morphology can be scaled up and objectively quantified , allowing standardization of morphological distinctions and replicability between experiments ., This system will facilitate future research on the relation between subcellular morphology and various physiological processes .

computer applications, computer science, biology, computational biology

journal.ppat.1001238

Interaction of c-Cbl with Myosin IIA Regulates Bleb Associated Macropinocytosis of Kaposis Sarcoma-Associated Herpesvirus

KSHV is etiologically associated with Kaposis sarcoma ( KS ) , the most common AIDS related malignancy , as well as with two lymphoproliferative diseases , primary effusion lymphoma ( PEL ) and multicentric Castlemans disease 1 , 2 ., KSHV infects a variety of target cells both in vivo and in vitro ., Entry into the target cells is the most crucial step in the establishment of a successful infection for all viruses ., KSHV utilizes different modes of endocytosis to enter different target cells in vitro 3 ., For example , KSHV enters human foreskin fibroblasts ( HFF ) via clathrin mediated endocytosis and enters HMVEC-d cells via macropinocytosis 3 , 4 , 5 ., During the early stages of infection of HMVEC-d cells , KSHV forms a multi-molecular complex with host cell heparan sulfate , integrins ( α3β1 , αVβ3 and αVβ5 ) and transporter protein xCT with the subsequent induction of overlapping signal cascades 3 ., Our studies show that KSHV induces a complex set of signaling molecules that are involved in diverse biological functions to regulate the various aspects of KSHV endocytosis including internalization , trafficking in the cytoplasm and nuclear delivery 3 ., KSHV activates FAK , Src , PI3-K , Rho-GTPases and cytoskeleton rearrangement which are all critical for entry of virus 6 , 7 , 8 , 9 ., KSHV also activates other downstream molecules such as PKC-ζ , MEK , ERK1/2 and NFkB which are essential for viral gene expression 6 , 7 , 8 , 9 ., The Cbl family of adaptor proteins include three mammalian isoforms , c-Cbl , Cbl-b and Cbl-c or Cbl-3 10 , 11 ., Cbl proteins play important roles in signal transduction as negative regulators by mediating the ubiquitinilation and down-regulation of proteins while it acts as a positive regulator through their scaffold function in assembling signaling complexes 10 , 11 ., c-Cbl has been shown to bind to several molecules critical in signal transduction 10 , 11 ., Tyrosine phosphorylation of c-Cbl has been shown to be crucial for c-Cbl mediated adaptor functions in most circumstances 11 , 12 , 13 ., However , the adaptor functions of c-Cbl and a c-Cbl mediated signaling pathway during virus infection has not been demonstrated ., Macropinocytosis provides a major route for the productive infection of many viruses including KSHV ., Macropinocytosis is an actin dependent membrane associated process which involves recruitment and integration of several signaling molecules necessary for cytoskeletal rearrangement and membrane remodeling ., However , there is little information about the molecules involved in the recruitment and integration of signaling during macropinosome formation ., Even though c-Cbl has been shown to recruit and link different signaling molecules in a signaling pathway , a direct role for c-Cbl in the process of macropinocytosis has not been established yet ., Here we identified that c-Cbl is involved in KSHV entry and critical for triggering the macropinocytic event ., Our data provide evidence that the interaction between c-Cbl and myosin IIA , a motor protein that binds to the proline rich domain of c-Cbl , regulates macropinocytosis of KSHV ., This study on the functional organization of the c-Cbl and myosin IIA complex and its effect on viral entry provide an important insight into understanding the role of c-Cbl in virus infection ., HMVEC-d cells ( CC-2543; Clonetics , Walkersville , Md ) were grown in endothelial cell medium ( EBM2; Cambrex , Walkersville , MD ) ., Induction of the KSHV lytic cycle in BCBL-1 cells , supernatant collection , and virus purification procedures were described previously 14 ., KSHV DNA was extracted from the virus , and the copies were quantitated by real-time DNA PCR using primers amplifying the KSHV ORF 73 gene as described previously 15 ., A pool of lentivirus shRNA specific for human c-Cbl and non-specific control shRNA were purchased from Santa Cruz Biotechnology ( Santa Cruz , CA ) ., HMVEC-d cells were transduced with control lentivirus shRNA and c-Cbl lentivirus shRNA according to the manufacturers instructions and selected by puromycin hydrochloride ( 10 µg ml−1; Santa Cruz Biotechnology ) ., The following antibodies were used: mouse anti-c-Cbl , mouse anti-phospho Cbl 700 ( phosphorylated at Tyr700 ) , and mouse anti-p85 ( PI-3K ) antibodies ( BD Transduction Laboratories , San Diego , CA ) ; anti-phospho MLC II , anti-phospho Cbl 731 , anti-phospho Cbl 774 ( phosphorylated at Tyr731 and Tyr774 ) , isoform specific anti-myosin II heavy chain antibodies myosin IIA , IIB and IIC ( Cell Signaling Technology , Danvers , MA ) ; mouse anti-phospho tyrosine ( 4G10 clone; Millipore , Temecula , CA ) ; mouse anti-tubulin , mouse anti-beta actin antibodies ( Sigma , St Louis , MO ) ; rabbit anti-lamin B ( Abcam , Cambridge , MA ) ; rabbit anti-HA ( Zymed , Invitrogen , Carlsbad , CA ) ; mouse anti-ubiquitin ( P4D1 ) , mouse ant-GFP , mouse anti-GST ( Santa Cruz , CA ) ; rabbit anti-gB and mouse anti-gpK8 . 1A antibodies were created in our laboratory16 , 17; anti-goat , anti-rabbit and anti-mouse antibodies linked to horseradish peroxidase ( KPL Inc . , Gaithersburg , Md . ) ; DAPI , rhodamine conjugated dextran , Alexa 594 or Alexa 488 conjugated phalloidin and anti-rabbit and anti-mouse secondary antibodies conjugated to Alexa 488 , Alexa 594 ( Invitrogen ) ; protein A and G–Sepharose CL-4B beads ( Amersham Pharmacia Biotech , Piscataway , NJ ) ; blebbistatin , U0126 ( Calbiochem , La Jolla , CA ) ; TPA , LY294002 ( Sigma ) ., Unless stated otherwise , cells were infected with KSHV at 10 DNA copies ( multiplicity of infection MOI ) per cell at 37°C ., Entry was measured by infecting the cells with KSHV for 30 min ., The cells were washed with HBSS to remove the unbound virus , treated with 0 . 25% trypsin-EDTA for 5 min at 37°C to remove the bound but non-internalized virus , and washed ., Cells were recovered by centrifugation and total DNA was isolated from infected or uninfected cells using a DNeasy kit ( QIAGEN , Valencia , CA ) as described previously 15 ., To calculate percent of inhibition of KSHV entry , internalized KSHV DNA was quantitated by amplification of the ORF73 gene by real-time DNA PCR 15 ., The KSHV ORF73 gene cloned in the pGEM-T vector ( Promega ) was used for the external standard ., The cycle threshold ( Ct ) values were used to generate the standard curve and to calculate the relative copy numbers of viral DNA in the samples ., Percentage inhibition was calculated by considering the ORF73 copy numbers in untransduced cells as 100% ., Total RNA was prepared from infected or uninfected cells using an RNeasy kit ( QIAGEN ) as described previously 15 ., To quantitate viral gene expression , isolated RNA was subjected to ORF73 and ORF50 RNA expression by real-time reverse transcription ( RT ) -PCR using gene specific real-time primers and specific TaqMan probes 15 ., The relative copy numbers of the transcripts were calculated from the standard curve plotted using the Ct values for different dilutions of in vitro-transcribed transcripts ., These values were normalized to each other using the values of the GAPDH control reactions ., Percentage inhibition was calculated by considering ORF73 and ORF50 gene expression in untransduced cells as 100% ., Cells were lysed in RIPA buffer ( 15 mM NaCl , 1 mM MgCl2 , 1 mM MnCl2 , 2 mM CaCl2 , 2 mM phenylmethylsulfonyl fluoride , and protease inhibitor mixture ( Sigma ) ) and centrifuged at 12 , 000 rpm at 4°C for 15 min ., Lysates were normalized to equal amounts of protein and the proteins were separated by 7 . 5–12 . 5% gradient SDS-PAGE , transferred to nitrocellulose and probed with the indicated primary antibodies ., Detection was by incubation with species-specific HRP-conjugated secondary antibodies ., Immunoreactive bands were visualized by enhanced chemiluminescence ( Pierce , Rockford , IL ) according to the manufacturers instructions ., The bands were scanned and quantitated using the FluorChem FC2 and Alpha-Imager Systems ( Alpha Innotech Corporation , San Leonardo , CA ) ., Two hundred micrograms of cell lysates prepared as described in the above section were incubated for 2 h with immunoprecipitating antibody at 4°C , and the immune complexes were captured by protein A or G-Sepharose ., The samples were tested by Western blot with specific primary and secondary antibodies ., HMVEC-d cells were infected with KSHV for different time points ., The samples were resolved on an SDS-PAGE gel and the gel was stained with Coomassie blue ., The bands of interest were excised , digested with trypsin , separated by reverse phase nano-chromatography and analyzed by mass spectrometry ., Immunofluorescence assay was performed using HMVEC-d cells seeded on 8 well chamber slides ( Nalge Nunc International ) ., Infected and uninfected cells were fixed with 3% paraformaldehyde for 15 min , permeabilized with 0 . 2% Triton X-100 , and blocked with Image-iTFX signal enhancer ( Invitrogen ) ., The cells were then immunostained with primary antibodies against the specific proteins , followed by fluorescent dye-conjugated secondary antibodies ., For colocalization with dextran and transferrin , cells were incubated with the fluid-phase marker dextran Texas Red ( 40 kD , 0 . 5 mg ml−1; Invitrogen ) or Alexa 594 transferrin ( 35 µg ml−1; Invitrogen ) at 37°C in the presence or absence of KSHV followed by immunostaining with the appropriate antibodies ., Cells were imaged with a Nikon fluorescence microscope equipped with a Metamorph digital imaging system ., DIC ( Differential Interference Contrast ) images were acquired with objectives equipped with DIC optics ., For confocal analysis , the Olympus Fluoview 300 fluorescence confocal microscope was used for imaging , and analysis was performed using Fluoview software ( Olympus , Melville , NY ) ., All experiments were performed at least three times ., HMVEC-d cells , incubated with dextran Texas Red ( 0 . 5 mg ml−1 , 40 kD; Invitrogen ) and KSHV for 30 min , were washed twice in HBSS ., To remove surface bound dextran , cells were treated with 0 . 25% trypsin-EDTA and the cells were harvested ., Quantitative analysis of dextran uptake was determined by counting the number of cells stained positive for dextran under immunofluorescence microscope ., At least 10 different microscopic fields of 50 cells each were counted for each experiment and the results displayed as percentage of dextran positive cells ., Flow cytometry analysis was used to quantify the uptake of dextran during KSHV internalization in control shRNA and c-Cbl shRNA transduced cells ., Cells were incubated with 500 µg ml−1 FITC-dextran in the presence or absence of virus at 37°C for 30 min ., The cells were washed , harvested using trypsin EDTA , fixed and analyzed by flow cytometry ., Mean fluorescence intensity was determined using a Becton Dickinson FACS system and CellQuest software ., Cells incubated with dextran alone were used as controls ., HeLa cells ( ATCC CCL-2 ) were cultured in DMEM containing 10% fetal bovine serum ., Wild type , mutants and deletion constructs of c-Cbl , c-Cbl C-terminal domain encompassing PRD ( Cbl-C ) and c-Cbl N-terminal domain ( Cbl-N ) constructs were generously provided by Dr . Hamid Band 18 ( Eppley Institute for Cancer and Allied Diseases , University of Nebraska Medical Center ) ., Cells were transiently transfected with wild type , mutants and deletion constructs ., Transfection was performed using 5 µg of plasmid DNA , lipofectamine 2000 ( Invitrogen ) , and Opti-MEM medium ( Invitrogen ) according to the manufacturers instructions ., After transfection , cells were cultured for 48 h ., Cells were then serum starved for 4 h and stimulated with TPA ( 100 ng ml−1 ) at 37°C for 5 min ., Lysis was performed in RIPA buffer plus protease inhibitors ., The cell lysate was used for immunoprecipitation and immunoblotting ., E . coli BL21 ( DE3 ) cells were transformed with pGEX4T . 1, GST-Cbl C ( Cbl residues 358–906 ) plasmids which encode regions encompassing the C-terminal PRD domain of c-Cbl and pGEX4T . 1, GST-Cbl N which encodes the N-terminal region of c-Cbl ., Expression of the GST-Cbl fusion proteins was induced with IPTG ( isopropyl-D-1-thiogalactopyranoside ) 1 mM for 3 h at 37°C ., The bacterial lysates ( 500 µg ) were incubated with glutathione-sepharose beads ( GE Healthcare , U . K . ) for 2 h at 4°C ., The beads were washed with lysis buffer three times ., 293T cells were transiently transfected with 2 µg pEGFP C3 myosin IIA plasmids ( Addgene ) ., After 48 h of transfection , cells were lysed in RIPA buffer and 500 µg of the lysates were incubated with the glutathione-Sepharose beads bound with the GST-Cbl fusion proteins ., The beads and the bound proteins were collected by centrifugation , washed and the interaction of GST-Cbl with myosin IIA was analyzed by SDS-PAGE and Western blotting using anti-GFP antibody ., HMVEC-d cells infected with KSHV for 5 min were fixed and stained with DAPI ., DIC images were acquired and the cells presenting blebs or no blebs were counted visually ., At least 10 random microscopic fields per experiment were counted and expressed as a proportion of the total number of DAPI stained cells ., Infected and uninfected cells were washed three times with HBSS and lysed in homogenization buffer ( 250 mM sucrose , 20 mM HEPES , 10 mM KCL , 1 mM EDTA , 1 mM EGTA and protease inhibitors ) ., The homogenate was subjected to centrifugation at 3 , 000 rpm for 5 min ., Post-nuclear supernatant was centrifuged at 8 , 000 rpm for 5 min at 4°C ., The supernatant was again centrifuged at 40 , 000 rpm for 1 h at 4°C , and the supernatant and the pellet were considered the cytosolic and the membrane fractions , respectively ., The membrane pellet was solubilized using RIPA buffer and used for Western blot ., To determine whether c-Cbl and the c-Cbl mediated signaling pathway play roles in KSHV infection , we first examined the early tyrosine phosphorylation kinetics of c-Cbl in KSHV infected cells ., HMVEC-d cells infected with KSHV induced rapid tyrosine phosphorylation of c-Cbl , which was detectable as early as 1 min post-infection ( p . i . ) , reaching maximum levels at 5 min ( 4 . 1-fold ) , followed by a decrease which was constent for as much as 30 min p . i . ( Figure 1a ) ., To determine whether the phosphorylation of c-Cbl is specifically induced by KSHV , cells were infected with KSHV pre-incubated with heparin ., Heparin is known to block the binding of KSHV to the target cells 19 ., Compared to the untreated virus , heparin treated virus considerably reduced the phosphorylation of c-Cbl ( Figure 1a ) which demonstrated the specificity of KSHV induced c-Cbl phosphorylation ., The efficient tyrosine phosphorylation of c-Cbl is suggestive of the possible involvement of a c-Cbl mediated signaling pathway in KSHV infection ., We next investigated the link between c-Cbl phosphorylation and other signaling molecules activated during KSHV infection ., It is well documented that the interaction of KSHV glycoproteins with integrins and other cellular receptors activate FAK and the downstream molecules Src and PI3-K 7 , 9 , 14 , 20 , 21 ., c-Cbl has been shown to form a complex with PI3-K p85 in the integrin mediated signaling pathway 12 ., We therefore examined whether the association of PI3-K with c-Cbl occurred during KSHV infection ., KSHV infection led to an increase in the interaction of c-Cbl with PI3-K p85 in a time dependent manner ( Figure 1b ) ., To verify that the c-Cbl-PI3-K interaction is specifically induced by virus , cells were infected with heparin treated virus which notably decreased the association of c-Cbl with PI3-K ( Figure 1b ) ., The c-Cbl-PI3-K association was further confirmed by confocal analysis ( Figure 1c ) ., Consistent with previous studies 13 , our results demonstrated that activated c-Cbl leads to the association of c-Cbl with PI3-K ., Previous studies have shown that ERK1/2 is activated during KSHV infection and is a key signaling molecule implicated in viral gene expression 22 ., To examine whether an ERK1/2 associated pathway is involved in c-Cbl mediated signaling , we investigated the association of ERK1/2 with c-Cbl in KSHV infected cells ., No colocalization was observed between ERK1/2 and c-Cbl ( Figure 1c ) which suggested that the ERK associated pathway is not involved in c-Cbl mediated signaling in KSHV infected cells ., Taken together , our data suggests that a signaling complex which contains c-Cbl and PI3-K but not ERK1/2 is involved in the integrin mediated signaling pathway of KSHV infection ., To further demonstrate the relationship between the interaction of c-Cb1 with PI3-K but not with ERK1/2 , we studied the effect of PI3-K and ERK1/2 inhibitors in KSHV induced c-Cbl phosphorylation ., HMVEC-d cells pretreated with the PI3-K inhibitor , LY294002 , and the ERK1/2 inhibitor , U0126 were infected with KSHV for 10 min and the lysates were analyzed for c-Cbl phosphorylation ., KSHV induced c-Cbl phosphorylation was abolished by the PI3-K inhibitor LY294002 , whereas the ERK1/2 inhibitor U0126 did not show any inhibition on c-Cbl phosphorylation ( Figure 1d ) ., The failure of ERK1/2 inhibitor to abolish c-Cbl phosphorylation confirmed that ERK1/2 is not associated with c-Cbl induction in KSHV infected cells ., We next used c-Cbl lentivirus encoding shRNAs to knockdown c-Cbl activity in HMVEC-d cells ( c-Cbl shRNA cells ) to analyze the functions of c-Cbl in KSHV infected cells ., The c-Cbl specific shRNA inhibited 90% of c-Cbl expression as detected by Western blotting with antibodies to c-Cbl ( Figure S1a ) ., Untransduced , control shRNA and c-Cbl shRNA transduced cells were infected with KSHV for 2 and 24 h , and viral gene expression was determined by real-time RT-PCR analysis ., Compared with control cells , c-Cbl shRNA transduced cells showed about 60–70% inhibition of the latency associated ORF 73 gene ( Figure 2a ) and 70–80% inhibition of the lytic switch ORF 50 gene ( Figure 2b ) expression ., We next determined whether the inhibition of viral-gene expression by c-Cbl shRNA was due to a blockage at the entry stage of the virus ., To determine c-Cbls role in KSHV entry , internalization of viral DNA was determined by measuring viral ORF 73 DNA copy numbers by real-time DNA PCR ., We observed ∼65% inhibition of KSHV entry in c-Cbl shRNA cells compared to control cells ( Figure 2c ) ., Internalized KSHV ORF 73 DNA copy numbers and ORF 50 and ORF 73 RNA copy numbers are shown as histograms in supplementary Figure S1 ., Taken together , these studies demonstrated that the decreased viral gene expression observed in c-Cbl shRNA cells was due to a decrease in the entry of KSHV ., These results further suggested that a c-Cbl containing signaling complex may be crucial for the initiation of entry and for a productive infection ., In our earlier studies we have demonstrated that macropinocytosis is the major pathway of KSHV entry leading to a productive infection in HMVEC-d cells 4 ., Since c-Cbl inhibited KSHVs entry , we theorized that c-Cbl might be playing a role in macropinocytosis and associated signaling events ., Viruses such as vaccinia virus that use macropinocytosis as a mode of entry induce signaling molecules and cytoskeletal rearrangements in the form of blebs which ultimately retract and ingest viral particles 23 , 24 ., To determine whether blebs were involved in KSHV infection , we used DIC image analysis and observed the association of KSHV with blebs ., As shown in Figure 3a , bleb formation and the association of individual blebs with KSHV was observed by 5 min p . i . ., The DIC microscopic analysis of a single bleb for viral particles confirmed that the blebs , formed during viral infection , were associated with viral particles ( Figure 3b ) ., To further investigate whether KSHV infection induces blebbing , HMVEC-d cells were infected with KSHV for 2 and 5 min and then actin , a well known determinant of cell shape and blebbing , was stained with phalloidin 25 ., Within 2 min of infection , membrane protrusions appeared along the cell surface which rapidly enlarged into well-formed blebs at 5 min ( Figure S2 ) ., We also observed the association of viral particles at the bleb forming site as well as with retracting blebs ( Figure S2 ) ., Unlike the well formed blebs , the retracting blebs were characterized by a thick actin cortex 26 ., These results demonstrated that early during infection , KSHV induces actin reorganization and the subsequent formation of blebs that may be involved in its entry ., Since we observed that c-Cbl shRNA inhibited KSHV entry , which involves bleb formation , we hypothesized that c-Cbl and its phosphorylation might be involved in the dynamics of virus induced blebbing ., To understand the function of c-Cbl in blebbing , we examined the localization of phosphorylated c-Cbl ( p-Cbl ) in KSHV infected cells ., Confocal microscopy analysis showed that KSHV induced p-Cbl localized to blebs as early as 2 min p . i . ( data not shown ) ., Bleb formation , and its association with p-Cbl , was maximal at 5 min , and by 10 min blebs containing p-Cbl started internalizing and p-Cbl was mostly observed at the nuclear periphery by 15 min p . i . ( Figure 3c ) ., A similar pattern of localization was exhibited by p-Cbl and virus at the blebs as early as 5 min p . i . , and accumulation of p-Cbl and virus around the nuclear periphery was observed at 15 min p . i . ( Figure 3d ) ., These results suggested that the recruitment of phosphorylated c-Cbl to the sites of bleb formation was involved in bleb associated entry of the virus ., To further explore the role of c-Cbl in bleb associated macropinocytosis , we performed a confocal immunofluorescence colocalization study between p-Cbl and the macropinocytosis marker dextran in infected cells ., This analysis showed that dextran colocalized with p-Cbl at 10 min p . i . ( Figure 4a ) in the infected cells ., Next , we performed a dextran uptake study since the uptake of dextran has been used as a biochemical marker of macropinocytosis ., We incubated the cells with dextran in the presence or absence of virus for 30 min and then quantitated the level of uptake ., As shown in Figure 4b and supplementary information Figure S3a , c-Cbl shRNA cells showed a drastic inhibition of dextran uptake compared to control shRNA cells infected with KSHV ., This indicated that the uptake of dextran or macropinocytosis in KSHV infected cells was a c-Cbl dependent process ., The uptake of dextran and colocalization with KSHV in non-specific control shRNA and c-Cbl shRNA cells were confirmed by immunofluorescence colocalization and DIC analysis ., In control shRNA cells infected with KSHV , intracellular KSHV was highly colocalized with dextran , whereas in c-Cbl shRNA cells infected with KSHV , most of the viral particles remained at the membrane periphery although minimal colocalization of KSHV with dextran was observed in some cells ( Figure 4c and Figure S3b ) ., Control shRNA cells incubated with KSHV and Alexa 594 transferrin , a marker for clathrin-mediated endocytosis , did not show any significant colocalization with KSHV ( Figure 4c and Figure S3b ) which demonstrated the specificity of macropinocytosis mediated entry in HMVEC-d cells 4 ., These results were consistent with the results of the dextran uptake study , confirming that c-Cbl was critical for inducing the macropinocytic process that promoted the internalization of KSHV ., The uptake of dextran in control shRNA and c-Cbl shRNA cells was further quantified by FACS analysis ., Cells were incubated with dextran in the presence or absence of virus for 30 min and the uptake was measured using flow cytometry ., As shown in Figure 4d , compared to the control shRNA cells , c-Cbl shRNA cells showed a notable decrease in mean fluorescence intensity ., These results are consistent with the results of the immunofluorescence analysis and thus confirmed the role of c-Cbl in KSHV induced macropinocytosis ., c-Cbl is a multi-domain protein that interacts with a number of signaling molecules and performs multiple functions 10 , 11 ., To decipher the molecular partners interacting with c-Cbl during KSHV infection , we used mass spectrometric analysis ., HMVEC-d cells were infected with KSHV for 1 , 5 and 10 min , lysed and the lysates were immunoprecipitated with anti-c-Cbl antibodies ., Samples were separated by SDS-PAGE , followed by Coomassie blue staining and mass spectrometry analysis ., Mass spectrometry identified several novel c-Cbl interacting proteins in the infected samples ( Supplementary Table S1 ) ., The most prominent protein identified in the infected samples was myosin IIA which is one of three isoforms of the non-muscle myosin II family of proteins 27 , 28 ., The other novel interacting partners of c-Cbl in the infected cells included vimentin , HSP70 , BiP protein , Rho GEF and a solute carrier anion exchanger ( Table S1 ) ., To confirm the mass spectrometry data , uninfected and KSHV infected cell lysates were immunoprecipitated with anti-c-Cbl antibody and blotted for the three isoforms of the non-muscle myosin II family , IIA , IIB and IIC , with isoform specific antibodies ., Our results confirmed that c-Cbl interacts with myosin IIA in the lysates of infected cells , whereas the other isoforms did not show any interaction with c-Cbl ( Figure 5a ) ., To elucidate the functional domain of c-Cbl involved in myosin IIA interaction , a series of truncated and mutant constructs of c-Cbl with HA epitope tags were used ( Figure S4 ) ., Since HMVEC-d cells are not easily transfectable , we used HeLa cells for this study ., HeLa cells were transfected with vector alone , Cbl wild-type , Cbl-tyrosine kinase binding domain ( TKB ) mutant , RING domain mutant , and two truncation mutants ( Cbl-Δ357 and Cbl-Δ421 ) ., As TPA ( phorbol ester ) has been shown to induce membrane blebbing 29 , we used TPA induced HeLa cells to analyze the interaction of over-expressed c-Cbl with endogenous myosin IIA ., Transfection of the Cbl-TKB mutant and RING mutant induced the interaction of c-Cbl with myosin IIA similar to full length wild-type Cbl ., The truncated versions Cbl-Δ357 and Cbl-Δ421 lacking a C-terminal proline rich domain ( PRD ) decreased the interaction with myosin IIA considerably ( Figure 5b ) ., Expression of all constructs determined by Western blotting with HA revealed comparable levels of protein ( Figure 5b ) ., Taken together , these results indicated that the C-terminal region encompassing the PRD of c-Cbl was sufficient for association with myosin IIA ., To further confirm that the C-terminal PRD of c-Cbl interacts with myosin , an in vitro binding assay was performed using bacterially expressed GST fusion proteins of c-Cbl C-terminal ( Cbl-C , encompassing PRD ) and N-terminal ( Cbl-N ) domains ., GST Cbl-C and Cbl-N proteins adsorbed on glutathione sepharose beads were incubated with 293T cell lysates expressing GFP-tagged myosin IIA ., The interaction between GFP-myosin IIA and GST-Cbl was analyzed by Western blotting with anti-GFP antibody ., Our results demonstrated that myosin IIA predominantly interacted with Cbl-C domains compared to Cbl-N ( Figure 5c ) ., The interaction of myosin IIA with c-Cbl suggested that their association could be playing a role in blebbing and macropinocytosis of KSHV ., To investigate this , we used blebbistatin , a specific inhibitor of myosin II ATPase activity that has been shown to inhibit myosin II induced blebbing 30 , 31 and macropinocytosis 23 ., As shown in Figure 6a , we observed a dose dependent inhibition of KSHV internalization in 25 µM ( ∼35% ) and 50 µM ( ∼60% ) concentrations of blebbistatin indicating that the entry process was dependent on myosin II activity ., As reported previously 4 , chlorpromazine , an inhibitor of clathrin dependent endocytosis , did not show any notable decrease in entry of KSHV ( Figure 6a ) ., These findings suggested that c-Cbl associated myosin IIA was involved in bleb mediated macropinocytosis of KSHV ., To determine whether blebbistatin treatment affects other internalization pathways , we investigated the effect of blebbistatin on clathrin-mediated internalization , a major and well characterized endocytic pathway of eukaryotic cells ., To study this , untreated or blebbistatin treated HMVEC-d cells were induced with FBS in the presence of Alexa 594 labelled transferrin or Texas Red labelled dextran ., The endocytic uptake of transferrin and dextran were then analyzed using immunofluorescence ., As indicated in Figure 6b and d , blebbistatin strongly inhibited the uptake of dextran , whereas the uptake of transferrin was unaffected ( Figure 6c and, e ) ., This demonstrated that blebbistatin specifically inhibits macropinocytosis but not clathrin mediated endocytosis pathways ., The above studies demonstrated that the c-Cbl interacting partner myosin IIA is a biologically significant component of the c-Cbl signaling pathway ., We then explored the role of c-Cbl in myosin II induced blebbing in KSHV infected cells ., If c-Cbl is an upstream molecule of myosin IIA , the loss of function of c-Cbl should prevent the formation of myosin II mediated blebs in c-Cbl shRNA cells ., To test this hypothesis , control shRNA and c-Cbl shRNA transduced HMVEC-d cells were infected with KSHV and the percentage of cells with blebs was quantitated ., As expected , in c-Cbl shRNA transduced cells , the blebs were considerably reduced compared to control shRNA cells ( Figure 7a and b ) ., This suggested that c-Cbl and associated myosin IIA molecules were linked to induce membrane blebbing in HMVEC-d cells ., Our results indicated that c-Cbl plays an upstream role in the regulation of bleb formation which occurs as a result of myosin II induced cortical contractility 25 , 26 ., To further demonstrate that c-Cbl is upstream to myosin IIA , we infected blebbistatin treated cells with KSHV and the membrane localization of c-Cbl was observed by immunofluorescence ., As shown in Figure 7c , blebbistatin did not inhibit the localization of c-Cbl to the plasma membrane , whereas it prevented the formation of blebs in the infected cells ., This suggested that myosin IIA was downstream to c-Cbl and was not involved in the localization of c-Cbl to the plasma membrane ., A subclass of myosins , the class II myosins are hexameric motor proteins composed of two identical heavy chains ( MHC ) , and two pairs of light chains ( MLC ) ., It has been well accepted that phosphorylation of the myosin light chain II is a major determinant of force generation and actomyosin dynamics during apoptotic membrane blebbing 32 , 33 ., Hence , we examined phosphorylation of myosin light chain II ( p-MLC II ) during KSHV infection ., Compared to the uninfected cells , KSHV infection results in rapid and strong phosphorylation of MLC II with maximal phosphorylation at 10 min p . i . ( 5 . 8-fold increase ) and decreased thereafter ( Figure 7d ) ., The specificity of virus induced MLC II phosphorylation was shown using heparin treated virus which did not induce MLC II phosphorylation ( Figure 7d ) ., Since light chain phosphorylation has been shown to regulate blebbing 32 , 33 , our results suggested that KSHV induced MLC II may be participating in the induction of blebbing during infection ., During virus induced and apoptotic membrane blebbing , the signaling molecules associated with cytoskeletal function are recruited to the blebs 23 , 26 ., It has been demonstrated that the recruitment of functional myosin II heavy and light chain complexes drive the process of bleb retraction 26; however , it is not clear how individual myosin II molecules are recruited to the blebs ., It is possible that c-Cbl interaction with myosin IIA leads to recruitment of the complex to the blebs ., Therefore , we infected control shRNA and c-Cbl shRNA cells with KSHV and tested the association of c-Cbl with myosin II in the blebs ., Punctate staining of p-Cbl and p-M

Introduction, Materials and Methods, Results, Discussion

KSHV is etiologically associated with Kaposis sarcoma ( KS ) , an angioproliferative endothelial cell malignancy ., Macropinocytosis is the predominant mode of in vitro entry of KSHV into its natural target cells , human dermal microvascular endothelial ( HMVEC-d ) cells ., Although macropinocytosis is known to be a major route of entry for many viruses , the molecule ( s ) involved in the recruitment and integration of signaling early during macropinosome formation is less well studied ., Here we demonstrate that tyrosine phosphorylation of the adaptor protein c-Cbl is required for KSHV induced membrane blebbing and macropinocytosis ., KSHV induced the tyrosine phosphorylation of c-Cbl as early as 1 min post-infection and was recruited to the sites of bleb formation ., Infection also led to an increase in the interaction of c-Cbl with PI3-K p85 in a time dependent manner ., c-Cbl shRNA decreased the formation of KSHV induced membrane blebs and macropinocytosis as well as virus entry ., Immunoprecipitation of c-Cbl followed by mass spectrometry identified the interaction of c-Cbl with a novel molecular partner , non-muscle myosin heavy chain IIA ( myosin IIA ) , in bleb associated macropinocytosis ., Phosphorylated c-Cbl colocalized with phospho-myosin light chain II in the interior of blebs of infected cells and this interaction was abolished by c-Cbl shRNA ., Studies with the myosin II inhibitor blebbistatin demonstrated that myosin IIA is a biologically significant component of the c-Cbl signaling pathway and c-Cbl plays a new role in the recruitment of myosin IIA to the blebs during KSHV infection ., Myosin II associates with actin in KSHV induced blebs and the absence of actin and myosin ubiquitination in c-Cbl ShRNA cells suggested that c-Cbl is also responsible for the ubiquitination of these proteins in the infected cells ., This is the first study demonstrating the role of c-Cbl in viral entry as well as macropinocytosis , and provides the evidence that a signaling complex containing c-Cbl and myosin IIA plays a crucial role in blebbing and macropinocytosis during viral infection and suggests that targeting c-Cbl could lead to a block in KSHV infection .

KSHV is etiologically associated with Kaposis sarcoma ( KS ) , the most common AIDS related neoplasm ., The first key step in KSHV infection is its initial contact with target cells and entry ., While it is known that KSHV uses macropinocytosis for its infectious entry into its natural target cells , HMVEC-d cells , we know little about the molecule ( s ) involved in this event ., Here , we show that the adaptor protein c-Cbl plays a major role in regulating bleb associated macropinocytosis of KSHV ., The results demonstrate that c-Cbl protein functions as an adaptor for the myosin II hexameric complex in macropinocytic events ., Knocking down c-Cbl by shRNA induces defects in myosin II dependent blebbing and KSHV entry , indicating that c-Cbl uses myosin II to coordinate signaling pathways , resulting in bleb formation and bleb retraction ., This work provides a clear understanding of the role of c-Cbl in the recruitment and integration of signaling molecules around the macropinosome during virus infection , and identifies potential targets to intervene in KSHV infection .

infectious diseases/viral infections, virology/host invasion and cell entry, virology/viruses and cancer, virology

journal.pntd.0001189

Identification of Peptide Mimotopes of Trypanosoma brucei gambiense Variant Surface Glycoproteins

Human African trypanosomiasis ( HAT ) , or sleeping sickness , is caused by the protozoan flagellar parasites Trypanosoma brucei ( T . b . ) gambiense and T . b . rhodesiense ., The disease is transmitted by tsetse flies ( Glossina spp . ) and therefore only occurs in sub-Saharan Africa ., The number of cases is currently estimated between 50 000 and 70 000 1 ., Control of T . b . gambiense HAT is largely based on accurate diagnosis and treatment of the human reservoir 2 ., Detection of parasites in blood , lymph node aspirate or cerebrospinal fluid is laborious and insensitive , and therefore only applied on suspected HAT patients ., In the absence of reliable antigen detection tests , the screening of the population at risk relies on the detection of antibodies against variant surface glycoproteins ( VSGs ) 2 ., These immunogenic VSGs form a dense monolayer of homodimers that completely covers the surface of bloodstream trypanosomes and determines the variable antigen type ( VAT ) of the individual trypanosome 3 ., The parasite genome contains hundreds of VSG genes and trypanosomes switch from the expression of one VSG gene to another ., This antigenic variation enables the parasite population to survive the hosts immune response ., Each VSG monomer contains 400–500 amino acids and consists of two domains , a variable N-terminal domain with little primary sequence homology and a relatively conserved C-terminal domain ., A glycosylphosphatidylinositol anchor links the C-terminal domain to the cell membrane ., All N-terminal domains fold in a similar three-dimensional structure , exposing only a limited subset of , probably discontinuous , epitopes 4–7 ., The current T . b . gambiense antibody detection tests are based on native VSGs from the VATs LiTat 1 . 3 , LiTat 1 . 5 and LiTat 1 . 6 of T . b . gambiense 8 ., These predominant VATs appear early during infection , and induce a strong and specific immune response in most patients 3 ., The card agglutination test for trypanosomiasis ( CATT ) 9 , most widely used for mass screening of populations at risk , consists of whole lyophilised trypanosomes of VAT LiTat 1 . 3 and has a sensitivity on whole blood of 87–98% and a specificity around 95% 2 ., Although VSG LiTat 1 . 3 is not expressed in all endemic HAT foci 10 , 11 , the low sensitivity of CATT in those foci can be overcome by combining different VATs in one test , as is the case in the LATEX/T . b . gambiense and the ELISA/T . b . gambiense where the combination of VSG LiTat 1 . 3 , 1 . 5 and 1 . 6 is used as antigen 12 , 13 ., The use of native VSGs as diagnostic antigens has several disadvantages ., Firstly , non-specific epitopes on the native antigens may cause cross-reactions and decrease test specificity ., Secondly , VSG production relies on culture of infective T . b . gambiense parasites in laboratory rodents and poses a risk of infection to the staff 14 ., These disadvantages might be avoided if native antigens are replaced by synthetic peptides ., The production of synthetic peptides is standardised , does not require laboratory animals and is without risk of infection 15 ., Peptide phage display is a selection technique based on DNA recombination , resulting in the expression of foreign peptide-variants on the outer surface of phage virions ., After an in vitro selection process based on binding affinity , called panning , the selected peptides are characterised by DNA sequencing ., Phage display is a powerful tool to identify mimotopes , small peptides that mimic linear , discontinuous and/or non-protein epitopes 16–18 ., Mimotopes with diagnostic potential have already been identified , e . g . for detection of specific antibodies for Lyme disease 19 , hepatitis C 15 , 20 , typhoid fever 21 , tuberculosis 22 and leishmaniasis 23 ., Some mimotopes have been patented to become incorporated in commercially available tests , e . g . for neurocysticercosis 24 ., In this study , we aimed to identify mimotopes for epitopes of T . b ., gambiense VSG LiTat 1 . 3 and LiTat 1 . 5 that may replace the native proteins in antibody detection tests for sleeping sickness ., Samples from HAT patients and endemic controls were collected within an observational study 13 ., All individuals gave their written informed consent before providing serum ., Permission for this study was obtained from the national ethical committee of DRC and from the ITM ethical committee , reference number 03 07 1 413 ., Monoclonal antibodies ( mAbs ) H12H3 ( IgG3 , anti-VSG LiTat 1 . 5 ) , H13F7 ( IgG3 , anti-VSG LiTat 1 . 3 ) and H18C11 ( IgG1 , anti-VSG LiTat 1 . 3 ) were generated by intraperitoneal infection of Balb/c mice with 106 T . b . gambiense LiTat 1 . 3 and 106 LiTat 1 . 5 cloned parasites ., After 2 weeks , splenocytes were isolated and fused with NS0 myeloma cells 25 ., Anti-VSG antibody producing hybridomas were identified by enzyme linked immunosorbent assay ( ELISA ) and further propagated ., The antibodies were purified from culture supernatant on protein A agarose ., The SBA Clonotyping™ system/HRP kit ( Southern Biotech ) was used for mAb isotyping ., Anti-VSG mAbs were coated onto anti-mouse IgG functionalised magnetic particles ( MP ) ( 1% w/v , 0 . 35 µm , Estapor/Merck ) at a concentration of 30 mg/g MP and stored in phosphate buffered saline ( PBS , 0 . 01 mol/L phosphate , 0 . 14 mol/L NaCl , pH 7 . 4 ) containing 0 . 1% ( w/v ) bovine serum albumin ( PBS-BSA ) ., The coated MP were washed eight times with PBS containing 0 . 25% w/v gelatine and 0 . 1% v/v Tween-20 ( PBSGT ) and resuspended 0 . 25% w/v gelatine in PBS ( PBSG ) ., Successful coating of the MP was confirmed by agglutination of VSG coated latex beads ( LATEX/T . b . gambiense ) 12 ., Anti-VSG mAb-free MPs were prepared by omitting the coating step ., Pannings were performed with the Ph . D . -12 ( 12-mer ) and the Ph . D . -C7C ( cyclic 7-mer ) phage display libraries ( New England Biolabs , NEB ) through two rounds consisting of 1 ) a positive selection with anti-VSG mAbs coated on MP , 2 ) a negative selection with anti-VSG mAb-free MP and 3 ) phage amplification ., After these two rounds a third positive selection was performed ., For positive selection , 10 µL of the phage display library ( for the 1st positive selection ) or 100 µL of amplified phage ( for the 2nd and 3rd positive selection ) were mixed overnight ( ON ) at 4°C with 1 mg mAb coated-MP in PBSG in a total volume of 1 mL ., MP were washed ten times with PBSGT and bound phages were eluted by antigen competition ( A ) followed by acid elution ( P ) ., For the antigen competition , the MP were incubated for 1 h with 700 µL PBS containing 0 . 23 mg of corresponding VSG ., After collection of the supernatant containing the eluted phages , MP were washed three times with PBSGT ., The remaining bound phages were eluted with 600 µL of 0 . 2 mol/L glycine-HCl containing 1 mg/mL BSA ( pH 2 . 2 ) and neutralised with 90 µL of Tris-HCl ( 1 mol/L , pH 9 . 1 ) ., The phages eluted from the positive selection were mixed overnight at 4°C with 1 mg of anti-VSG mAb-free MP in a total volume of 1 mL of PBSG ., The phages in the supernatant were amplified ., Phages were amplified in Escherichia ( E . ) coli ( strain ER2738 , NEB ) in lysogeny broth ( LB ) , supplemented with tetracycline ( 20 mg/mL ) 26 ., After 4 . 5 h shaking at 37°C , bacteria were pelleted by centrifugation ( 30 min , 1811 g ) ., The phages in the supernatant were precipitated overnight at 4°C with 25% w/v polyethylene glycol-6000 in 2 . 5 mol/L NaCl ( PEG-NaCl ) ., Phages were pelleted by centrifugation ( 45 min , 1811 g , 4°C ) and resuspended in 1 mL of PBS ., Residual bacteria were pelleted by centrifugation ( 5 min , 15700 g ) and phage precipitation with PEG-NaCl was repeated for 2 to 4 hours at 4°C ., After centrifugation ( 20 min , 15700 g , 4°C ) the phage pellet was resuspended in 200 µL of PBS with 0 . 02% w/v NaN3 ., After a third positive selection , phages selected through three antigen competition elutions ( AAA ) and three acid elutions ( PPP ) were titered as described below ., Phages were titered to obtain well separated , single plaques for analysis ., Phages were diluted 103 to 107 in PBS , mixed with an E . coli culture and plated on agar plates containing 1 mL/L IPTG/X-gal ( 1 . 25 g isopropyl β-D-thiogalactoside , 1 g 5-bromo-4-chloro-3-indolyl-β-D-galactoside , 25 mL dimethylformamide ) ., After the 3rd positive selection 94 blue clones were picked and each clone was inoculated in 200 µL of LB in a sterile culture plate ( BD Falcon™ Clear 96-well Microtest™ Plate ) ., This plate was shaken overnight at 30°C , where after the bacteria were pelleted by 5 min centrifugation at 1312, g . The supernatant was tested in a sandwich ELISA with the homologous mAb ., Based on these results , twenty phages per elution method were amplified , tested in a sandwich ELISA with all three mAbs and amplified for DNA extraction ., Purification of phage DNA was performed according to the NEB manual 26 ., Sequence determination was performed at the VIB Genetic Service Facility of the University of Antwerp with the −96 gIII sequencing primer , 5′-H0CCC TCA TAG TTA GCG TAA CG-3′ ( NEB ) ., The obtained sequence chromatograms were read with Chromas 2 . 33 ( Technelysium Pty Ltd ) ., Sequence alignment was performed manually and with RELIC software 27 ., We searched for discontinuous epitopes with the 3D-Epitope-Explorer ( 3DEX ) 28 and visualised the sequences on the protein model with PyMOL pdb viewer ( PyMOL Molecular Graphics System , Schrödinger , LLC ) ., ELISA plates ( Nunc MaxiSorp™ ) were coated with 5 µg/mL anti-VSG mAb in PBS ( 100 µL/well ) and incubated ON at 4°C ., Plates were saturated for 1 h at rT with 350 µl PBS-Blotto ( 0 . 01 mol/L phosphate , 0 . 2 mol/L NaCl , 1% w/v skimmed milk powder , 0 . 05% w/v NaN3 ) and washed three times with 0 . 05% v/v Tween-20 in PBS ( PBST ) ( ELx50 , Bio-Tek ELISA washer ) ., Wells were incubated for 4 h at rT with 100 µL of phage dilution in PBS-Blotto ( 1/3 for culture plate supernatant or 1/20 for PEG-NaCl purified phage ) ., After three washes , horse radish peroxidase ( HRP ) -labelled anti-M13 pVIII mAb ( GE Healthcare ) , diluted 1/2000 in PBST was added to the wells for 1 h at rT ( 100 µl/well ) ., After another five washes , wells were incubated for 1 h at rT with 100 µL/well of 2 . 2′-azino-bis- ( 3-ethylbenzthiazoline-6-sulfonic acid ) ( ABTS ) chromogen/substrate solution ( 50 mg tablet/100 mL of ABTS buffer , Roche ) ., The plate was shaken for 10 seconds and the optical density ( OD ) was read at 414 nm ( Labsystems Multiskan RC 351 ) ., Peptide selection for synthesis ( Eurogentec , Belgium ) was based on strong , reproducible and specific reaction of phage clones with their homologous anti-VSG mAb in ELISA , prediction of peptide hydrophilicity and common motive groups in the sequences ., The 12-mer linear peptides were synthesised at >70 or >85% purity ., The C-terminus was elongated with a GGGS-CONH2 tail to mimic the GGGS-peptide spacer between the random peptide sequence and the phage protein pIII and to block the negative charge of the carboxyl terminus ., The cyclic 7-mer sequences were synthesised at >90% purity ., The 7-mer peptides were flanked by two cysteines , and the C-terminus was elongated with GGGS-CONH2 ., The biotinylated peptides were synthesised at >85% purity , with an additional lysine-biotin added to the C-terminus ., All synthetic peptides were reconstituted in sterile deionised H2O to a concentration of 2 mg/mL ., The capacity of the synthetic peptides to mimic the natural VSG epitopes was assessed in an inhibition ELISA ., Peptide dilution series of 200 , 67 , 22 , 7 and 0 µg/ml or 67 , 22 , 7 , 2 and 0 µg/ml were prepared in PBS containing 1% BSA ( PBS-BSA ) with mAb H12H3 at 1 µg/mL , and in PBS-Blotto with mAb H13F7 at 11 µg/mL or mAb H18C11 at 0 . 04 µg/mL ., The same dilution series were prepared in PBS-BSA or PBS-Blotto but without mAb ., These dilutions were rotated ON at 4°C ., All samples were tested in duplicate ., For mAb H12H3 , ELISA plates were coated ON at 4°C with 100 µL/well of 2 µg/mL VSG LiTat 1 . 5 in PBS and antigen-free wells served as an antigen negative control ., The next day , ELISA plates were blocked at rT for 1 h with PBS-BSA and washed three times with PBST ., The wells were incubated for 1 h at rT with 100 µl of the dilutions ., Plates were washed three times and HRP-labelled goat Fc-specific anti-mouse IgG conjugate ( Jackson ) , diluted 1/500 in PBST was added for 1 h ( 100 µL/well , rT ) ., The colour reaction was performed as previously described for the sandwich ELISA ., For the mAbs H13F7 and H18C11 the same protocol was followed , using VSG LiTat 1 . 3 , PBS-Blotto 1% and a 1/1000 dilution of the anti-mouse conjugate ., The remaining activity ( % RA ) was calculated: first , for each peptide dilution , the average OD measured in the antigen-free control wells was subtracted from the average OD in the corresponding antigen containing wells , thus yielding ODa ., Second , the corrected ODc was obtained by subtracting the ODa obtained in the mAb-free wells from the ODa in the corresponding mAb containing wells ., Finally , the % remaining activity was calculated as 100×ODc/ODmax , where ODmax was the ODc of wells receiving the peptide-free mAb dilution ., Nunc MaxiSorp™ plates were coated with 100 µL/well of 10 µg/mL streptavidin ( NEB ) in carbonate buffer ( CB , 0 . 1 mol/L , pH 9 . 2 ) ., Plates were incubated ON at 4°C ., After saturation with PBS-BSA ( for mAb H12H3 ) or PBS-Blotto ( for mAb H13F7 ) and three washes with PBST , wells were filled with 100 µL of biotinylated peptide in PBS at a concentration of 5 ( only for peptide 24 , C57 and C59 ) , 1 , 0 . 6 , 0 . 3 and 0 µg/ml ., After 1 h at rT , wells were washed three times and incubated for 1 h with 100 µL mAb ( mAb H12H3 at 1 . 07 , 0 . 53 , 0 . 27 µg/mL in PBS-BSA or mAb H13F7 at 0 . 53 and 0 . 27 µg/mL in PBS-Blotto ) ., After three washes , HRP-labelled goat anti-mouse IgG ( Fcγ ) conjugate was diluted 1/1000 in PBST and added at 100 µL/well for 1, h . After five washes , plates were incubated for 1 h with 100 µL/well of ABTS ., The OD was read at 414 nm and corrected by subtracting the OD obtained in the peptide-free wells from the OD obtained in the corresponding peptide containing wells ., The capacity of the synthetic peptides to specifically bind antibodies in serum from HAT patients was assessed in an inhibition ELISA with nine sera from gambiense sleeping sickness patients and ten sera from endemic controls ., ELISA plates were coated ON at 4°C with 100 µL/well of 10 µg/mL streptavidin in CB ., All samples were tested in duplicate ., The following day the plates were saturated with PBS-BSA ( for mAb H12H3 ) or PBS-Blotto ( for mAb H13F7 ) ., The dilutions of the biotinylated peptides were made in PBS ( 0 . 01 mol/L phosphate , 0 . 14 mol/L NaCl , pH 6 ) , ranging from 20 µg/mL to 0 . 3 µg/mL , depending on the peptide ., After three washes with PBST , the peptide dilutions were added at 100 µL/well and left for 1 h at rT ., Plates were tapped dry , sealed and stored frozen at −80°C ., Just before use , plates were thawed and washed ( 3× PBST ) ., Wells were incubated with 100 µL of a 1/5 human serum dilution in PBS-BSA or PBS-Blotto , depending on the mAb ( 1 h at rT ) ., After three washes with PBST , mAb H12H3 ( in PBS-BSA ) or mAb H13F7 ( in PBS-Blotto ) was added at a concentration ranging from 1 to 0 . 25 µg/mL , depending on the peptide ( 100 µl/well , 1 h , rT ) ., Plates were washed three times and a 1/1000 dilution in PBST of HRP labelled goat anti-mouse IgG ( Fcγ ) conjugate was added at 100 µL/well ( 1 h , rT ) ., Plates were washed five times and incubated for 1 h at rT with 100 µL/well of ABTS ., The OD was read at 414 nm and corrected by subtracting the average OD obtained in the antigen-free wells from the average OD obtained in the corresponding antigen containing wells , thus yielding ODc ., The percent remaining activity was calculated as 100×ODc/ODmax ., P-values were tested with the Wilcoxon rank test ( positive versus negative ) and corrected for multiple comparisons with the Bonferroni method ., Based on the results in indirect ELISA with VSG LiTat 1 . 5 and LiTat 1 . 3 ( data not shown ) we selected nine HAT positive ( OD>1 . 5 ) and 10 endemic negative ( OD<0 . 2 ) human serum samples originating from a study on detection of specific antibodies in serum and saliva in the Democratic Republic of Congo ( DRC ) 13 ., Panning of the Ph . D . -12 library yielded phage clones for which median ODs in the sandwich ELISA with the homologous mAb were 1 . 352 ( interquartile range , IQR , 1 . 119–1 . 608 ) for the AAA elution and 2 . 759 ( IQR 2 . 384–2 . 943 ) for the PPP elution ., Twenty phage clones eluted with AAA and OD>1 and twenty phage clones eluted with PPP and OD>2 , were selected for amplification , cross reactivity testing and sequencing ., All forty amplified phage clones reacted specifically with mAb H12H3 ( median OD 1 . 055 with H12H3 versus 0 . 100 with the heterologous mAb H13F7 and H18C11 ) , therefore excluding that the selected phage clones bound to the conserved Fc part of the mAbs ., Phages obtained with the AAA elution expressed six different amino acid ( AA ) sequences ( fig . 1 ) ., One sequence could not be read ., Phages obtained with the PPP elution expressed nine different sequences ., Panning of the Ph . D . -C7C library yielded phage clones for which median ODs in the sandwich ELISA with the homologous mAb were 0 . 713 ( IQR 0 . 485–2 . 338 ) for the AAA elution and 3 . 218 ( IQR 3 . 138–3 . 324 ) for the PPP elution ., Twenty phage clones eluted with AAA and OD>2 and twenty phage clones eluted with PPP and OD>3 , were selected for amplification , cross reactivity testing and sequencing ., All amplified phage clones reacted specifically with mAb H12H3 ( median OD 3 . 106 with H12H3 versus 0 . 138 with the heterologous mAbs H13F7 and H18C11 ) ., Five phage clones corresponding to three sequences did not amplify well ( OD with mAb H12H3<0 . 4 ) and were not withheld for further experiments ., With each elution method , phages expressing twelve different amino acid sequences were obtained ( fig . 1 ) ., Two of these sequences were found in both the AAA and the PPP elution , bringing the total of different C7C-sequences to twenty-two ., Amongst the thirty-seven different sequences obtained through panning with mAb H12H3 , four groups of common motives could be distinguished ( fig . 1 ) ., Amino acids with similar structure and characteristics were considered homologous , such as arginine ( R ) and lysine ( K ) ; serine ( S ) and threonine ( T ) ; glutamine ( Q ) and asparagine ( N ) ; alanine ( A ) , valine ( V ) , leucine ( L ) and isoleucine ( I ) ; phenylalanine ( F ) , tyrosine ( Y ) and tryptophan ( W ) ; and aspartic acid ( D ) and glutamic acid ( E ) ., Group 1 with common motive SAP ( W/Y ) ( V/A , S or N ) ( L/A or Y/F/W ) ( R/K ) DH ( L/A or Y/F ) ( P or S/T ) L/AxG contained all the Ph . D . -12 sequences , and part of the Ph . D . -C7C sequences ., Group 2 , 3 and 4 consisted of Ph . D . -C7C sequences only and had as a common motive AxxxT ( S/T or A ) ( P or L ) ( N/Q ) QWL , AxPVYExHWxxxG , and AxQxPHxxxxG respectively ., The sequence CTDFEGMLC did not have more than two AA in common with one of the other sequences and is displayed separately ., Homology between peptides and the protein sequence of VSG LiTat 1 . 5 GenBank HQ662603 was found within AA 268 to 281 of the protein sequence ( maximum 42 . 86% or 6/14 identical AA , fig . 1 ) , in the variable N-terminal domain of VSG LiTat 1 . 5 ., Out of the thirty-seven obtained sequences , ten peptides were synthesised ., All ten synthetic peptides strongly inhibited the binding of mAb H12H3 to native VSG LiTat 1 . 5 in a dose dependent manner ( <25% remaining activity at a peptide concentration of 67 µg/ml ) ( fig . 1 ) and were resynthesised with a C-terminal lysine-biotin ., Peptides selected with mAb H13F7 and H18C11 ( see below ) did not inhibit binding of H12H3 to VSG LiTat 1 . 5 , ( data not shown ) ., All biotinylated peptides were recognised by mAb H12H3 in an indirect ELISA ., A peptide concentration of >10 µg/mL was necessary to obtain an OD>0 . 5 with peptides C57 and C59 while 5 µg/mL for peptide 24 and concentrations ranging from 0 . 3 to 0 . 6 µg/mL for the other seven peptides , were sufficient to obtain an OD>1 ( data not shown ) ., Panning of the Ph . D . -12 library yielded phage clones for which median ODs in the sandwich ELISA with the homologous mAb were 3 . 219 ( IQR 3 . 000–3 . 350 ) for the AAA elution and 0 . 148 ( IQR 0 . 123–2 . 999 ) for the PPP elution ., Per elution method , twenty phage clones with OD>2 were selected ., All forty amplified clones reacted specifically with mAb H13F7 ( median OD 3 . 022 with H13F7 versus 0 . 113 with the heterologous mAbs H12H3 and H18C11 ) ; this excludes that the selected phage clones bound to the conserved Fc part of the mAbs ., Phages obtained with the AAA elution , expressed seven different amino acid sequences ., Phages obtained with the PPP elution , expressed three different sequences ., One sequence was found in both the AAA and the PPP elution , thus bringing the total of different sequences to nine ( fig . 2 ) ., Panning of the Ph . D . -C7C library yielded phage clones for which median ODs in the sandwich ELISA with the homologous mAb were 1 . 481 ( IQR 0 . 739–1 . 869 ) with the AAA eluted phage clones and 0 . 107 ( IQR 0 . 087–0 . 122 ) with the PPP eluted phage clones ., None of the PPP eluted clones were withheld ., Twenty phage clones of the AAA elution with OD>1 were amplified ., All reacted specifically with mAb H13F7 ( median OD 1 . 942 with H13F7 versus 0 . 106 with the heterologous mAbs H12H3 and H18C11 ) ., Seven different amino acid sequences were expressed ( fig . 2 ) ., Amongst the sixteen different sequences obtained with mAb H13F7 , three groups of common motives could be distinguished ( fig . 2 ) ., Group 1 with common motive PPxWINPFPxF contained only 12-mer sequences ., Group 2 contained some of the 12-mer and all of the 7-mer sequences and had as common motive PW ( W or L ) PLQ ( W/Y ) ( I/V/L ) F or , with “WPL” in reverse order , Ax ( I/V ) L ( P or S ) WLH ( I/V ) ., Peptide 59 and peptide C63 share a common motive , but in reverse order ( F/W ) LPL ., Group 3 consisted of two sequences with common motive SPxMLH ., Alignment of these sequences with the protein sequence of VSG LiTat 1 . 3 GenBank AJ304413 , only gave results for group 2 and one sequence of group 3 ., These had maximum 28 . 57% identical AA ( 4/14 ) within respectively AA stretch 196 to 210 and AA stretch 338 to 351 of VSG LiTat 1 . 3 ( fig . 2 ) ., The motive W ( AA 291 ) P ( AA 290 ) L ( AA 292 ) L ( AA 234 ) T ( AA 230 ) of peptide 64 could be mapped onto the three-dimensional VSG LiTat 1 . 3 protein structure , in the N-terminal domain ( results not shown ) ., Out of the sixteen sequences selected with mAb H13F7 , twelve peptides were synthesised ., Six synthetic peptides strongly inhibited the binding of the mAb to the native VSG LiTat 1 . 3 ( <50% remaining activity at a peptide concentration of 67 µg/ml ) , and one peptide was a weaker inhibitor ( 76% remaining activity ) ( fig . 2 ) ., These seven peptides were resynthesised with a C-terminal lysine-biotin and their reactivity with mAb H13F7 was assessed by indirect ELISA ., All biotinylated peptides had an OD>1 with mAb H13F7 at a concentration of 0 . 6 to 0 . 3 µg/mL peptide ( data not shown ) ., Panning of the Ph . D . -12 library yielded phage clones for which median ODs in the sandwich ELISA with the homologous mAb were 0 . 124 ( IQR 0 . 109–0 . 185 ) for the AAA elution and 0 . 130 ( IQR 0 . 108–0 . 148 ) for the PPP elution ., None of the clones from the Ph . D-C7C-library gave a sufficiently high OD to be amplified ., Twenty AAA eluted clones of the Ph . D . -12-library with OD>0 . 5 were amplified ., All twenty clones reacted specifically with mAb H18C11 ( median OD 3 . 271 with H18C11 versus 0 . 106 with the heterologous mAbs H12H3 and H13F7 ) ., Only one amino acid sequence was expressed: SHSTPYYWKGYI ., We could not identify any homology between this peptide and the protein sequence of VSG LiTat 1 . 3 ., The corresponding synthetic peptide did not react with mAb H18C11 in indirect ELISA ( data not shown ) and did not inhibit the binding of mAb H18C11 to native VSG LiTat 1 . 3 and was therefore not withheld for further experiments ., The diagnostic potential of the biotinylated peptides was assessed in an inhibition ELISA with human sera from nine gambiense HAT patients and ten negative controls ., Compared to the HAT negative sera , the HAT positive sera significantly inhibited binding of mAb H12H3 to peptide 23 , C59 and C60 ( p<0 . 05 ) and 21 , 22 , and 28 ( p<0 . 01 ) ( fig . 3 ) ., Five of these peptides belong to common motive group 1 , peptide C60 belongs to group 4 ( fig . 1 ) ., The HAT positive sera also significantly inhibited binding of mAb H13F7 to peptide 25 ( p<0 . 05 ) and peptides 60 and 61 ( p<0 . 01 ) ( fig . 3 ) ., All three peptides belong to common motive group 1 ( fig . 2 ) ., By means of phage display technology we successfully identified peptides that mimic epitopes on the native trypanosomal variant surface glycoproteins LiTat 1 . 5 and LiTat 1 . 3 of T . b . gambiense ., These mimotopic peptides were recognised by the monoclonal anti-VSG antibodies H12H3 and H13F7 that were used for panning , and might have potential for the diagnosis of human African trypanosomiasis ., Our results indicate that a linear region in the protein sequence of VSG LiTat 1 . 5 was identified ., This region is localised in the N-terminal domain of the VSG near the surface of the trypanosome and is therefore a candidate for further testing as a synthetic , linear peptide ., Antibodies specific to linear , continuous epitopes on protein antigens typically contact three to four critical amino acids over a six residue segment 16 ., The peptide sequences selected with mAb H12H3 had up to six amino acids ( % identity 42 . 86 ) in common with the variable N-terminal domain ( AA 268 to 281 ) of VSG LiTat 1 . 5 ., Glycine from the GGGS-spacer , inserted between the peptide sequence and the pIII phage protein , was part of this common motive ., It is possible to define the exact residues in the peptide sequences that are essential for binding with the antibody by alanine scanning mutagenesis and recreate the epitope of each monoclonal ., A recent example has been the identification of a linear epitope on the VP1 protein of foot-and mouth disease virus by Yang et al . 29 by screening a 12-mer phage display library with a mAb ., Contrary to VSG LiTat 1 . 5 , we suspect the epitope of VSG LiTat 1 . 3 to be discontinuous ., Alignment of the peptide sequences with the protein sequence of VSG LiTat 1 . 3 located the common motives in different parts of the protein sequence with a maximum % identity of only 28 . 57 ., Also , when the VSG was ( partly ) denatured , the OD in ELISA with this mAb dropped ., Many protein epitopes are discontinuous and comprise critical binding residues that are distant in the primary sequence but close in the folded native tertiary protein structure ., Indeed , WPLLT , the motive of peptide 64 was mapped onto the three-dimensional VSG LiTat 1 . 3 protein structure , near the surface of the trypanosome ., As “WPL” or “LPW” is part of the common motive in the peptide sequences of group 2 ( fig . 2 ) it is possible that the discontinuous epitope of mAb H13F7 is localised in this region ., All 12-mer peptides that strongly inhibited the binding of the mAbs to the VSG contain one or more proline residues ., Proline limits the flexibility of the peptide and may therefore favour the forming of the mAb-peptide complex 30 ., The 7-mer peptides are already constrained by two flanking cysteines , which may account for the fact that the sequence of four of the best cyclic inhibitors contains no proline ., Cortese 31 reports that many mAbs fail to select specific peptides ., Due to the limitation of library complexity it is often impossible to isolate peptides of high affinity and there is no general rule applicable as to what type of library suits a certain application 18 ., In our study mAb H18C11 selected specific peptides , but only with the 12-mer library ., Additionally , all phage clones selected with this mAb expressed the same peptide sequence ., It is possible that one phage clone overgrew other , higher-affinity phage clones , during amplification ., The peptide selected with mAb H18C11 and some of the synthetic peptides selected with mAb H12H3 and mAb H13F7 failed to react with the corresponding mAb in indirect ELISA ., It has been described before that phage-born peptides can lose their ability to bind the target molecule when synthesised chemically 17 ., Furthermore the conformation of peptides in binding assays may differ from the presentation on the phage ., Although direct coating was only successful for some peptides ( data not shown ) , we successfully demonstrated the capacity of several of these peptides in solution to inhibit the binding of their mAb to the corresponding VSG ., Based on these results we selected some peptides for resynthesis and biotinylation ., The biotinylated peptides were bound to streptavidin , which was coated onto the ELISA plate , this improved the peptide presentation to such an extent that all of the biotinylated peptides were able to bind their corresponding mAb ., Although the aim of our study was to identify mimotopes for antibody detection in human serum , we opted to perform the panning with mouse monoclonal antibodies ., Indeed , sera from sleeping sickness patients contain an important fraction of trypanosome unrelated antibodies as a consequence of polyclonal B cell stimulation 32 , 33 ., Therefore , the risk to select mimotopes unrelated to sleeping sickness by applying human sera for the panning is considerable , unless only the trypanosome specific antibody fraction of these sera is used ., Moreover , it was demonstrated that mAbs identified peptide mimotopes similar to those selected with pooled sera of typhus patients 21 ., Also , mAb H13F7 was able to cause lysis of trypanosomes of VAT LiTat 1 . 3 in the immuno-trypanolysis test , which demonstrates that this mAb recognises VSG epitopes , exposed on living bloodstream trypanosomes , similar to those recognised by human sleeping sickness sera 3 ., Finally , we chose to perform the screening with mAbs that bind to different epitopes on the VSGs to increase the chance that the selected mimotopes would bind different antibodies in the polyclonal patient sera ., Human sleeping sickness sera inhibited the binding of anti-LiTat 1 . 5 mAb H12H3 to peptides 21 , 22 , 23 , 28 , C59 and C60 and the binding of anti-LiTat 1 . 3 mAb H13F7 to peptides 25 , 60 and 61 , auguring for their value as diagnostic antigens ., Not all peptides were equally well recognised by human sera ., It is possible that some peptides react weakly with positive sera in spite of their specificity for immunodominant regions , which can be explained if these peptides mimic only part of the structure of the corresponding region on the antigen 20 ., This may be the reason why the mimotopes corresponding to the linear region in the protein sequence of VSG LiTat 1 . 5 that interacts with the mAb seem to be more easily recognised by human antibodies than the mimotopes of the discontinuous LiTat 1 . 3 epitope , where a correct conformational presentation is crucial ., The mouse and human immune system may as well react with different vigour to certain epitopes or recognise different principal epitopes ., The fraction of antibodies in HAT sera that bind the same epitope as the mAbs may therefore be relatively small ., By using an inhibition ELISA , low serum dilutions could be applied , maximising the reaction with the pep

Introduction, Materials and Methods, Results, Discussion

The current antibody detection tests for the diagnosis of gambiense human African trypanosomiasis ( HAT ) are based on native variant surface glycoproteins ( VSGs ) of Trypanosoma brucei ( T . b . ) gambiense ., These native VSGs are difficult to produce , and contain non-specific epitopes that may cause cross-reactions ., We aimed to identify mimotopic peptides for epitopes of T . b . gambiense VSGs that , when produced synthetically , can replace the native proteins in antibody detection tests ., PhD ., -12 and PhD ., -C7C phage display peptide libraries were screened with mouse monoclonal antibodies against the predominant VSGs LiTat 1 . 3 and LiTat 1 . 5 of T . b . gambiense ., Thirty seven different peptide sequences corresponding to a linear LiTat 1 . 5 VSG epitope and 17 sequences corresponding to a discontinuous LiTat 1 . 3 VSG epitope were identified ., Seventeen of 22 synthetic peptides inhibited the binding of their homologous monoclonal to VSG LiTat 1 . 5 or LiTat 1 . 3 ., Binding of these monoclonal antibodies to respectively six and three synthetic mimotopic peptides of LiTat 1 . 5 and LiTat 1 . 3 was significantly inhibited by HAT sera ( p<0 . 05 ) ., We successfully identified peptides that mimic epitopes on the native trypanosomal VSGs LiTat 1 . 5 and LiTat 1 . 3 ., These mimotopes might have potential for the diagnosis of human African trypanosomiasis but require further evaluation and testing with a large panel of HAT positive and negative sera .

The control of human African trypanosomiasis or sleeping sickness , a deadly disease in sub-Saharan Africa , mainly depends on a correct diagnosis and treatment ., The aim of our study was to identify mimotopic peptides ( mimotopes ) that may replace the native proteins in antibody detection tests for sleeping sickness and hereby improve the diagnostic sensitivity and specificity ., We selected peptide expressing phages from the PhD ., -12 and PhD ., -C7C phage display libraries with mouse monoclonal antibodies specific to variant surface glycoprotein ( VSG ) LiTat 1 . 3 or LiTat 1 . 5 of Trypanosoma brucei gambiense ., The peptide coding genes of the selected phages were sequenced and the corresponding peptides were synthesised ., Several of the synthetic peptides were confirmed as mimotopes for VSG LiTat 1 . 3 or LiTat 1 . 5 since they were able to inhibit the binding of their homologous monoclonal to the corresponding VSG ., These peptides were biotinylated and their diagnostic potential was assessed with human sera ., We successfully demonstrated that human sleeping sickness sera recognise some of the mimotopes of VSG LiTat 1 . 3 and LiTat 1 . 5 , indicating the diagnostic potential of such peptides .

humoral immunity, sequencing, medicine, clinical laboratory sciences, protein interactions, immunology, microbiology, parastic protozoans, glycoproteins, sequence analysis, synthetic peptide, biology, proteomics, biochemistry, trypanosoma, diagnostic medicine, clinical immunology, immunity, protozoology, glycobiology, immunoglobulins

journal.pcbi.1005395

Dysfunctions of the basal ganglia-cerebellar-thalamo-cortical system produce motor tics in Tourette syndrome

Tourette syndrome ( TS ) is a neuropsychiatric disorder characterized by the presence of sudden and repetitive involuntary movements or vocalizations , generally termed as “tics” , having differing degrees of intensity and frequency , and unpredictable duration 1 , 2 ., Tics can be simple , for example involving eye blinking , facial grimacing , shoulder shrugging , sniffing , or complex , involving more elaborated manifestations like touching objects , clapping , obscene gestures , or repetition of words 3 , 4 ., The typical age of onset of TS is around five to seven years and the course of the disease can be quite variable ., In addition to tics , children with TS can show a variety of comorbid psychopathologies , including learning difficulties , sleep abnormalities , anxiety , obsessive-compulsive disorder ( OCD ) , and attention deficit hyperactivity disorder ( ADHD ) 5 , 6 ( see S1 Table in the Supporting Information for all the main abbreviations used in the article ) ., Usually , most TS symptoms decline during adolescence or early adulthood 7 ., Motor tics are a cardinal symptom of TS shared with several neurological impairments including dystonia 8 , Huntington’s disease 9 , 10 and OCD 11 , 12 ., Traditionally , tics in TS are associated with basal ganglia abnormalities and in particular with a dysfunction of the striatal GABAergic networks leading to an excess of striatal dopamine 13–16 ., This excess might cause an abnormal functioning of the basal ganglia-thalamo-cortical circuit leading to the production of tics 17 ., To understand how this circuit may operate in TS , we first briefly describe how it typically works in healthy subjects ( see section “The basal ganglia and their loops with the thalamo-cortical system: anatomy and physiology” for more details ) ., In general , the basal ganglia promote movement generation of some specific motor patterns within primary motor cortex via a double-inhibition mechanism while maintaining tonic inhibitory control over other patterns 18–21 ., In non-pathological conditions , the inhibition of specific GABAergic output nuclei of the basal ganglia leads to release the activity within the target thalamus areas forming loops with primary motor cortex , thus allowing the focused disinhibition of specific motor patterns ., The basal-ganglia double-inhibition mechanism also targets sub-cortical areas , although in this case without the mediation of the thalamus , for example the superior colliculus for eye movements 22 , 23 ., An alteration in striatal dopamine release as in TS may induce the production of tics as a consequence of a focal excitatory abnormality in the striatum that causes an undesired disinhibition of thalamo-cortical circuits 15 , 17 whose effect is the production of tics ., The basal ganglia are strongly linked , both anatomically and functionally , with several cortical regions and with the cerebellum ., The basal ganglia and cerebellum receive input from , and send output to , cortex through multisynaptic anatomically partially segregated loops performing distinct functional operations within the motor and cognitive realms 24–27 ., Studies in rats 28 and monkeys 29 have demonstrated that the cerebellum has a strong disynaptic projection to the striatum mediated by the intralaminar nuclei of the thalamus ., Complementary to this , recent investigations on monkeys have shown that the subthalamic nucleus , an important component of the basal ganglia , has a disynaptic projection to the cerebellar cortex by way of the pontine nuclei 30 ., Similar data have been found in humans 31 ., These data have stimulated new research to investigate the role of the cerebellum and basal ganglia in functions typically associated with cortex ( e . g . , action understanding , 32–35 ) , and the involvement of cortical and cerebellar regions in impairments typically associated with basal ganglia such as Parkinson’s disease 36–46 and TS 47–49 ., This system-level perspective 50 , 51 , according to which the basal ganglia work in concert with cortex and cerebellum to produce motor and cognitive behaviours of various complexity 26 , 35 , 52–55 , renders the whole picture of TS pathophysiology more complex 56 ., In particular , the specific contribution of cerebellar and cortical areas to basal ganglia-mediated tic expression remains unknown ., The cerebellar activation found in several studies on tics may reflect an increase of afferent sensory input driven by overt tic movements or , rather , may be due to the transmission of descending signals originating from primary motor cortex 57 ., Another possibility is that cerebellar neurons fire before tic movements and their discharge takes place no later than that of primary motor cortex neurons 49 ., Recently , McCairn and colleagues 49 have explicitly adopted a system-level approach to investigate the role of basal ganglia , cortical , and cerebellar areas in TS ., The authors generated a pharmacologic motor tic/TS model with two monkeys by microinjecting the GABA antagonist bicuculline into the sensorimotor striatum ( putamen ) 57 , 58 ., In this way , the increased striatal inhibition caused abnormalities in the dopamine release 3 , 59 , 60 that , in turn , led to motor tics 13–16 ( see section “Simulation settings” for more details ) ., Neural activity was recorded from several areas of the basal ganglia , cerebellum , and primary motor cortex simultaneously to investigate their relationship ., The results confirmed that aberrant activity leading to motor tics was initiated in the basal ganglia ., However , they also showed how the occurrence of tics was closely associated with enhanced activity involving both the motor cortex and the cerebellum , implying that these may act in concert to produce overt tic movements ., The time latencies of pathological activity in the cerebellum and primary motor cortex substantially overlapped and followed that of basal ganglia ., This suggests that aberrant signals may travel along divergent pathways from the basal ganglia to the cortex and cerebellum ., In this respect , the authors suggest that the basal ganglia might , presumably , influence cerebellar activity via the subthalamic-pons-cerebellar disynaptic link 30 , with a latency that is sufficiently short to allow cerebellum to affect abnormal movements ., However , the authors did not support this claim empirically ., Building on the results obtained in 49 , in this paper we propose a computational model reproducing key anatomical and functional features of the system formed by the basal ganglia , thalamus , primary motor cortex , and cerebellum to investigate within a system-level perspective how motor tics are generated in TS ., The model yields several results and predictions ., First , it reproduces the main results obtained in 49 about the differences in basal ganglia/primary motor cortex/cerebellum neural activity recorded during tic/no-tic events ., Second , and remarkably , the model shows that in order to reproduce and explain these data it is important to study the interplay between striatal dopamine signals and cortical activity , and the role played by the recently discovered subthalamic-pons-cerebellar pathway 30 working in synergy with the cerebello-thalamo-cortical circuit ., In particular , the model predicts that the interplay between dopaminergic signals and cortical activity may underlie the emergence of tic events , and that the anatomical connection linking subthalamic nucleus and cerebellum may support the involvement of the cerebellum in tic production ., In this way , the model supports the claim of 49 about a possible involvement of the subthalamic-pons-cerebellar circuit in tic generation , while specifying what functions it might accomplish ., These predictions could form the basis for future experiments ., Third , the model predicts that tic production could be reduced by externally stimulating or inhibiting the primary motor cortex ., These predictions could be important for identifying new target areas , aside the traditional ones 6 , 61 , 62 , to design innovative system-level therapeutic actions ., Finally , the model investigates the role of the recently discovered disynaptic bi-directional connections linking the basal ganglia with the cerebellum 29 , 30 ., To the best of our knowledge , there are no computational models investigating the role of these connections ., Previous computational and conceptual models have , indeed , mainly studied the indirect interactions between basal ganglia and cerebellum mediated by cortical areas 63–69 ., In view of recent empirical studies , attention to non-cortical-mediated basal ganglia-cerebellum interaction could radically change our perspective about how these subcortical areas interact with each other and with the cortex to regulate motor and non-motor behaviours 31 , 35 , 43 , 55 ., The computational model proposed here starts to address this issue by developing a simplified computational implementation of such links and by suggesting the possible involvement of the subthalamic-pons-cerebellar circuit in motor tic production ., Fig 1 summarizes the brain areas mainly involved in tic production ., The rest of the paper is organized as follows ., Section “Methods” describes the computational features of the model and the biological support of its assumptions ., Section “Results” illustrates the results of the target empirical experiments with monkeys performed in 49 and how the model reproduces and explains them ., It also presents the predictions of the model ., Section “Discussion” discusses the system-level mechanisms through which the model explains the motor tic production and presents some limitations of the model while also suggesting possible future work to overcome them ., The system-level architecture of the model is formed by four main components ( see Fig 2 ) : the basal ganglia component ( BG ) reproduces the key anatomical and functional features of the basal ganglia building on the computational models proposed in 21 , 70–72; the cerebellum component ( Cer ) captures some critical anatomical and functional aspects of the cerebellum pivoting on the models proposed in 68 , 73 , 74; the motor thalamus and the primary motor cortex components ( respectively Th and M1 ) , which do not focus on anatomical features , only reproduce functional aspects related to the activity of distinct neural populations ., Indeed , as it was non-trivial to reproduce the dynamics of the complex system formed by the basal ganglia-thalamo-cortical loops , the loops linking the cerebellum with the cortex through thalamus , and the circuits linking the basal ganglia with the cerebellum , we used simplified models of the primary motor cortex and thalamus that allowed an easier study of the structures considered important for the generation of tics ., This follows a strategy previously proposed for building system-level models more amenable to analysis 75 ( cf . also 76 , 77 ) ., At the same time , due to the key role of the basal ganglia in triggering motor tics in TS 49 we considered a more sophisticated model of these nuclei with respect to the other components of the model ., The possible effects of introducing finer grained anatomical and physiological details in the model are discussed in section “Conclusions and future work” ., With the exception of Cer components , each of the other model components is formed by three neural units representing three distinct neural populations encoding different information contents ., From a behavioural point of view , it would have been sufficient to include just one neural unit for each component to address the target experiment of McCairn and colleagues 49 ., Indeed , this experiment involved monkeys not solving any specific task but rather producing motor tics as spontaneous input-free behaviors under neural noise ( as detailed below , in the model such noise is intended to capture the spurious effects on neural activation due to the signals supplied by other cortices as well as the effect of intrinsic neural noise 77–80 ) ., However , it was important to include a larger number of neural units to reproduce in a realistic way the circuitry implementing the competitive dynamics typical of some components of the model , in particular of the BG 21 , 81 , relevant to the production of tics ( see sections “The basal ganglia and their loops with the thalamo-cortical system: anatomy and physiology” and “The model predicts that the interplay between dopaminergic signal and cortical activity triggers the tic event” ) ., The neural units within each component of the model are represented by leaky integrator units 82 , 83 ., The activation of a single leaky unit represents the average firing rate of a population of real neurons ., The neural population approach based on leaky integrator units is suitable for representing system-level features that are not immediately apparent at the level of individual neurons but manifest at higher levels 77 ., This approach facilitates the comparison between the data on neural activation recorded in the model and the data obtained in the target experiment proposed in 49 ., In addition , it allows a dimensionality reduction that increases the computational efficiency of simulations 84 , and this is important for running sensitivity analyses of large models such as the one performed here ., The chosen granularity of the model was also suitable for this work since it did not aim to reproduce detailed neural spatio-temporal patterns supporting the selection and performance of specific movements ( cf . section “Simulation settings” ) ., The model has been implemented , as described here , based on a technique that was proposed in 85 , 86 ( see also 87 ) ., This technique , suitable to illustrate neural system-level models formed by homogeneous neurons , aims to standardise all equations of the model so as to simplify its explanation , understanding , implementation , analysis , and reproducibility ., The model is in particular fully described by the few equations presented in this section , the values of the equation parameters reported in the S2 Table ( see Supporting Information ) , and the diagram of Fig 2 showing the architecture and connectivity of the model ., Each leaky integrator unit of the model components has an activation a and an activation potential ( hereafter “potential” ) u at time t having the following dynamics 82 , 83:, τ u ˙ = - u + I ( 1 ) a = f ( u ) ( 2 ), where τ is the unit decay coefficient; I is the input to the unit that , depending on the component to which the unit belongs , could take into account the effects of the different pre-synaptic connections received from other components , the effects of noise , and the effects of dopamine ., In particular , the term I of the post-synaptic unit j of the component post is computed as follows ( the effects of dopamine are discussed below ) :, I p o s t j = r p o s t j + ∑ p r e ∑ i w p r e i → p o s t j · a p r e i + n ( 3 ), where rpostj is the resting potential of the post-synaptic unit j of the component post; wprei → postj is the weight of the connection from the pre-synaptic unit i of the component pre to the post-synaptic unit j of the component post; aprei is the activity of the pre-synaptic unit i of the component pre computed according to Eq 2 , and n is a noise value independently sampled from a Gaussian distribution for each unit ., The pre-synaptic and post-synaptic units are those respectively sending and receiving signals as indicated in Fig 2 . The function f ( . ) = tanh ( . ) − thr+ is the activation function of neural units , where tanh ( . ) is the hyperbolic tangent function , whose values were remapped to the range −400 , 400 , thr is a parameter used to reproduce the effects of the threshold potential of real neurons 88 , and . + is a function returning the value of the function argument if this is positive , and zero otherwise ., The differential equations related to the u of all units are numerically integrated using the Euler method ., Before presenting the computational details of the model components , this section highlights some features of the anatomy and physiology of the basal ganglia , and their loops with the thalamo-cortical system , as they are particularly important for tic production ., The description uses the same abbreviations adopted for the model components shown in Fig 2 . In the model , the BG component includes five regions , each formed by a layer of three leaky integrator units ., The two main inputs of the BG component are Str and STN ., Str is formed by two subregions , StrD1 and StrD2 , with units expressing D1R and D2R dopamine receptors ., STN works in a loop with GPe and receives most of its afferent projections from M1 ., Similarly , StrD1 and StrD2 receive afferent projections from M1 and Th ., StrD1 , StrD2 , STN and GPe send efferent projections to the GPi or SNr , which are the GABAergic output nuclei of the BG ( hereafter , GPi and SNr , represented as one component in the model , will be indicated as GPi/SNr ) ., The excitatory and inhibitory connections between the regions of the BG component are feedforward links between one unit and the topologically corresponding unit in the following layer ( thin lines in Fig 2 ) ., This connectivity reproduces in an abstract fashion the structure of the BG channels ( one-to-one connections ) ., The units of STN are connected with all GPi and GPe units ( all-to-all connections ) ., This simulates the diffused action of the STN over its target regions 24 , 70 ., BG project to the Th through inhibitory links ( GPi/SNr-Th ) 21 and to Cer through excitatory connections ( STN-Cer ) 30 , 35 ., For the striatal sub-component StrD1 , the I term is calculated by multiplying the right side of Eq 3 for the dopaminergic term aDAD1 used to account for the dopaminergic modulation on the activity of StrD1 and computed as follows:, a D A D 1 = b S t r D 1 + d S t r D 1 · a D A ( 4 ), where bStrD1 is a baseline StrD1 potential modulation not due to DA , dStrD1 is the StrD1 DA factor amplitude , and aDA is the activity of a leaky integrator unit ( Eq 2 ) used to simulate the dopamine efflux ., The dopamine efflux was simulated through an activation potential uDA of the DA leaky unit that rapidly reaches a maximum level DAMAX = 0 . 5 around 1 sec from the beginning of each trial , and then decays toward DAMIN = 0 . 01 ., Similarly , for the striatal sub-component StrD2 the I term is calculated by multiplying the right side of Eq 3 by the dopaminergic term aDAD2 used to account for the dopaminergic modulation on the activity of StrD2 and computed as follows:, a D A D 2 = a D A D 1 b S t r D 2 + d S t r D 2 · a D A ( 5 ), where bStrD2 is a baseline StrD2 potential modulation not due to DA and dStrD2 is the StrD2 DA factor amplitude ., While the contribution of the dopaminergic efflux on the activity of StrD1 units was implemented as a multiplicative excitatory effect ( Eq 4 ) , the modulation of dopaminergic efflux on the activity of StrD2 units was implemented as a multiplicative inhibitory effect ( Eq 5 ) ., It has been shown that these two different types of dopaminergic modulations reflected what happens in the real BG ( cf . 103 ) ., Hence the term aDAD1 in the Eq 5 takes into account the recent data showing a possible combined effect of D1 and D2 receptors 89 ., For the other sub-components of BG ( STN , GPe and GPi ) the I term was computed by simply using the Eq 3 . The Th component is formed by two regions: ThBC , representing the thalamic parts where both BG and Cer project; ThC , representing the thalamic areas where only Cer projects ., Each region includes three leaky integrator units ., This organization in two subregions is based on anatomical data showing the presence of both partially segregated and overlapping projections from the BG and Cer output regions to Th 104 , 105 ., ThBC receives inhibitory signals from the BG component ( GPi/SNr region ) and excitatory signals from the Cer component 106 , 107 ., By contrast , ThC only receives excitatory signals from the Cer component 26 , 105 ., In addition , ThBC and ThC send excitatory signals to the input stages of the BG component ( StrD1 , StrD2 , STN ) 29 , 55 , 108 , 109 and are bi-directionally connected with M1 through excitatory links 26 , 27 , 53 ., The I terms of ThBC and ThC were computed using Eq 3 . The Cer component was built starting from the Marr-Albus type of model 110 , 111 proposed in 68 , 73 , 74 , as these are implemented with a level of abstraction that was similar to the one of the BG component ., In particular , the Cer includes four regions , each formed by a layer of leaky integrator units: the granule cells ( GC ) formed by 100 units; the Golgi cells ( GO ) formed by one inhibitory unit; the Purkinje cells ( PC ) formed by three units; the dentate nuclei ( DN ) formed by three units ., These numbers approximate the proportion of neurons observed in the real Cer 110–112 ., There is also a mossy fibers ( MF ) layer which receives excitatory connections from M1 and STN ., These circuits reproduce the functional effects of the M1 and STN activities on the cerebellar areas due to the pons-cerebellar link 26 , 30 ., GC transform the signal from MF for further processing by the PC ., According to the Marr-Albus theory , GC provide a sparse code , that is , a code with only a small fraction ( less than 10% in the model used here ) of cells active at any time ., In this way , the functioning of the cerebellum is facilitated because different MF inputs create highly dissimilar sparse GC activity patterns , which are easily recognizable by PC ., GO receives excitatory input from MF and GC , and provides a feedback inhibition to GC ., GO firing suppresses MF excitation of GC and thus tends to shorten the duration of bursts in the connections linking GC to PC ., This mechanism further supports the sparse coding of the input 73 ., PC show a spontaneous activity 112 that is influenced by parallel fibers—these are excitatory afferent inputs from GC ., PC also receive an input signal from M1 through the inferior olive-climbing fiber system—a climbing fiber is an axon of a neuron of the inferior olive ., This circuit is important for implementing Cer learning processes 113 ., In this respect , the inferior olive is commonly thought to compute an error signal conveyed to PC through nucleo-olivary projections ( refer to 114 for a detailed computational model ) ., In particular , in a model that would take into account the Cer learning processes , the output of DN should be subtracted from the M1 input to PC ., The inferior olive-climbing fiber system is also relevant to managing the timing of the input 115 ., Since the model did not aim to study the effects of Cer learning processes on tics , we abstracted the timing effect of such a system with a simple connection from M1 to PC ( see Fig 2 ) ., This link contributes to modulate the PC activity in a synchronous way with respect to the M1 activity ., The activity of the units of DN is modulated by the inhibitory connections from the corresponding units of PC ( one-to-one connections ) and by the excitatory collaterals from MF supplying a baseline activation for DN 116 ., The three units of DN , in turn , send excitatory signals to M1 ( through Th ) 26 and to StrD1 and StrD2 ( through ThC ) 29 , 105 ., The basic functioning of the Cer component is organized around the inhibitory PC , whose axons provide the only output of the cerebellar cortex ., Each unit of PC modulates the selection of a particular motor pattern within the dentate-thalamo-cortical system 117 ., In other words , similarly to what happens to BG , parallel sub-loops with the Cer component independently modulate a motor pattern allowing the selective facilitation of one response and the concurrent suppression of the others 26 , 35 , 45 ., When MF are silent ( i . e . , no input is received by Cer ) , PC show spontaneous activity and their inhibitory output prevents DN cells from firing ., This in turn prevents the selection of responses at such times ., We assumed that a previous learning process based on long-term depression ( LTD ) and long-term potentiation ( LTP ) 68 , 118 has led to having the GC-PC connections assume a high negative value when a motor pattern has to be selected by the input , and a small negative or positive value when a motor pattern should be inhibited ., The high negative value for the GC-PC synapse assures that the activity of the corresponding PC unit is close to zero and this in turn makes the corresponding DN unit positively activated 111 ., Consequently , excitation from MF collaterals predominates over inhibition from PC to DN related to the correct response ., DN neurons excite the thalamus that , in turn , excites the region in the motor cortex related to the correct response ., The I terms for the units of the Cer were computed using Eq 3 . The noise term n was set to zero for GC , GO , PC and DN ., We set by hand the value of the elements of wGC → PC by assuming that a previous learning process had led activity from GC to PC having a zero value when a motor pattern has to be selected by the input , and a positive value when no motor pattern has to be selected ., The values of the parameters of the equations are shown in the S2 Table ( see Supporting Information ) ., The activation recorded in the GC and PC layers is assumed to correspond to the firing rates measured within the cerebellar cortex of the monkeys ( labeled as “CbllCx” in 49 ) ., The M1 component is formed by three leaky units whose activity is assumed to correspond to the firing rate recorded in the primary motor cortex of the monkeys in the target experiment of McCairn and colleagues 49 ., M1 is bi-directionally connected with Th and projects to BG and Cer through excitatory links 26 ., The I term of the units of the M1 component was computed using Eq 3 ., This section compares the data on neural activity collected in 49 in the brain of one monkey and the data on neural activity collected in the brain of one subject simulated with the model ( data for other subjects are qualitatively similar ) ., Figs 3 and 4 show respectively the firing rate in the BG and in the M1 and Cer during TIC and NO-TIC trials ( i . e . , intertic intervals ) recorded in the monkey and in the model ., The model curves are obtained with, ( a ) an activation of cortex affected by the intrinsic neural noise of the various regions of the model;, ( b ) a further activation mimicking possible inputs to M1 from other cortical areas ( here captured , in the case of no-tic and tic cases , with a Gaussian-like input with a height of respectively 30 and 17 , and a standard deviation of respectively 0 . 040 and 0 . 250 sec ) ;, ( c ) dopaminergic bursts that capture the possible dopamine dysregulation caused by bicuculline ( here captured , in the case of no-tic and tic cases , with a Gaussian-like input with a height of respectively 1 and 50 and a standard deviation of respectively 0 . 600 and 0 . 020 sec ) ., The figures show that real and simulated data are very similar ., In both cases , in the Dorsal putamen and GPi there are no relevant differences in the firing rate amplitudes between the tic and no-tic state whereas there is a partial preservation of the response for GPe , with the early inhibitory peak maintained and the later excitatory peak increased during a tic ., By contrast , for M1 and Cer ( CbllCx in the figure ) the firing rate amplitudes during the tic state are greater than those measured during the no-tic state ., The model allows the simulation of the activity of other key areas not monitored in the target experiment 49 ., In particular , we measured the activity in STN and Th based on the hypothesis that these regions might be involved in tic production due to their potential role as mediators between M1 , BG , and Cer signals 26 , 55 , 105 ., Fig 5 shows that , similarly to what happens for M1 and CbllCx , in the STN and Th there is a remarkable difference in the activity amplitudes between tic and no-tic states ., This result represents a prediction of the model that could be tested in new experiments ., The abnormal activation of M1 in case of a tic supports the increase of activity in STN and Th ., The enhanced activity of STN , in turn , contributes to get a larger excitatory peak in GPe ( cf . section “Propagation of aberrant basal ganglia activity to primary motor cortex and cerebellum” ) ., In section “Discussion” , we further discuss the possible neural processes based on which STN and Th may be involved in tic production ., The results obtained with the model and presented in sections “The model reproduces data on firing rate during tic/intertic intervals” and “The model predicts an abnormal tic-related activity in the subthalamic nucleus and in the thalamus” are supported by statistical analysis of the data collected across 40 simulated subjects ., For each subject , we considered one trial randomly selected from the 10 trials ., In this way , we got 40 different measures across all the simulated subjects ., A two-way analysis of variance ( ANOVA ) was performed using the function aov of the statistical analysis software R . In more detail , the ANOVA was performed with two factors , namely the peak activity in the different areas ( i . e . , Dorsal putamen , GPi , GPe , STN , Th , M1 , CbllCx ) and the movement state ( i . e . , NO-TIC vs . TIC ) ., A post hoc test was also applied using the function TukeyHSD of R . A result was considered statistically significant if the p value was less than 0 . 001 ., The average value of the peak amplitude of the activity and its standard deviation for the areas of the model in TIC and NO-TIC trials are reported in the S3 Table , visually summarised in Fig 6 ., The ANOVA shows a statistically significant interaction between the activity in the different areas and the movement state ( p < 0 . 001 ) ., In addition , the post hoc tests show that , as in the experiment of McCairn and colleagues 49 , the differences in the activity amplitudes between TIC and NO-TIC trials are not statistical significant for the Dorsal putamen ( p = 0 . 990 ) and GPi ( p = 0 . 970 ) , whereas they are statistically significant for all other regions , in particular GPe , STN , Th , M1 , and CbllCx ( p < 0 . 001 for all of them ) ., Fig 7 shows the firing rate of the Dorsal putamen and M1 cells presented in 49 and obtained by recording neuron activity in the monkey model of tics ., The authors found that the striatal burst occurs 0 . 29 sec before the tic initiation ., This is followed by the activation of GPe and GPi , occurring respectively 0 . 26 sec and 0 . 19 sec before the tic onset , and by the activation of Cer and M1 respectively happening 0 . 11 sec and 0 . 12 sec before the tic onset ., The authors also found significant differences in the latency distribution of BG areas versus M1 and CbllCx , whereas they did not find significant differences in this distribution between M1 and CbllCx ., Overall , these findings suggest that in the animal model of 49 the tic event is triggered by the putamen as the activation of BG precedes that of M1 and Cer ., We obtained similar results in the model ., In more detail , to study the causality of the signal propagation in the model we computed the delay of the onset of the average activity in M1 with respect to the onset of the average activities in the other areas ., The delay was calculated by using the cross-correlation function ccf of the statistical analysis software R applied to the derivative of the signals ., The results of the cross-correlations are summarized in Fig 8 . The figure shows that in the tic state the onset of the average activity in the Dorsal putamen takes place 0 . 126 sec before the onset of the same signal in M1 ., Similarly , the onset of the average activity in GPe and GPi anticipates the onset of the same signal in M1 of respectively 0 . 116 sec and 0 . 124 sec .

Introduction, Methods, Results, Discussion

Motor tics are a cardinal feature of Tourette syndrome and are traditionally associated with an excess of striatal dopamine in the basal ganglia ., Recent evidence increasingly supports a more articulated view where cerebellum and cortex , working closely in concert with basal ganglia , are also involved in tic production ., Building on such evidence , this article proposes a computational model of the basal ganglia-cerebellar-thalamo-cortical system to study how motor tics are generated in Tourette syndrome ., In particular , the model:, ( i ) reproduces the main results of recent experiments about the involvement of the basal ganglia-cerebellar-thalamo-cortical system in tic generation;, ( ii ) suggests an explanation of the system-level mechanisms underlying motor tic production: in this respect , the model predicts that the interplay between dopaminergic signal and cortical activity contributes to triggering the tic event and that the recently discovered basal ganglia-cerebellar anatomical pathway may support the involvement of the cerebellum in tic production;, ( iii ) furnishes predictions on the amount of tics generated when striatal dopamine increases and when the cortex is externally stimulated ., These predictions could be important in identifying new brain target areas for future therapies ., Finally , the model represents the first computational attempt to study the role of the recently discovered basal ganglia-cerebellar anatomical links ., Studying this non-cortex-mediated basal ganglia-cerebellar interaction could radically change our perspective about how these areas interact with each other and with the cortex ., Overall , the model also shows the utility of casting Tourette syndrome within a system-level perspective rather than viewing it as related to the dysfunction of a single brain area .

Tourette syndrome is a neuropsychiatric disorder characterized by vocal and motor tics ., Tics represent a cardinal symptom traditionally associated with a dysfunction of the basal ganglia leading to an excess of the dopamine neurotransmitter ., This view gives a restricted clinical picture and limits therapeutic approaches because it ignores the influence of altered interactions between the basal ganglia and other brain areas ., In this respect , recent evidence supports a more articulated framework where cerebellum and cortex are also involved in tic production ., Building on these data , we propose a computational model of the basal ganglia-cerebellar-thalamo-cortical network to investigate the specific mechanisms underlying motor tic production ., The model reproduces the results of recent experiments and suggests an explanation of the system-level processes underlying tic production ., Moreover , it furnishes predictions related to the amount of tics generated when there are dysfunctions in the basal ganglia-cerebellar-thalamo-cortical circuits ., These predictions could be important in identifying new brain target areas for future therapies based on a system-level view of Tourette syndrome .

neuropsychiatric disorders, medicine and health sciences, neurochemistry, chemical compounds, dopaminergics, brain, vertebrates, neuroscience, organic compounds, animals, mammals, hormones, primates, amines, neurotransmitters, cerebellum, catecholamines, dopamine, monkeys, basal ganglia, neurochemicals, chemistry, motor cortex, thalamus, mental health and psychiatry, biochemistry, organic chemistry, anatomy, tourette syndrome, biogenic amines, biology and life sciences, physical sciences, amniotes, cerebral cortex, organisms

journal.pcbi.1005267

Role of Autoregulation and Relative Synthesis of Operon Partners in Alternative Sigma Factor Networks

Bacteria survive in stressful environmental conditions by inducing dramatic changes in their gene expression patterns 1 , 2 ., For a variety of stresses , these global changes in gene expression are brought about by the activation of alternative σ-factors that bind the RNA polymerase core enzyme and direct it towards the appropriate stress response regulons 3 ., Consequently , to ensure that these σ-factors are only active under specific environmental conditions , bacteria have evolved regulatory systems to control their production , activity and availability 3 , 4 ., These regulatory networks can be highly complex but frequently share features such as anti-σ-factors , partner switching mechanisms and proteolytic activation 4 ., The complexity of these networks has impeded a clear mechanistic understanding of the resulting dynamical properties ., In this study , we focus on one of the best studied examples of alternative σ-factors , the general stress-response regulating σB in Bacillus subtilis 5 to understand how the structure of the σ-factor regulatory networks is related to their functional response ., The σB-mediated response is triggered by diverse energy and environmental stress signals and activates expression of a broad array of genes needed for cell survival in these conditions 5 ., Activity of σB is tightly regulated by a partner-switching network ( Fig 1A and 1B ) comprising σB , its antagonist anti-σ-factor RsbW , and anti-anti-σ-factor RsbV ., In the absence of stress , RsbW dimer ( RsbW2 ) binds to σB and prevents its association with RNA polymerase thereby keeping the σB regulon OFF ., Under these conditions most of RsbV is kept in the phosphorylated form ( RsbV~P ) by the kinase activity of RsbW2 ., RsbV~P has a low affinity for RsbW2 and cannot interact with it effectively 6 ., However , in the presence of stress , RsbV~P is dephosphorylated by one or both of the dedicated phosphatase complexes: RsbQP for energy stress and RsbTU for environmental stress 7–10 ., Dephosphorylated RsbV attacks the σB-RsbW2 complex to induce σB release , thereby turning the σB regulon ON 11 ., Notably , the genes encoding σB and its regulators lie within a σB-controlled operon 12 , thereby resulting in positive and negative feedback loops ., Recently , it was shown that under energy stress σB is activated in a stochastic series of transient pulses and increasing stress resulted in higher pulse frequencies 13 ., It has also been shown that increase in environmental stressor such as ethanol leads to a single σB pulse with an amplitude that is sensitive to the rate of stressor increase 14 ., While it is clear that the pulsatile activation of σB is rooted in the complex architecture of its regulatory network ( Fig 1A and 1B ) its mechanism is not fully understood ., Previous mathematical models of the σB network either did not produce the pulsatile response 15 or made simplifications to the network 13 that are somewhat inconsistent with experimentally observed details ., As a result , it remains unclear which design features of the σB network enable its functional properties ., To address these issues we develop a detailed mathematical model of the σB network and examine its dynamics to understand the mechanistic principles underlying the pulsatile response ., By decoupling the post-translational and transcriptional components of the network we show that an ultrasensitive negative feedback between the two is the basis for σB pulsing ., Moreover we find that the relative synthesis rates of σB and its operon partners RsbW and RsbV , plays a critical role in determining the nature of the σB response ., We also use our model , together with previously published experimental data from 13 , 14 , to explain how the σB network is able to encode the rate of stress increase and the size of stochastic bursts of stress phosphatase into the amplitudes of σB pulses ., We further develop this model to investigate how the network functions in the context of other σ-factors ., As in many other bacteria , σB is one of the many σ-factors that complex with RNA-polymerase core that is present in limited amounts 3 , 16 ., Therefore , when induced these alternative σ-factors compete with one another and the housekeeping σ-factor σA for RNA polymerase ., We use our model to investigate how the design of this network enables it to function even in the presence of competition from σA which has a significantly higher affinity for RNA polymerase 17 ., Lastly , we investigate how multiple alternative σ-factors compete when cells are exposed to multiple stresses simultaneously ., Using our model we identify design features that are ubiquitous in stress σ-factor regulation and critical to bacterial survival under diverse types of stresses ., In a recent study , Locke et al . 13 demonstrated that a step-increase in energy stress results in pulsatile activation of σB ., The study also proposed a minimal mathematical model of the network which reproduced pulsing in σB ., However , this model included several assumptions inconsistent with experimentally observed details:, ( i ) Phosphorylation and dephosphorylation reactions were assumed to follow Michaelis-Menten kinetics despite the fact that kinase ( RsbW ) and phosphatase concentrations are known to be comparable to substrate ( RsbV ) concentrations 18 so the approximation breaks down 19 ,, ( ii ) σB and RsbV are represented as a single lumped variable rather than separate species and ,, ( iii ) partner-switching , and the formation and dissociation of various RsbW2 complexes were not included explicitly ., Though this minimal model produces pulses resembling their experimental observations , it does not depict a biochemically accurate picture of the σB network ., Consequently it cannot be used to uncover the design features that enable σB pulsing ., To understand the σB network response we built on our earlier study 15 to develop a detailed mathematical model that explicitly includes all known molecular interactions in the network ., Note that we made one significant change to the model discussed in 15 ., The model in 15 assumed that the synthesis rates for σB and its operon partners ( RsbW and RsbV ) follow the stoichiometry of their binding ratios ( i . e . RsbWT/BT = 2 and RsbWT/RsbVT = 1; where BT , RsbWT and RsbVT represent total σB , RsbW and RsbV concentrations respectively ) ., However experimental measurements have shown that σB , RsbW and RsbV are produced in non-stoichiometric ratios 18 ., The exact mechanism underlying these non-stoichiometric ratios is currently only incompletely understood ., However , analysis of the open-reading frames in the operon showed that rsbV and rsbW may be translationally coupled due to overlapping termination and initiation codons 20 which may ensure that they are expressed in similar amounts ., The same analysis also showed that the rsbW and sigB reading frames overlapped and that this overlap was preceded by a region of dyad symmetry which may form a stem-loop structure 20 ., These features may interfere with sigB translation and lead to lower expression of σB than its binding partners RsbV and RsbW ., To account for these features , in contrast to our earlier study , we assumed σB , RsbW and RsbV can be produced in non-stoichiometric ratios and studied how changes in relative synthesis rates of σB operon partners affect the response of the σB network to step-increases in energy stress phosphatase levels ., We note that RsbX , a negative regulator of RsbTU phosphatase 21 , is not included in our model ., RsbX was excluded for simplicity since it is not essential for the pulsatile response of the σB network 14 ., Simulations of this detailed model showed that different combinations of RsbW:σB ( λW ) and RsbV:σB ( λV ) relative synthesis rates lead to qualitatively different dynamical responses of the σB network ., For operon partner synthesis ratios similar to those estimated in 18 , ( i . e . RsbWT > 2BT and RsbWT ≈ RsbVT ) our model responded to a step-up increase of the phosphatase with a pulsatile σB response ( Fig 1C ) that resembled the experimentally observed behavior 13 ., In contrast , when RsbW:σB and RsbV:σB relative synthesis rates follow the stoichiometry of their binding ratios pulsing is not observed and the σB activity monotonically increases over time ( Fig 1D ) ., Pulsing also disappears when RsbW synthesis is high enough to neutralize both its binding partners ( Fig 1E ) ., To understand why the pulsatile response is only observed for certain operon partner synthesis rates , we investigated our mathematical model by decoupling the network’s transcriptional and post-translational responses ( as shown in Fig 1A ) ., By varying the σB operon transcription rate , while keeping the relative synthesis rates of RsbW:σB ( λW ) and RsbV:σB ( λV ) fixed , we were able to calculate the post-translational response ( Fig 2A , blue curve ) of the σB network: σB = Fp ( BT , PT ) ., This function describes how the free σB concentration varies as a function of BT ( total concentration of σB ) and PT ( total phosphatase concentration ) ., Note that although we refer only to BT for brevity , RsbWT and RsbVT are always assumed to increase in proportion to the BT for this post-translational response ., This post-translational function is analogous to an in vitro assay wherein various combinations of total σB ( BT−and proportional amounts of RsbWT and RsbVT ) and total phosphatase ( PT ) are mixed together and then the resulting free σB concentration is measured ., In parallel , we calculated the transcriptional response ( Fig 2A , black curve ) BT = FT ( σB ) which analogous to a transcriptional reporter construct in vivo , describes how changes in the free σB concentration affect total σB concentrations ( and RsbWT and RsbVT concentrations which are always proportional to BT ) ., In this analysis framework , the steady state of the complete closed loop network can be determined by simultaneously solving the post-translational and transcriptional equations , σB = FP ( BT , PT ) and BT = FT ( σB ) at each phosphatase concentration PT ., Graphing both functions provided the steady-state solution as their intersection point ( Fig 2A , red circle ) ., This decoupling approximation allows us to quantify the sign and strength of feedback in the full model ., The effective sign of the feedback in the σB network is given by the sign of the product of the sensitivities of two response functions , i . e . sign ( ( ∂FT / ∂σB ) · ( ∂FP / ∂BT ) ) ., Since σ-factors function as activators of transcription , FT ( σB ) is a monotonically increasing function of σB ( i . e . ∂FT / ∂σB > 0 ) ., Consequently , the sign of the feedback in the σB network is given by the sign of the sensitivity of the post-translational response to RsbBT ( i . e . ∂FP / ∂BT ) ., In other words , if increase in the operon production leads to an increase in free σB then the feedback is positive , whereas if increase in the operon production leads to a decrease in free σB then the feedback is negative ., Our results show that for the parameters chosen in Fig 1C FP is a non-monotonic function of BT ( Fig 2A , blue curve ) ., At low RsbBT , free σB increases as a function of BT because RsbW is sequestered in the W2V2 complex ., However at higher BT , the kinase flux dominates the phosphatase flux resulting in an increased RsbV~P and the freeing of RsbW2 from RsbV ., Freed RsbW2 sequesters σB in the W2σB complex ., Furthermore , in the total σB concentration range where ∂FP / ∂BT < 0 in Fig 2B , the post-translational response is quite steep ( Fig 2A ) , i . e . small changes in BT lead to significant decreases in free σB ., This ultrasensitivity can be quantified by calculating the slope in logarithmic space , i . e . This dimensionless quantity characterizes the ratio of relative changes in σB and BT at steady state ( Fig 2B ) ., The sign of LGP defines the effective sign of the feedback loop and if the magnitude of |LGP| > 1 defines an ultrasensitive response ., For the σB network , in the region around the steady state LGP < −1 indicating that the σB network operates in an ultrasensitive negative feedback regime ., Two types of post-translational reactions that are known to produce ultrasensitivity play a role here ( S1A and S1B Fig ) : ( 1 ) Zero-order ultrasensitivity due to competition between RsbW kinase and RsbQP/RsbTU phosphatases for RsbV and ( 2 ) molecular titration due to sequestration of σB by RsbW ., Notably around the steady state , whereas both the fraction of unphosphorylated RsbV and the fraction of free σB decrease ultrasensitively as a function of increase in operon expression ( proportional to BT ) the latter is far more sensitive ( S1C Fig ) ., This indicates that molecular titration between σB and its binding partners may contribute more to the ultrasensitivity of the post-translational response than the zero-order competition between RsbW and stress phosphatases ., Irrespective of their relative contributions however , our results show that both mechanisms combine to ensure that near the steady state the σB network operates in an ultrasensitive negative feedback regime ., Notably , negative feedback is one of the few network motifs capable of producing adaption-like pulsatile responses 22 ., Moreover , ultrasensitivity of the feedback ensures homeostatic behavior—making the steady state robust to variations of parameters 22 ., This explains why in Fig 1C a step-increase in the phosphatase concentration in our model leads to a σB pulse followed by return to nearly the same steady state ., Plotting the trajectory of the σB pulse ( green curve , Fig 2C ) on the ( σB , BT ) plane and over the post-translational and transcriptional responses ( Fig 2C ) illustrates the mechanism driving this pulsatile response ., Starting at the initial steady state ( red circle ) , an increase in phosphatase shifts the ultrasensitive post-translational response ( cyan to blue curve ) so that free σB is rapidly released from the RsbW2-σB complex whereas total σB levels remain relatively unchanged ., The increase in σB operon transcription eventually causes accumulation of total σB and the anti-σ-factor RsbW ., This in turn forces the σB level to decrease , following the post-translational response curve , to the new steady state ( gray circle ) which has very little free σB thereby completing the σB pulse ., The same analysis can be applied for different values of relative synthesis rates , i . e . those that correspond to Fig 1D and 1E ., As shown in S2 Fig these parameter values do not produce an ultrasensitive non-monotonic post-translational response ., Consequently they do not lead to the emergence of overall negative feedback explaining their non-pulsing dynamics ., To determine if the presence or absence of negative feedback more generally explains the different dynamical responses in Fig 1C–1E , we sampled different combinations of relative synthesis rates ( RsbWT / BT = λW and RsbVT / BT = λV ) and calculated the post-translational sensitivities ., Our calculations showed that based on the sign of post-translational sensitivity ( LGP ) the relative synthesis parameter space can be divided into three regions ( Fig 2D ) ., For ( λW , λV ) combinations in Region I the sensitivity is always positive ., Increase in λW leads the system into an ultrasensitive negative regime ( LGP < 0 and |LGP| ≫ 1 ) in Region II ., A further increase in λW or a decrease in λV transitions the system into a non-responsive ( LGP ∼ 0 ) state in Region III ., Dynamic simulations for sampled ( λW , λV ) combinations confirm that pulsatile responses to step-up in phosphatase concentration are restricted to Region II where the effective feedback is negative ( S2 Fig ) ., To understand the boundaries between the three regions and how the level of the phosphatase affects the network , we developed a simplified analytical model that is based on the observation that RsbW and RsbV bind strongly to each other 18 ( see S1 Text for details ) ., This approximation allowed us to determine the boundaries in Fig 2D ( black and red lines ) and resulted in a clear biological interpretation of the three regions ., In Region I the amount of RsbW , irrespective of phosphatase level , is insufficient to bind all of its partners and consequently some fraction of σB always remains free or unbound to RsbW ., In contrast in Region II , the amount of phosphatase determines how much RsbV is in its inactive phosphorylated form RsbV~P and therefore whether the amount of RsbW is sufficient to bind all of its partners depends on the levels of RsbV~P ., As a result , for this region , the ratio of kinase and phosphatase ( PT ) fluxes determines the post-translational response ., Lastly , Region III is the opposite of Region I in that the amount of RsbW is more than sufficient to bind all of its partners , even when all RsbV is unphosphorylated ., As a result , irrespective of phosphatase levels , very little σB is free and its level is nearly insensitive to changes in total σB ., Thus negative feedback and consequently pulsing are only possible in Region II where changes in phosphatase can shift the balance between the prevalent partner complexes ., The role of negative feedback in producing a pulsatile response also explains why pulsing does not occur in strains where σB operon is transcribed constitutively 13 ., In this case , the σB network lacks the negative feedback necessary to produce a pulsatile response ., A step-increase in phosphatase still leads to an increase in free σB due to the change in the post-translational response; however , this not followed by an increase in total σB levels ( S2C Fig ) ., Consequently , an increase in phosphatase results in a monotonic increase in free σB rather than a pulse ( S2F Fig ) ., The only actual measurements of λW and λV were made by Delumeau et al . 18 using a quantitative western blot assay ., Interestingly they report that λW = 2 . 9 , λV = 1 . 7 in the absence of stress and λW = 2 . 4 , λV = 2 . 65 in the presence of stress ., These measurements suggest that the ratios might change depending on whether cells are under stress ., Although the mechanism underlying this change is unclear , our model predicts that both measured ratio pairs lie within the negative feedback regime shown in Fig 2D ., Accordingly our simulations show that the network responds to step-increases in phosphatase levels with a pulsatile response for both pre-stress and post-stress ( λW , λV ) values ( S3 Fig ) ., However , due to reduced ultrasensitivity of the system for these parameters , concertation of free σB following increase in the stress ( phosphatase ) does not perfectly adapt to the pre-stress value ( S3 Fig ) ., In an attempt to match the near-perfect adaptation reported in Refs ., 13 , 14 we’ve chosen to do further analysis with λW = 4 and λV = 4 . 5 ., Notably , our simulations also showed that it is not essential for the phosphatase level to remain fixed after a stress-induced step-increase ., In fact , we found that a dilution mediated decline in phosphatase level post-step-increase has little impact on the pulse amplitude ( S4 Fig ) ., This observation can be explained by the relatively rapid dynamics of the post-translational response as compared to the gradual nature of dilution and suggests that pulsatile dynamics are relevant even for experimental conditions where phosphatase levels do not remain fixed in stressful conditions 14 , 23 ., Further our decoupling method also sheds light on another experimental observation by Locke et al . 13: the dependence of σB pulse amplitude on the phosphatase level ., Specifically , we found that σB pulse amplitude is a threshold-linear function of the phosphatase concentration ( S5 Fig ) ., Our decoupling method shows that this threshold-linear behavior arises because the σB network only operates in a negative feedback regime for phosphatase concentrations higher than a threshold ., Below the phosphatase threshold , the post-translational response σB = FP ( BT , PT ) ∼ 0 and is insensitive to RsbBT ( S5B and S5C Fig ) ., Thus , the full system lacks the negative feedback and as a result σB does not pulse ., Using our analytical approximation we found that this phosphatase threshold is proportional to the basal level of RsbW kinase synthesis rate and the ratio of the kinase and phosphatase catalytic rate constants ( S5D and S5E Fig ) ., Increase in the basal σB operon expression rate increases the phosphatase threshold ., Further , an increase in the relative synthesis rate of RsbW ( λW = RsbWT / BT ) makes the phosphatase threshold more sensitive to the σB operon expression rate , whereas a decrease in ratio of the kinase and phosphatase catalytic rate constants makes it less sensitive ( S5D and S5E Fig ) ., This shows that the phosphatase threshold represents the concentration at which the phosphatase is able to match the basal kinase flux ., Altogether these results show how the ultrasensitive negative feedback plays a critical role in determining many properties of the σB network pulsatile response and how the decoupling method can facilitate the identification of essential design features that enable the existence of this negative feedback ., In the preceding sections we have shown how the σB network responds to a step-increase in RsbQP or RsbTU phosphatases by producing a single pulse of activity ., However , Locke et al . 13 have shown that an increase in energy stress leads to a sustained response with a series of stochastic pulses in σB activity ., This study further showed that this sustained pulsing response is driven by noisy fluctuations in level of energy-stress-sensing phosphatase RsbQP ., While the mean level of RsbQP is regulated transcriptionally by energy stress , its concentration in single cells can fluctuate due to the stochasticity of gene expression 8 , 13 ., To determine if our model could explain this response to stochastic fluctuations in RsbQP , we modified it to include fluctuations in the concentration of this phosphatase ., Based on previous theoretical 24 , 25 and experimental 26 studies we assume that fluctuating phosphatase level follows a gamma distribution which is described by two parameters—burst size ( b , average number of molecules produced per burst ) and burst frequency ( a , number of bursts per cell cycle ) ., The mean phosphatase in this case is the product of burst size and burst frequency ( 〈PT〉 = ab ) ., Thus , energy stress can increase mean phosphatase by changing burst size or burst frequency or both ., In other words , stress conditions can increase phosphatase levels by either producing more phosphatase molecules per transcription-translation event or by making these events more frequent ., While the results of 13 cannot exclude either mechanism , we can use our model to uncover which mechanisms is dominant ., First , we performed stochastic simulations in which mean phosphatase concentration was varied by changing burst size ., These simulations reproduced all the experimentally-observed features of the σB pulsatile response ., Specifically our results show that stochastic bursts in stress phosphatase levels lead to pulses of σB activity ( Fig 3A ) ., Moreover , consistent with the experimental observations of 13 , our model showed that the amplitude of σB pulses increases linearly with the stress phosphatase level ( Fig 3A and 3B ) ., Finally , we found that stress-mediated increases in phosphatase concentration lead to an ultrasensitive ( effective Hill coefficient ~5 . 6 ) increase in the frequency of σB pulsing ( Fig 3C ) and an ultrasensitive ( effective Hill coefficient ~2 ) increase in the level of σB target expression ( Fig 3D ) ., Next , we compared these results with stochastic simulations in which burst frequency was modulated ( Fig 3E–3H ) ., These simulations also led to an increase in σB pulsing ( Fig 3E ) and a non-linear increase in the level of σB target expression as mean phosphatase level was increased with more frequent bursts ( Fig 3H ) ., However , we found that σB pulse amplitude remains constant for burst frequency modulation ( Fig 3E and 3F ) unlike the ~5-fold increase for burst-size modulation ( Fig 3B ) ., Moreover , the frequency of σB pulses increase linearly with phosphatase level unlike the non-linear increase observed with burst-size-increase simulations ( compare Fig 3C and 3G ) ., Notably the experimental observations reported in 13 show that σB pulse amplitude does increase ( ~3-fold ) with an increase in energy stress thus suggesting that increase in phosphatase concentration at high stress is primarily the result of increase in burst size ., To further reinforce the role of mean burst-size modulation in controlling the σB pulsatile response we next examined the cumulative histograms of pulse amplitudes at different phosphatase concentrations ., These histograms carry different signatures for burst-size or burst-frequency encoding ., The distribution of pulse amplitudes is unchanged with increase in burst frequency ( S6A Fig ) because σB pulse amplitude is determined by phosphatase burst size and not burst frequency ., In contrast , if phosphatase levels are controlled by changing mean burst size then the distribution of pulse amplitudes changes accordingly ., Consequently , the normalized cumulative histograms of pulse amplitudes overlap for burst-frequency encoding ( S6A Fig ) but not burst-size encoding ( S6B Fig ) ., Applying this test to the data from 13 , we found that the normalized cumulative pulse amplitudes histograms do not overlap ( S6C Fig ) ., These results predict that stress affects the σB network via burst-size modulation of phosphatase production which is then encoded into σB pulse amplitudes ., While the molecular mechanism that introduces energy stress to the network is still not fully understood , our prediction places an important constraint on it ., Our model can also be used to study the response of σB network to environmental stress ., Unlike the energy stress phosphatase , the environmental stress phosphatase RsbU is regulated post-translationally by binding of RsbT 27–29 ., RsbT is trapped by its negative regulators under unstressed conditions but is released upon stress ., Consequently , the concentration of RsbTU complex is tightly controlled at the post-translational level and is therefore expected to be relatively insensitive to gene expression fluctuations but sensitive to the level of environmental stress ., As a result , step-up increases in environmental stress agents like ethanol produce rapid increases in RsbTU and result in only a single pulse of σB activity 14 ., However it has been shown that for gradual increases in stress , σB pulse amplitude depends on the rate of stress increase 14 ., To explain this response , we modeled gradual stress with ramped increase in RsbTU complex concentration ( Fig 4A ) ., Our simulations showed that the detailed model of σB network is indeed able to capture the effect of rate of stress increase on σB pulse amplitudes ., Specifically for a fixed increase in RsbTU complex , the pulse amplitude decreases non-linearly as a function of the duration of phosphatase ramp ( Fig 4B and 4E ) ., We hypothesized that this ramp rate encoding is the result of the timescale separation between the fast post-translational and the slow transcriptional responses of the σB network ., During the pulsed σB activation , post-translational response is rate-limited by the phosphatase ramp ., In contrast , the transcriptional response is slow and its rate is set by the degradation rate of σB operon proteins ., Following a step-increase in phosphatase , the fast post-translational response ensures that σB reaches its post-translational steady state before the slow increase in RsbW sequesters σB and turns off the pulse ( Fig 4A and 4B ) ., However , for a ramped increase in phosphatase the post-translational increase in σB is limited by the rate of phosphatase ramp ., This allows RsbW to catch up and terminate the σB pulse earlier , thereby decreasing the pulse amplitude ., To test this , we varied the degradation rate of σB operon proteins and proportionally changed the operon transcription rate to ensure that the total concentrations of σB , RsbW and RsbV are kept fixed ., We found that indeed pulse amplitude decreases with increase in degradation/dilution rate ( Fig 4C and 4D ) ., Our simulations showed that Kramp , the half-maximal constant for the dependence of pulse amplitude on ramp duration , was indeed sensitive to the degradation rate ( Fig 4E and 4F ) ., This suggests that the timescale separation between the post-translational and transcriptional responses is the basis of ramp rate encoding into pulse amplitude ., The results thus far indicate that σB network functions in the effectively negative feedback regime where increase in the operon expression decreases σB activity ., Negative feedback loops have been shown to increase the robustness of the system to perturbations ., We therefore decided to investigate how the σB network design affects its performance when it faces competition for RNA polymerase from other σ-factors , e . g . from the housekeeping σ-factor σA 16 , 30 , 31 ., Since σA has a much higher affinity for RNA polymerase 17 , a small increase in σA can dramatically increase the amount of σB necessary to activate the transcription of the σB regulon ., Thus , changes in σA can alter the input-output relationship of a stress-response σ-factor like σB ( S7A and S7B Fig ) and thereby adversely affect the survival of cells under stress ., To understand how the σB network handles competition for RNA polymerase , we expanded our model to explicitly include σA , RNA polymerase ( RNApol ) and its complexes with both σ-factors ., The presence of σA will affect transcriptional activity of σB but not post-translational interactions between σB operon partners ( Fig 5A , left panel ) ., Therefore , post-translational response σB = FP ( BT , PT ) is not affected by σA ., In contrast , in the transcription response , an increase in σA decreased the ‘effective affinity’ of σB for RNApol and consequently higher levels of free σB are necessary to achieve the same production rate for σB target genes ., Using our model , we examined how changes in σA level affect the network response to energy stress signal , i . e . under stochastically fluctuating RsbQP phosphatase levels ., Our simulations showed that phosphatase bursts lead to pulses of free σB and pulsatile transcription of σB-controlled promoters ( Fig 5B and 5C ) as the presence of σA does not affect the effective feedback sign ., Notably our results also showed that the amplitudes of σB target promoter pulses are hardly affected by a ~30% increase in σA ( Fig 5C , left panel ) ., This surprising insensitivity of the phosphatase-σB target dose-response to RNApol competition is the result of the ultrasensitive negative feedback between free σB and total σB ., Due to the ultrasensitivity of this feedback , a small decrease in total σB levels resulting from the increase in σA causes a large increase in σB pulse amplitude ( Fig 5B left panel , Fig 5D green line ) ., This increased amplitude compensates for the increased competition for RNApol and insulates the network from perturbations ( Fig 5D and 5E , green curves ) ., To further illustrate the importance of the negative feedback in insulating the network , we compared the response of the wildtype network to an “in silico” mutant network wherein the σB operon is constitutive rather than σB dependent ( Fig 5A ) ., Consequently this network lacks any feedback between free σB and total σB ., Our simulations ( Fig 5B , right panel ) show that the free σB concentration of the no-feedback-network does not show adaptive pulsing and therefore σB concentration fluctuates along with the phosphatase levels ., Increase in σA did not affect this response ., This is expected since in the absence of feedback σA only af

Introduction, Results, Methods

Despite the central role of alternative sigma factors in bacterial stress response and virulence their regulation remains incompletely understood ., Here we investigate one of the best-studied examples of alternative sigma factors: the σB network that controls the general stress response of Bacillus subtilis to uncover widely relevant general design principles that describe the structure-function relationship of alternative sigma factor regulatory networks ., We show that the relative stoichiometry of the synthesis rates of σB , its anti-sigma factor RsbW and the anti-anti-sigma factor RsbV plays a critical role in shaping the network behavior by forcing the σB network to function as an ultrasensitive negative feedback loop ., We further demonstrate how this negative feedback regulation insulates alternative sigma factor activity from competition with the housekeeping sigma factor for RNA polymerase and allows multiple stress sigma factors to function simultaneously with little competitive interference .

Understanding the regulation of bacterial stress response holds the key to tackling the problems of emerging resistance to anti-bacteria’s and antibiotics ., To this end , here we study one of the longest serving model systems of bacterial stress response: the σB pathway of Bacillus subtilis ., The sigma factor σB controls the general stress response of Bacillus subtilis to a variety of stress conditions including starvation , antibiotics and harmful environmental perturbations ., Recent studies have demonstrated that an increase in stress triggers pulsatile activation of σB ., Using mathematical modeling we identify the core structural design feature of the network that are responsible for its pulsatile response ., We further demonstrate how the same core design features are common to a variety of stress response pathways ., As a result of these features , cells can respond to multiple simultaneous stresses without interference or competition between the different pathways .

cellular stress responses, engineering and technology, enzymes, signal processing, cell processes, enzymology, operons, dna-binding proteins, phosphatases, polymerases, network analysis, dna, computer and information sciences, stress signaling cascade, proteins, rna polymerase, modulation, biochemistry, signal transduction, biochemical simulations, cell biology, nucleic acids, genetics, biology and life sciences, computational biology, cell signaling, signaling cascades

journal.pgen.0030058

Novel Crohn Disease Locus Identified by Genome-Wide Association Maps to a Gene Desert on 5p13.1 and Modulates Expression of PTGER4

Crohn disease ( CD ) is a chronic relapsing inflammatory disorder of the intestinal tract , described for the first time in the 1920s 1 ., Lifetime prevalence has increased to current estimates of ∼0 . 15% in Caucasians ., The precise environmental causes underlying this rise remain essentially unknown , but familial clustering and twin studies clearly identify an inherited component to predisposition ., More than ten susceptibility loci have been identified by linkage and/or association studies and convincing causative mutations have been reported , particularly in CARD15 2–3 ., As known loci dont fully account for the genetic risk for CD we performed a genome-wide association scan ( GWA ) to contribute to the identification of additional susceptibility loci ., Genotype data from the Illumina HumanHap300 Genotyping Beadchip 4 were obtained on 547 Caucasian CD patients from Belgium and compared to genotypes for 928 healthy controls from Belgium and France ., Genotype call rates were >93% for all individuals included in the study ., Of the total 317 , 497 SNPs available , 15 , 046 with genotyping success rate of less than 96% or deviating from Hardy-Weinberg proportions in controls ( Fishers exact test p ≤ 10−3 ) were eliminated from further analysis as it is known that less reliable markers generate spurious associations ., For the remaining 302 , 451 SNPs , we compared allele frequencies between cases and controls as outlined in Methods ., Figure 1 shows the 10 , 000 most significant p-values obtained across the human genome ., Regions on Chromosomes 1 , 5 , and 16 harbored clusters of markers with suggestive evidence of association at significance levels between 10−6 and 10−10 ., The significance of tests of association with these markers remained within this range after controlling for possible effects of population structure using a backwards stepwise regression 5 ., The strongest association was found with markers of the IL23R gene on Chromosome 1 , which has recently been identified as a novel CD susceptibility locus in a case-control and family-based association study of Caucasian and Jewish cohorts 6 ., In our data , two markers of the IL23R gene , rs11209026 and rs11465804 , gave the most significant association signals ( p < 10−9 ) ., Rs11209026 corresponds to an Arg381Gln substitution in IL23R while rs11465804 is intronic and in strong linkage disequilibrium ( LD ) with the former marker ., A marker within the CARD15 gene on Chromosome 16 , which is the first susceptibility gene to have been identified in CD 3 , also showed suggestive evidence of association ( rs5743289; p < 10−6 ) ., We also examined the results of the GWA with respect to other previously reported susceptibility loci , including OCTN 7 , DLG5 8 , TNFSF15 9 , and ATG16L1 10 ., None of these obtained a similar level of significance for association in our study ., Genotyping our cohorts for other SNPs at these loci that are reported in the literature to be associated with CD did not improve the signals , with the exception of rs224188 corresponding to a Thr-to-Ala substitution within ATGL16L1 ( p < 2 × 10−4 ) , thus providing confirmation of this novel susceptibility locus for the first time 10 ., On Chromosome 5p13 . 1 , we identified a region of approximately 250 kb that contained six markers with p < 10−6 in the association test ( Figure S1 ) ., This region has not previously been reported as a CD susceptibility locus ., We selected ten markers from the regions of IL23R and 5p13 . 1 for confirmation genotyping in up to 1 , 266 additional Caucasian CD patients and 559 additional controls ., The IL23R locus was included in the confirmation genotyping as it had not yet been reported at the time of our study 6 ., The associations at these two loci were clearly replicated with p-values as low as 4 . 2 × 10−7 at the IL23R and 3 . 7 × 10−4 at 5p13 . 1 ( Table 1 ) ., In the combined data from the GWA and replication studies , we obtained p-values as low as 2 . 2 × 10−18 at IL23R and 2 . 1 × 10−12 at the 5p13 . 1 locus ., In addition , we genotyped trios with non-affected parents for the same SNPs to perform a transmission disequilibrium test ( TDT ) ., The ten SNPs were typed on 137 trios with affected offspring included in the case-control study , while two of the 5p13 . 1 SNPs were typed on an additional 291 independent trios , also originating from Belgium ., Significant over-transmission of the associated alleles were found at both loci , thus providing additional confirmatory evidence in support of the IL23R1 and 5p13 . 1 susceptibility loci ( Table 1 ) ., To further characterize the novel 5p13 . 1 locus , we genotyped a subset of 1 , 092 CD patients and 374 Belgian controls for 111 markers ( average interval: 2 . 3 kb ) spanning the 250-kb segment ., We determined the most likely linkage phase for each individual using PHASE 11 , and used the corresponding haplotype frequencies to quantify the level of LD between all marker pairs ., The 250 kb encompass five clearly delineated LD blocks , the central one ( block III ) being the largest and spanning 122 kb ( Figure 2A ) ., We first performed single-marker association analyses ., The strongest effects were observed within the 122-kb block III with several SNPs yielding p-values <10−5 ., p-values <10−3 and 10−4 were observed in flanking blocks II and IV , respectively ( Figure 2B ) ., We then performed haplotype analysis of the region spanned by blocks II to IV ., For block III , 20 haplotypes accounted for 93% of the observed chromosomes ., These could be grouped in three clades comprising respectively six ( IIIA ) , six ( IIIB ) , and two ( IIIC ) haplotypes , plus a group of six haplotypes that apparently originated from various recombination events ., Likewise , evaluation of block II revealed three clades ( with respectively two IIA , three IIB , and two IIC haplotypes ) and two recombinant haplotypes , while block IV was characterized by two clades with two ( IVA ) and one ( IVB ) haplotype respectively ., We compared the clade frequencies in cases and controls at intervals bounded by ancestral recombination events ( Figure 2C ) ., In agreement with the results of the single-marker analysis , the most significant associations were found in block III followed by IV and II ., To verify whether the entire 5p13 . 1 effect could be attributed to block III ( i . e . , the effects observed for blocks II and IV would be mere echos of the block III effect ) , we performed a multi-variate analysis as described in Methods ., The clade effects of blocks II and IV conditional on the effect of block III and vice versa remained significant ( p ( II|III ) = 0 . 023 , p ( III|II ) = 0 . 0004 , p ( IV|III ) = 0 . 003 , and p ( III|IV ) = 0 . 026 ) , suggesting that multiple variants in the region may jointly account for the observed effect on CD ., Commonly occurring recombinant haplotypes in blocks II and III caused local drops in significance , thus suggesting that causal variants lie outside the corresponding subsegments ( Figure 2C ) ., No known genes or CpG islands were found within the region of association on 5p13 . 1 after examination with the Ensembl and UCSC genome browsers ., The region has an average GC content of 38% , and an excess of interspersed repeats given GC content ( 58 . 36% versus 42 . 3% ) , which is mainly due to an excess of LINE1 ( 33 . 05% versus 19 . 6% ) and LTR elements ( 15 . 36% versus 7 . 70% ) 12 ., It contains 98 highly conserved elements 13 ., It is part of a 1 . 25-Mb gene desert between DAB2 ( 850 kb distally from the block ) and PTGER4 ( 270 kb proximally from the block ) ., Interestingly , several of the genes flanking the region have been implicated in pathogenesis of CD , or are related to genes that have been implicated in the disease ., These include a member of the caspase recruitment domain family ( CARD6 ) , three complement factors ( C6 , C7 , and C9 ) , and—most notably—the prostaglandin receptor EP4 ( PTGER4 ) , which resides closest to the group of disease associated markers ( Figure S1 ) ., One hypothesis is that the disease-associated region contains cis-acting regulatory elements that control the expression levels of the causal gene ( s ) located in the vicinity , and that the causal variants modulate the activity of these elements ., As a first step to test this , we studied the effect of SNPs in the disease-associated region on the expression levels of neighboring genes ., To that end we exploited a database of genome-wide gene expression ( Affymetrix HG-U133 Plus 2 . 0 chips; http://www . affymetrix . com ) measured in EBV-transformed lymphoblastoid cell lines from 378 individuals genotyped with the Illumina HumanHap300 Genotyping Beadchip ( W . Cookson , unpublished data ) ., Remarkably , eight of the 26 Illumina markers spanning 264 kb coinciding precisely with the CD-associated region yielded p-values <2 × 10−3 for PTGER4 ( Figure 2B ) ., Three of the markers influencing PTGER4 expression are located in block III ( rs16869977 , rs10512739 , and rs6880934 ) ., The first two are tagging the IIIBa sub-clade ( Figure 2C ) , while the third one is in complete LD with it ( IIIA + IIIBa ) ., The corresponding SNPs did not show convincing evidence for association with CD ., Two strongly associated SNPs ( D′ = 0 . 84 ) located respectively in block IV ( rs4495224 ) and V ( rs7720838 ) showed the most significant effect on PTGER4 expression and were also associated with CD ( Table 1 ) ., The rs4495224 A and rs7720838 T risk alleles were associated with increased PTGER4 expression ., Although these results must be treated as preliminary , they tend to support the hypothesis that the disease-associated polymorphisms may be related to the expression levels of one or more genes in the region ., CD is the most common form of inflammatory bowel disease , the other being ulcerative colitis ( UC ) ., We genotyped a cohort of 246 Belgian UC patients ( Caucasians ) for IL23R ( rs11209026 ) , ATG16L1 ( rs2241880 ) , and the novel 5p13 . 1 locus ( rs4613763 ) ., Consistent with published results 6 , 10 we found a significant association for IL23R ( p = 1 . 2 × 10−3; odds ratio: 2 . 51 ) but not for ATG16L1 ( p = 0 . 78 ) ., There was no effect of the novel 5p13 . 1 locus on UC ( p = 0 . 54 ) ., While additional studies will be needed to exclude completely a role in UC , these results suggest that the principal susceptibility effects of the 5p13 . 1 locus are for CD ., The restriction to CD risk observed for ATG16L1 and the 5p13 . 1 locus is similar to that found for CARD15 3 ., We herein describe the localization of a novel major susceptibility locus for CD on 5p13 . 1 by GWA ., The region of strongest association coincides with a gene desert devoid of known protein-coding genes ., The observed effect may be mediated by as-yet unknown transcripts mapping within the region ., As a matter of fact , limited numbers of spliced and unspliced ESTs originating from the HT1080 fibrosarcoma cell line or medulla ( e . g . , BG182136 and BG184600 ) map to the region ., An alternative explanation , however , is that the disease-associated region contains cis-acting elements controlling the expression of more distant genes ., We provide evidence in support of this hypothesis by demonstrating that genetic variants in the CD-associated region differentially regulate the expression levels of PTGER4 , the closest known gene located 270 kb proximally ., PTGER4 is a strong candidate gene for CD , as it is known that knock-out mice develop severe colitis upon dextran sodium sulphate treatment , unlike mice deficient in any of the seven other types of prostanoid receptors ., Increased susceptibility to colitis is also observed in wild-type mice administered an EP4-selective antagonist , while EP4-selective agonists are protective 14 ., We observe in particular that the CD susceptibility allele at marker rs4495224 is associated with increased PTGER4 transcript levels in lymphoblastoid cell lines ., This finding establishes a direct link between disease susceptibility and PTGER4 expression , although the direction of the effect apparently contradicts the results in knock-out mice ., Detailed studies of the effect of genetic variants in the disease-associated region on PTGER4 expression in different tissues and of a possible connection between PTGER4 levels and CD susceptibility are certainly needed and work towards that goal is in progress ., The hypothesis that the 5p13 . 1 CD-susceptibility locus operates by modulating PTGER4 expression levels could—at least in theory—be tested by replacing the corresponding murine sequences with the human orthologous variants and quantitatively complement the murine knock-out allele 15 ., Our results suggest that the 5p13 . 1 effect on CD could result from the combined action of multiple susceptibility variants ., Extensive sequencing of the most common haplotypes in the region of association is being conducted towards their identification ., Genotyping for the whole genome scan was performed on a Illumina HumanHap300 Genotyping Beadchip ( http://www . illumina . com ) 4 ., Genotyping of individual SNPs was performed on an ABI7900HT Sequence Detection System using TaqMan MGB probes from Pre-designed SNP Genotyping or Custom TaqMan SNP Genotyping assays ( Applied Biosystems , http://www . appliedbiosystems . com ) ., Association analyses were conducted using Fishers exact test ( whole genome scan ) or chi-squared tests of independence ( confirmation analysis ) ., We applied the logistic regression method of Setakis et al . 5 to test for the possible effect of population structure on the most significant association results ., The 110 control markers included in the logistic regression had 100% genotype success rate with minor allele frequency >30% , and no two markers were within 20 Mb of one another ., To test for an effect of block I conditional on the effect of an adjacent block II , we compared the proportion of I haplotype clades nested within a given II clade ( for instance , proportion of IA , IB and IC within IIA ) between cases and controls by chi-squared ., Chi-squared values ( and degrees of freedom ) were summed across II clades to yield an overall ( I|II ) test statistic ., The database genome-wide expression analysis data was provided by W . Cookson ( Imperial College , London , United Kingdom ) ., Briefly , expression data were generated from RNA extracted from EBV-transformed cells from 378 genotyped offspring in nuclear families ., Annotations for individual transcripts on the Affymetrix arrays were extracted from the Affymetrix NetAffx database ( http://www . affymetrix . com ) ., Data from the gene expression experiment was normalized together using the RMA ( Robust Multi-Array Average ) package 16 , 17 to remove any technical or spurious background variation ., An inverse normalization transformation step was also applied to each trait to avoid any outliers ., A variance components method was used to estimate heritability of each trait using Merlin-regress ( RandomSample option ) 18 , 19 ., For PTGER4 , we obtained a mean quantitative expression value of −0 . 017 and a variance of 0 . 722 while the heritability estimate for PTGER4 estimated using the sibship data was 0 . 844 ., Association analysis was applied with Merlin ( FASTASSOC option ) ., We estimated an additive effect for SNPs and tested its significance using a score test that adjusts for familiality and takes into account uncertainty in the inference of missing genotypes .

Introduction, Results/Discussion, Materials and Methods

To identify novel susceptibility loci for Crohn disease ( CD ) , we undertook a genome-wide association study with more than 300 , 000 SNPs characterized in 547 patients and 928 controls ., We found three chromosome regions that provided evidence of disease association with p-values between 10−6 and 10−9 ., Two of these ( IL23R on Chromosome 1 and CARD15 on Chromosome 16 ) correspond to genes previously reported to be associated with CD ., In addition , a 250-kb region of Chromosome 5p13 . 1 was found to contain multiple markers with strongly suggestive evidence of disease association ( including four markers with p < 10−7 ) ., We replicated the results for 5p13 . 1 by studying 1 , 266 additional CD patients , 559 additional controls , and 428 trios ., Significant evidence of association ( p < 4 × 10−4 ) was found in case/control comparisons with the replication data , while associated alleles were over-transmitted to affected offspring ( p < 0 . 05 ) , thus confirming that the 5p13 . 1 locus contributes to CD susceptibility ., The CD-associated 250-kb region was saturated with 111 SNP markers ., Haplotype analysis supports a complex locus architecture with multiple variants contributing to disease susceptibility ., The novel 5p13 . 1 CD locus is contained within a 1 . 25-Mb gene desert ., We present evidence that disease-associated alleles correlate with quantitative expression levels of the prostaglandin receptor EP4 , PTGER4 , the gene that resides closest to the associated region ., Our results identify a major new susceptibility locus for CD , and suggest that genetic variants associated with disease risk at this locus could modulate cis-acting regulatory elements of PTGER4 .

Individual susceptibility to many common diseases is determined by a combination of environmental and genetic factors ., Identifying these genetic risk factors is one of the most important objectives of modern medical genetics , as it paves the way towards personalized medicine and drug target identification ., Recent advances in SNP genotyping technology allows systematic association scanning of the entire genome for the detection of novel susceptibility loci ., We herein apply this approach to Crohn disease , which afflicts an estimated 0 . 15% of the people in the developed world and identify a novel susceptibility locus on Chromosome 5 ., A unique feature of the novel 5p13 . 1 locus is that it coincides with a 1 . 25-Mb gene desert ., We present evidence that genetic variants at this locus influence the expression levels of the closest gene , PTGER4 , located 270 kb away , in the direction of the centromere ., PTGER4 encodes the prostaglandin receptor EP4 and is a strong candidate susceptibility gene for Crohn disease as PTGER4 knock-out mice have increased susceptibility to colitis .

homo (human), genetics and genomics

journal.pgen.1003905

Robust Demographic Inference from Genomic and SNP Data

Reconstructing the past history of a given species is important not only for its own sake , but for disentangling demographic from selective effects 1 , 2 ., Demography is indeed often estimated on a set of markers and the best neutral model is used as a null for evidencing markers under selection 3 , 4 or for finding global patterns of selection across the genome e . g . 5 ., Various methods have been proposed to estimate demography from genetic data , including full-likelihood methods 6–9 , summary-statistics likelihood based methods 10 , 11 , or different flavours of Approximate Bayesian Computation 12–16 ., With some exceptions , these methods are relatively slow and do not scale up very well with new genomic data , as computation time increases with the number of loci ., In contrast , recently developed composite-likelihood methods based on the site frequency spectrum SFS , 17 have computing times that do not depend on the amount of available genomic data 18–21 , and several approaches have been proposed to estimate demographic parameters from the SFS e . g . 11 , 17 , 20 , 21–24 ., Among these latter methods , the most widely used is 21 , which estimates the expected joint site frequency spectrum for an arbitrary set of parameters by a diffusion approach ., Whereas the estimation of the expected SFS is relatively fast , the optimization of the parameters is still time-consuming , which prevents to tackle models with more than three populations at the same time ., While some methods can extract demographic information from single whole-genomes per population 25 , 26 , SFS-based methods , when applied to multiple individuals , do not require whole genome data because correct estimates of the SFS can be obtained from a few Mb 21 ., However , with few exceptions 11 , the accuracy of SFS-based methods has not been properly assessed , and their ability to infer demographic parameters has been questioned 27 ., One advantage of SFS-based inference methods is that they can handle large next generation sequencing ( NGS ) data sets 28–30 ., However , the computation of the SFS from NGS data is not always trivial ., An empirical Bayes approach has been proposed to estimate the joint 2D SFS from low coverage data 31 and an unbiased maximum likelihood approach has been developed to recover the SFS for a single population 32 ., SFS obtained from low-coverage genomic data often show a deficit of rare alleles because a given allele needs to be observed in several individuals to exclude read errors 28 , 33 ., These missing low frequency variants can lead to imprecisions and biases in population genetic inferences 34 ., Several approaches have been proposed to correct for this bias 32 , 35 , either during the process of genotype calling itself e . g . 31 , 36 , 37 or later by applying quality filters on called genotypes e . g . 38 ., Gravel et al . 28 have also proposed to predict the SFS from low-coverage data by using an overlapping subset of high quality data to derive a generalized correction of the SFS ., It appears likely that SFS estimation will improve with higher coverage NGS data , and that such data will become increasingly available and used in the near future ., As an alternative to deep sequencing , one could use information from a few tens of thousands SNP scattered over the whole genome to make demographic inference , but most SNP chips have complex and often unknown ascertainment schemes that bias the SFS if not properly taken into account 39–41 ., However , a new SNP chip has recently been introduced 42 , 43 , which implements a known and simple ascertainment scheme where SNPs are selected at random from sites that are heterozygous in a single individual of a given population ., Whereas this ascertainment scheme has no major effect on statistics designed to infer admixture 42 , it biases the site frequency spectrum 44 , 45 and thus potentially alters the estimation of other parameters ., Using simple combinatorics , the SFS can be unbiased 44 in a single population , and this strategy could be extended to unbias joint SFS under complex models involving more populations ., A diffusion approach has been recently proposed to estimate divergence times between two populations based on the fraction of SNPs having occurred recently in the ascertained population 45 , but this approach is currently restricted to the sole estimation of divergence time and cannot be applied if gene flow occurred between populations ., In this paper , we introduce a flexible and robust way to estimate demographic parameters from the SFS inferred from sequence or SNP chip data that we implemented in the fastsimcoal2 software ., Our method is based on Nielsens approach 17 , which estimates the expected SFS from simulations under any demographic model ., We compare the performance of this approach to 21 under a variety of evolutionary models with simulated data , and we show that it can successfully handle models including more than three populations ., We also show how this approach can be extended to deal with ascertained SNP panels by explicitly modelling the ascertainment bias and computing likelihoods based on expected ascertained SFSs ., We first apply our method to a large human genomic data set from which we estimate the demography of four populations , and then to two separate Affymetrix ascertained SNP panels 43 from which we estimate the demography of two African populations ., We performed parameter estimations for 10 data sets generated under each of the 3 evolutionary scenarios shown in Figures 1A–1C ., We took two approaches for estimating demography: our new approach based on a composite multinomial likelihood where the expected SFS is obtained using coalescent simulations and 21 , which computes a composite Poisson likelihood where the expected SFS is obtained by a diffusion approximation ., The two approaches have a very similar accuracy under a simple bottleneck scenario ( Figure S4 ) and under a scenario of population isolation with migration 46 ( IM model , Figure S5 ) ., For both approaches we report the estimates leading to the maximum likelihood obtained among 50 independent runs ., Under these conditions , leads to extremely accurate estimations for most data sets ., However , in a few cases ( 1/10 for the bottleneck scenario , and 2/10 for the IM model ) , the best likelihood obtained from 50 runs led to very divergent estimates , which were not reported in Figures S4 , S5 ., For those cases , the log likelihood appeared orders of magnitude smaller than those inferred for other data sets and could be easily spotted ., Although it is possible to recognize that additional runs are necessary to get meaningful estimates , we did not follow this procedure here , as we wanted to allocate similar resources to the two programs and get results using an automated procedure not requiring further user tweaks ., Contrastingly , fastsimcoal2 estimations seem to converge to correct values for all data sets in Figure S4 and S5 , even though the variances of the estimators are slightly larger than s for those cases where both approaches agree on the correct demographic model ., Parameter estimations under the more complex scenario of Figure 1C , mimicking a simple model of human evolution , are reported in Figure 2 ., In this case , results obtained by fastsimcoal2 are again very accurate and close to the true values for all 10 data sets ., With , we report results for only 8 data sets due to potential lack of convergence , as explained above ., However , even for these 8 data sets , the best estimates can be quite far from the true parameters , especially for parameters related to the ancestral bottleneck ., It suggests that for complex scenarios involving three populations and more than 5 parameters , needs to be run from many more than 50 initial conditions and that some iterative refinements of search ranges might be necessary to obtain correct solutions ( R . Gutenkunst , personal communication ) ., Note that a lack of robustness of under certain conditions ( e . g . high migration rates between populations ) had already been reported before 11 , 24 ., We have estimated parameters for the more complex hierarchical continent-island model shown in Figure 1D , involving samples from 10 different populations ( islands ) , a model that cannot handle ., Continent-island models are equivalent to infinite islands models , and have been used to model recent spatial expansions see e . g . 47 ., This model could therefore represent two successive spatial expansions , the first one stemming from an ancestral refuge area , and the second one starting more recently from a single deme belonging to the first expansion wave ., The parameters of interest are here the immigrations rates in each sampled deme , the timing of the spatial expansions and the ancestral population size ., As shown in Figure 3 , all these parameters are extremely well estimated by fastsimcoal2 when we maximize the multiple pairwise composite-likelihood shown in eq ., ( 7 ) ., We note that we can also recover very well the immigration rate to the unsampled deme ( rightmost column in Fig . 3 ) from which the second expansion started ., The accuracy of the immigration rate estimations is quite remarkable , given that they span over two orders of magnitude and that we specified the same search intervals covering four orders of magnitude for each parameter ., We first applied our methodology to the problem of estimating the past demography of two African , one European and one African-American populations ., The multidimensional SFS for these 4 populations was estimated from more than 220 , 000 non-coding SNPs , each located more than 5 Kb away from its closest neighbour , such as to minimize linkage disequilibrium between SNPs ., We examined three evolutionary scenarios shown in Figure 4 to explain observed patterns of diversity ., In the first and simplest scenario ( Figure 4A ) , the South Western African American population ( ASW ) was assumed to have been formed 16 generations ago ( around 1600 AD ) with initial input from one European ( CEU ) and two Niger-Congo speaking African populations ( Yoruba from Nigeria: YRI; Luhya from Kenya: LWK ) having diverged earlier ., In order to calibrate the other parameters , we assumed that the European population diverged from the ancestral African population 50 Ky ago 28 , 48 ., Under this scenario , we find that the ASW population would have initially received 16% ( CI95%\u200a=\u200a15–17% ) of its gene pool from the CEU population , 83 . 8% from the YRI population and almost nothing ( 0 . 2% ) from the LWK population ( see Table 1 , Model A ) ., This European contribution is in line with previous estimates obtained from SNP-chip allele frequencies ( 17% for Southwest African Americans 49 ) ., Under model A , the two Niger-Congo populations would have diverged very recently ( 70 generations ago , CI95%\u200a=\u200a56–197 ) , and the CEU and YRI populations have the smallest effective population sizes ( around 4000 individuals ) , whereas the ASW population has the largest ( NASW\u200a=\u200a170 , 000 individuals ) ., The inferred human ancestral population size is relatively small ( about 8000 individuals ) and there is no real signal of an ancestral bottleneck since the estimated bottleneck size ( NBOT\u200a=\u200a7083 ) is only 12% smaller than the ancestral size , in line with recent results showing no evidence for a strong Pleistocene bottleneck in humans 50 ., Whereas model A captures some obvious features of the past demography of these populations ( see Table S1 ) , it seems relatively unrealistic for some other features ( i . e . a direct contribution of the CEU and YRI populations to ASW ) ., We therefore investigated a more realistic but more complex and parameter-rich model involving several other unsampled populations , as shown in Figure 4B ( see Material and Methods for a complete description of this model ) ., The multiple continent-island model B1 assumes that the ASW population was founded by migrants originating from a Niger-Congo and from a European metapopulations , from which the two Niger-Congo and the CEU populations currently receive migrants ., It also assumes that the Niger-Congo and the European metapopulations passed through a bottleneck when they diverged from an ancestral African population ., An even more complex scenario B2 includes a potential admixture of the Luhya population ( a Niger-Congo speaking population from Kenya ) with an unsampled ( potentially East-African ) population , which also diverged earlier ago from the ancestral African population ., The model parameters estimates and their confidence intervals obtained by a parametric bootstrap approach are listed in Table 1 ., The two models show overall very congruent values and overlapping 95% confidence intervals for their common parameters ., The agreement is especially good for the human ancestral size ( NANC\u200a=\u200a12–13 , 000 individuals ) , the ancestral African population size ( NAFR\u200a=\u200a25–27 , 000 ) , the continental European size ( NEUR\u200a=\u200a14 , 500–16 , 500 individuals ) , the European strong bottleneck intensity ( IBEUR\u200a= =\u200a0 . 42–0 . 43 , where is the bottleneck duration , and is the bottleneck size ) , the Niger-Congo milder bottleneck intensity ( INC\u200a=\u200a0 . 027–0 . 028 ) , the divergence time of the Niger-Congo metapopulation ( TNC\u200a=\u200a793–797 generations ) , the time to the shift to the ancestral human population size ( TBOT∼10 , 000 generations ) , and the European contribution to the ASW population ( aE\u200a=\u200a0 . 16–0 . 17 ) ., The other parameters show different point estimates but all have overlapping confidence intervals ., We have plotted the marginal SFS for each of the four populations in Figure S6 , to visualize the fit of the expected and observed SFS for each model ., Whereas the expected population specific marginal SFSs show some discrepancies with the observation for the four populations under model A , the fit is much better for model B1 , except for LWK , which still shows an underestimation of singletons and doubletons ., Model B2 , which allows for LWK admixture , leads to a much better fit for the LWK population , as shown by the cumulative distribution of differences between the expected and observed marginal SFS ( see 3rd row in Figure S6 ) ., Under this model B2 , we estimate the LWK population to have 17% admixture from an unspecified but probably East African ( see e . g . Figure 1 in ref . 51 ) population ., This East African population would have diverged from the ancestral African population more than 2200 generations ago ( 95% CI 1274–3586 ) , thus potentially before the out-of-Africa dispersal ., Even though the different models can be conveniently compared on the basis of their marginal SFSs , these 1D SFSs only capture a small fraction of the total ( multidimensional ) SFS ., Therefore the models are better compared on the basis of their likelihood ., This is formalized here by a model comparison procedure based on AIC 52 , revealing that the relative likelihood of models A and B1 are almost 0 as compared to that of model B2 ( see Table S2 ) ., We estimated the parameters of African past demographies shown in Figure 5 based on Yoruba and San samples for which we have independent SNP panels ( see Methods section ) ., In model A ( shown in Figure 5A ) , we assumed that the Yoruba and San samples were taken from large populations that expanded after their divergence , and we allowed for a single pulse of gene flow between them at a given time Ta in the past ., The model B ( shown in Figure 5B ) includes the divergence of two-continent island metapopulations , and assume that the sampled populations are each an island attached to these continents and that the two continents exchanged migrants some time ago in a single pulse of gene flow , like in model A , but also earlier in time ( see Figure 5B and material and methods for a complete description of the model ) ., The point estimates of the two models and their associated 95% confidence intervals ( CI ) inferred from 100 parametric bootstraps are reported in Table 2 for both SNP panels ., Overall , the two SNP panels show congruent point estimators and CI widths under the two models ., There is only one parameter ( NAY ) for which the CI do not overlap under model A , which suggests that the two panels provide broadly compatible scenarios of African demography ., Estimations from data simulated under the same model for parameter values similar to those inferred in Figure 5A show ( see Figure S8 ) that, i ) both panels should perform very similarly for estimating parameters ,, ii ) all parameters of the model should be well estimated , except those related to a very recent expansion of one of the ascertained population ,, iii ) ancestral population sizes and divergence times are particularly well estimated , and, iv ) the addition of a single Denisovan sequence allows one to recover the absolute values of the parameters ., Concentrating on the parameters common to both models , we see in Table 2 that the ancestral size NANC shows very similar estimates across models and panels , with an estimated value around 9 , 000–9 , 500 individuals ( in line with estimates obtained with non-ascertained data set ) ., The African population size is also consistently estimated to be around 18 , 000–28 , 000 individuals across models , and the ancestral Yoruban size appears smaller and between 5 , 500 and 13 , 000 individuals ., These estimates fit well with previous Bayesian estimations of African demography from nuclear markers under slightly different models ., Based on microsatellites , Wegmann et al . 13 estimated the ancestral size of Niger-Congo ( NC ) populations ( to which Yoruba belong ) to be 12 , 500 individuals and that of the ancestral African population to be 15 , 000 individuals ., More recently , the analysis of 40 non-coding regions of 2 Kb 53 led to estimates of NC and African ancestral size to be 17 , 500 and 11 , 000 individuals , respectively , as well as a San effective size of the order of 20 , 000 individuals ., The differences between these estimations and ours might be due to the fact that these previous analyses were based on slightly different models that assumed constant sizes for all current populations and the same population size before the split with Denisovans ., In addition , we find evidence for some asymmetrical gene flow between San and Yoruba , around 500–600 generations ago ( 12 . 5–15 Kya ) under model A , and much more recently ( 60–80 generations ago ) under model B . Interestingly , this is the only parameter common to the two models that shows such drastic difference ., Despite this disparity , which could be due to the fact that we allow for earlier migration between the two metapopulations in model B , we obtain very similar estimates for the admixture rates between populations both between panels and across models ., Overall , we find a slightly larger extent of gene flow from Yoruba to San than the reverse , but the confidence intervals of the two parameters seem quite overlapping under both models ., Under model A , the point estimates for the divergence time TDIV are much more different than what was obtained under our simulations ( Figure S8 ) , with a much younger divergence suggested by the San panel ( 2 , 600 generations or 65 Kya ) than for the Yoruba panel ( 4 , 700 generations or 117 . 5 Kya ) ., Taking the middle of the overlap between the two CI would lead to a divergence time of 4 , 500 generations or 112 . 5 Kya ( Table 2 ) , in keeping with a recent estimate of the divergence of Khoisan populations obtained by an ABC approach 110 Ky , 53 , and compatible with the divergence time estimated between San and other West African population ( 65–120 Ky in 54 , or ∼100 Ky in 55 ) ., Under model B , the two estimates obtained for panel 4 and 5 , show a similar discrepancy , but the estimated values are much higher ( 5 , 530 and 10 , 330 generations for panels 4 and 5 , respectively ) , which can also be due to the fact that we authorize some gene flow between the two metapopulations after their divergence ., If we again take the middle of the overlap between the two CI , we obtain a value of 7 , 500 generations ( 180 Kya ) , substantially larger than the value obtained under model A ( 4 , 500 generations ) ., An examination of the parameters restricted to model B suggests that the Yoruban continent expanded recently 170–300 generations ago ( 4250–7500 ya ) , from a relatively small population of 600–3600 individuals , and that the Yoruban island receives more migrants ( around 18 per generation ) than the San island ( 2–3 individuals per generation ) ., The age of the expansion is slightly older than the divergence time between two Western Niger-Congo populations estimated previously ( 140 generations , 13 ) , and intermediate between the age of the Niger-Congo languages ( ∼10 Kya , 56 ) , and that of the Bantu expansion ( ∼5 Kya , 57 ) ., The larger immigration rate seen in Yorubans is compatible with the fact that farmer populations generally maintain higher levels of gene flow with their neighbours than hunter-gatherers due to their larger effective size 47 ., Note however that all parameter estimates mentioned above assume that the Denisova divergence time is correctly estimated at 16 , 000 generations or 400 Kya 58 , even though there is still a large uncertainty attached to this divergence time , which could range from 230 to 650 Kya 58 or even between 170 and 700 Kya in a more recent study 59 ., Reported estimates and CI in Table 2 do not take this uncertainty into account , and should thus be rescaled if a different divergence time between Denisovans and Humans was proposed ., Like in the case of non-ascertained data , we find that the more complex model is much better supported by the data ., Even though this better fit is barely visible when considering the marginal 1D expected SFS ( see Figure S10 ) , this is more exactly quantified by an AIC analysis ( Table S3 ) revealing that the relative likelihood of model A is close to zero for both panels when compared to model B ., We have introduced a new and flexible simulation-based approach to estimating demographic parameters ., For the tested scenarios , our composite-likelihood approach is as precise as 21 , which is the current standard in the field ., Our approach seems more robust than since it is more likely to converge towards the correct solution when starting from the same number ( 50 ) of initial conditions ( see Figures 2 , 3 , S4 , S5 ) ., In terms of computational speed , point estimates are very quickly obtained by for simple models ( on average 15 seconds and 6 minutes for models in Fig . 1A and 1B , respectively , compared to 15 minutes and 2h30 for fastsimcoal2 , respectively ) ., However , fastsimcoal2 is much faster for more complex models with three populations and migration ( 4–5 h per run for fastsimcoal2 for model on Fig . 1C , compared to 34 h on average for ) ., By maximizing the fit of two-dimensional SFS , fastsimcoal2 can also explore very complex models involving more than 10 populations with migration , which cannot be tackled by any other current method ., Since fastsimcoal2 and use a very similar likelihood function ( see Figure S3 ) , it seems that the improved convergence of our approach lies in the use of the ECM optimization scheme , which compensates for the use of non-optimal approximate likelihoods ., Note that our robust ECM maximization technique and the maximization of the product of pairwise composite likelihoods could also be used by methods deriving the SFS analytically or by a diffusion approximation ( like ) , thus potentially enabling the analysis of models as complex as those studied here ., Also note that recent progress in the computation of joint SFS using coalescent or diffusion approaches 18 , 23 have led to the development of promising demographic inference methods applied to the study of relatively complex evolutionary models see e . g . 24 ., Even though different demographic trajectories can lead to exactly the same SFS in a single population 27 , we do not find any evidence of parameter non-identifiability in our investigated cases ., This is probably because we restricted our search to a limited set of possible histories , defined by few-parameter models ., Our results confirm that if the true history lies within the models considered , the parameters of relatively complex scenarios can be well recovered from the ( joint ) SFS ., However , we must keep in mind that histories outside our model family might have identical likelihoods ., One disadvantage of our method ( and of any other simulation-based method ) is that we are approximating the likelihood , implying that two runs from identical initial parameter values can results in different estimations ( see Figure S2 ) ., Using more simulations for the estimation of the likelihood would lessen but not totally suppress this problem , but our results show that our maximization procedure leads to almost completely unbiased estimates and converges to correct values ., Another disadvantage of our approach is its dependence on composite likelihoods ., More powerful full likelihood approaches explicitly take into account linkage disequilibrium ( LD ) between sites 60 , and therefore might reveal useful to infer recent migration events ( see e . g . 61 ) ., That being said , our applied data sets consist of SNPs randomly distributed across the whole genome , and so patterns of LD between sites are minimal ., Whereas confidence intervals of demographic parameters based on composite likelihood ratios should in principle be too narrow ( see e . g . 21 , 60 , 62 , 63 ) , a study based on short stretches of DNA sequences has empirically shown that they were extremely similar to those obtained by explicitly modeling patterns of recombination 54 ., This appears unlikely to be true in general , and certainly not if products of pairwise composite likelihoods were used ( as with eq ., ( 7 ) , which was actually not used for our test cases ) ., Similarly , the use of composite likelihoods in model tests based on AIC can overestimate the support for the most likely model 64 ., However , the composite likelihoods in our test cases are quasi likelihoods due to the global independence between SNPs , and the differences in relative likelihood of alternative models are so huge ( see Tables S2 and S3 ) that some residual patterns of LD are unlikely to change our conclusions ., As an alternative to our composite likelihood maximization approach , Garrigan 22 has proposed to integrate an approximate likelihood computed in a way similar to ours into an MCMC algorithm , allowing him to get posterior distributions and credible intervals ., Whereas MCMC algorithms generally assume that the likelihood is computed accurately , it has been shown that MCMC procedure should lead to correct posterior distributions even if the likelihood is approximated , provided that there is no systematic error in its computation 65 , 66 ., This Bayesian approach could be worth exploring as a possible extension of our likelihood maximization procedure ., However , our current implementation has the advantage of quickly getting point estimates , around which CIs can be obtained later by repeating the estimation on bootstrapped samples ., For instance , a point estimate for the IM model shown in Figure 1B is obtained in about 2h30 on a single core machine , whereas 40–80 h are necessary to get posterior distributions for the parameters of a similar IM model from a single MCMC run using a specialized coalescent program on a multi-core machine see 22 ., The additional versatility of our simulation-based likelihood approach is well exemplified by its handling of ascertained SNP chips , and the inference of several parameters from the SFS under complex demographic scenarios ., Previous ways of handling ascertained SNP chips either consisted in removing the bias induced by the ascertainment 44 or taking it into account in the estimation procedure 39 , 45 ., However , these methods are usually not as general as our implementation , as they are either restricted to models including a single population 44 , or to the case of the sole estimation of divergence time between two populations 45 ., Contrastingly , our method can be applied to various types of demographic models including several populations , bottlenecks and migration ., Our simulation results suggest that parameters of complex models can be correctly recovered when the ascertainment consists of randomly chosen SNPs heterozygous in a single individual ( Figures S8 and S9 ) ., Interestingly , we find that some parameters of unascertained populations that diverged a long time ago either with ( Figure S8 ) or without ( Figure S9 ) admixture can also be quite well estimated when the model is well specified ., This suggests that a given ascertainment panel of the GWHO Affymetrix chip could be used to infer parameters in several related populations ., It is also worth noting that our calibration of parameters relied on the assumption that the divergence time with an outgroup population was known , but a different divergence time would only require a rescaling of the estimated parameters ., The use of an outgroup species with fixed divergence time is a standard way to calibrate mutation rates ( as e . g . in 21 ) , but we note it could also be used within species for DNA sequence data when some uncertainty exist on mutation rates , which is currently the case in humans 67 , 68 ., Most parameters inferred from real African populations have very similar estimates and confidence intervals irrespective of which SNP panel is used ( Figure 5 , Table 2 ) , which agrees with our simulation results ( Figures S8 , S9 ) ., However , a few parameters seem to provide relatively divergent estimates , like the Yoruba and the African ancestral size , as well as the Yoruba-San divergence time , a discrepancy that is not really expected from the simulations ., This discrepancy could stem from either an unknown source of ascertainment , from a misspecification of the model for one of the two ascertained population , or from an ascertained individual that is not representative of its population , the latter case being possibly due to inbreeding or admixture ., It currently appears difficult to disentangle these cases , and the inclusion of additional parameters in model B only seems to marginally improve the fit of the expected SFS to the data ., It suggests that our models still do not capture all aspect of the true demography of these populations , which might also affect our ability to reproduce the ascertained SFS , and have a negative impact on our estimations ., We note however that previous estimates of African demography e . g . 53 are more in line with those inferred from the Yoruba than from the San panel , which could suggest that our demographic models are more appropriate for the Yoruba than for the San population ., Overall , our results nevertheless show that meaningful demographic estimates can be obtained from ascertained SNP chips , suggesting a useful and cheap alternative to large scale sequencing for demographic inference ., Our methodology has the potential to infer demographic parameters from large scale genomic data under a much wider range of neutral evolutionary models than either the current implementation of , current Approximate Bayesian Computation ( ABC ) implementations 69 , summary statistics based approaches 11 , or other existing likelihood-based methods 22 ., Whereas ABC has the potential to be applied to genomic data , it has rarely been done since it usually requires the simulations of data sets as large as those analysed , which is computationally very costly ., Our approach could thus be seen as a powerful likelihood-based alternative to the study of complex evolutionary mode

Introduction, Results, Discussion, Methods

We introduce a flexible and robust simulation-based framework to infer demographic parameters from the site frequency spectrum ( SFS ) computed on large genomic datasets ., We show that our composite-likelihood approach allows one to study evolutionary models of arbitrary complexity , which cannot be tackled by other current likelihood-based methods ., For simple scenarios , our approach compares favorably in terms of accuracy and speed with , the current reference in the field , while showing better convergence properties for complex models ., We first apply our methodology to non-coding genomic SNP data from four human populations ., To infer their demographic history , we compare neutral evolutionary models of increasing complexity , including unsampled populations ., We further show the versatility of our framework by extending it to the inference of demographic parameters from SNP chips with known ascertainment , such as that recently released by Affymetrix to study human origins ., Whereas previous ways of handling ascertained SNPs were either restricted to a single population or only allowed the inference of divergence time between a pair of populations , our framework can correctly infer parameters of more complex models including the divergence of several populations , bottlenecks and migration ., We apply this approach to the reconstruction of African demography using two distinct ascertained human SNP panels studied under two evolutionary models ., The two SNP panels lead to globally very similar estimates and confidence intervals , and suggest an ancient divergence ( >110 Ky ) between Yoruba and San populations ., Our methodology appears well suited to the study of complex scenarios from large genomic data sets .

We present a new likelihood-based method to infer the past demography of a set of populations from large genomic datasets ., Our method can be applied to arbitrarily complex models as the likelihood is estimated by coalescent simulations ., Under simple scenarios , our method behaves similarly to a widely used diffusion-based method while showing better convergence properties ., In addition , our approach can be applied to very complex models including as many as a dozen populations , and still retrieve parameters very accurately in a reasonable time ., We apply our approach to estimate the past demography of four human populations for which non-coding whole genome diversity is available , estimating the degree of European admixture of a southwest African American population and that of a Kenyan population with an unsampled East African population ., We also show the versatility of our framework by inferring the demographic history of African populations from SNP chip data with known ascertainment bias , and find a very old divergence time ( >110 Ky ) between Yorubas from Western Africa and Sans from Southern Africa .

journal.pcbi.0030108

A Mass Conserved Reaction–Diffusion System Captures Properties of Cell Polarity

Eukaryotic cells such as neutrophils and Dictyostelium cells respond to temporal and spatial gradients of extracellular signals with directional movements 1–6 ., This process , known as chemotaxis , is a fundamental cellular process 5 , 7–9 ., In a migrating cell , specific molecular events take place at the front and back edges 1 , 2 , 5 , 10 ., The spatially distinctive molecular accumulation inside cells is known as cell polarity ., The front–back polarity usually has one axis , and this uniqueness is an important property because a migrating cell with two fronts could not move effectively 11 ., Another behavior of the front–back polarity is higher sensitivity of the front to a gradient of extracellular signals 10 , 12 ., This would also be important because the direction of movement should be controlled at the front edge ., Many molecules that are involved in chemotaxis in mammalian cells have been identified 4 , 5 ., Some molecules , including phosphoinositide 3-kinase ( PI3K ) , phosphatidylinositol 3 , 4 , 5-triphosphate ( PIP3 ) , Cdc42 , Rac , and F-actin , are specifically localized at the front , whereas others , including phosphatase and tensin homologue deleted on Chromosome 10 ( PTEN ) and RhoA , are at the back of migrating cells 1 , 4 , 10 , 13–15 ., The Rho family of small GTP ( guanosine 5′-tri phosphate ) ases in particular play a central role in chemotaxis and in establishing cell polarity 15–17 ., However , the mechanism of generating spatial accumulation of the Rho GTPases in cell polarity has yet to be clarified ., Many mathematical models that account for gradient sensing and signal amplification in cell polarity have been proposed 12 ., The local excitation and global inhibition model has been proposed to explain spatial gradient sensing 6 , 18 ., Some models involve positive feedback loops for amplified accumulation of signaling molecules 19–23 ., A reaction–diffusion model that includes local self-enhancement and long-range antagonistic effects has been proposed for directional sensitivity 24 ., Most of the reported models of cell polarity , which involve the detailed parameters such as concentrations or rate constants , have been constructed with many parameters and equations ., Although these detailed models are at least partially successful in reproducing experimental observations in cell polarity , the theoretical essence underlying cell polarity has not been explicitly demonstrated; thus , a simple conceptual model that can be used for analytical study is needed to extract common principles in cell polarity ., Although the reported models consist of distinct molecular species or networks , it should be especially emphasized that many of them are able to exhibit similar behaviors of cell polarity regardless of their different frameworks ., This fact indicates that a common principle should underlie the models , and a conceptual model is suitable for extracting common principles in cell polarity ., Because the Rho small GTPases are key regulators for cell polarity 16 , 17 , we first developed a reaction–diffusion model of the Rho GTPases on the basis of an earlier model 25 to examine the spatial properties of the Rho GTPases ., We found that the interaction of the Rho GTPases per se can generate specific spatial accumulation of the Rho GTPases , and that our model shows important behaviors of cell polarity ., We also found that our model exhibits behaviors similar to the model by Narang and Subramanian 22 , 23 , which is based on the molecular networks that are different from ours ., This suggests that common principles should underlie both models ., We found that a mass conservation of components and diffusion-driven instability are commonly conserved in the Narang and Subramanian models and in our model ., Based on these common properties , we established conceptual models of a mass conserved reaction–diffusion system , and found that such properties can account for the critical behaviors of cell polarity ., These results strongly suggest that a mass conservation of components with diffusion-driven instability is one of the fundamental principles of cell polarity ., We developed a reaction–diffusion model of the Rho GTPases ( Rac , Cdc42 , RhoA ) on the basis of the earlier model of the Rho GTPase 25 , which explains the temporal behaviors , to examine whether the interaction of the Rho GTPases can generate the spatial behaviors in the cell polarity of migrating cells ., The Rho GTPases exhibit guanine nucleotide–binding activity and function as molecular switches , cycling between an inactive GDP ( guanosine 5′-bis phosphate ) -bound form and an active GTP-bound form ., The Rho GTPases in active forms are located in the plasma membrane , and those in inactive forms are in the cytosol ( Figure 1A ) 26 ., It is likely that molecules in the cytosol have larger diffusivity than those in the plasma membrane ., According to some studies , Cdc42 activates Rac 27–29 , and RhoA has mutual inhibitory interactions with Cdc42 and Rac 29–33 ., In addition , Rac plays a dominant role in a positive feedback loop , which involves PI3K , PIP3 , and F-actin 13 , 34–36 ., Based on these experimental findings , we developed a diagram of the Rho GTPases interaction ( Figure 1B ) ., We assume that molecules of Rac , Cdc42 , and RhoA are activated by guanine nucleotide exchange factors ( GEFs; kai ) and are inactivated by GTPase-activating proteins ( GAPs; kii ) , and that interactions between molecules ( kij ) are additive to GEFs or GAPs ., Some molecule–molecule interactions are stimulation dependent ., Activations of molecules by the stimulation ( ksi ) are also assumed to be additive to GEFs ., As in many previous models 18–23 , we describe the spatial kinetics of molecules by simple diffusion equations ., A recent study in which the diffusion coefficients of the Rho GTPases in the plasma membrane are determined 37 may support this assumption ., The model of the interaction of the Rho GTPases is as follows ( see also Materials and Methods ) :, where Rac , Cdc , and Rho with suffixes m and c denote the concentrations of Rac , Cdc42 , and RhoA in the active state ( membrane ) and inactive state ( cytosol ) , respectively ., The numerical suffixes represent the following: 1 , Rac; 2 , Cdc42; and 3 , RhoA ., Dmi and Dci denote the diffusion coefficients of molecules in the active state and inactive state , respectively ( Dmi < Dci ) ., The position-dependent parameter , S , denotes the intensity of stimulation ., Because the parameters have not been fully obtained experimentally , we set parameters to reproduce the behaviors of cell polarity ( Figure 2 ) , and further analyzed the generality of such behaviors in detail with conceptual models ( see below ) ., The behaviors in the Rho GTPases model , such as switch-like reversible accumulation , uniqueness of axis , and sensing of the stimulation gradient by the polarized peak , were similar to those observed in the models of Subramanian and Narang 22 , 23 ., Despite the differences of the molecular species and networks among the models , the similarity in behaviors among them raises the possibility that a common principle could underlie them ., Therefore , we examined whether common properties can be seen . These models belong to reaction–diffusion systems with a periodic boundary condition and exhibit switch-like response , implying that instability is important for accumulation of the components ., In addition , these models involve components whose masses are conserved ., The total amount of phosphoinositides between the plasma membrane and the endoplasmic reticulum is conserved in Narang and Subramanians model 22 , 23 , and the total amounts of the Rho GTPase between the membrane ( e . g . , Rhom ) and cytosol ( e . g . , Rhoc ) are conserved in our Rho GTPases model ., Based on these common properties , we derived a new concept ( i . e . , mass conserved reaction–diffusion system with diffusion-driven instability ) and hypothesized that this system is a fundamental principle of cell polarity ., We therefore developed a simple conceptual model with two components ( u and v ) , which belong to a mass conserved reaction–diffusion system with instability , and examined whether the model can cause the behaviors of cell polarity to emerge sufficiently ., where a1 and a2 are parameters of the model ., The position-dependent parameter , S , is intensity of stimulation ( see Materials and Methods ) ., The stability of the homogenous solution in this model depends on the value of S ( see Materials and Methods ) ., We found that the conceptual model exhibits behaviors similar to the Rho GTPases model , such as switch-like reversible accumulation ( Figure 3A–3C ) , uniqueness of axis ( Figure 3D–3F ) , and sensing of the stimulation gradient by the polarized peak ( Figure 3G–3I ) ., This finding indicates that the conceptual model retains the essential behaviors in the Rho GTPases model ., We further used this conceptual model to examine in detail two behaviors of cell polarity: uniqueness of axis and sensitivity of the polarized peak ., To better understand the results of the numerical simulations , we investigated the following model ( see Equations 1a–1c ) , by analytical approximations: where Du = αDv ., Here , a1 , a2 , and α are parameters of the models ., This model belongs to the mass conserved reaction–diffusion system , and is more advantageous for analytical examination ., The homogenous solution was unstable regardless of the values of parameters in this model ( see Materials and Methods ) , so this model did not show reversible accumulation ., However , the model still retained the important properties such as uniqueness of axis and localization of sensitivity , so we can use this model to analytically examine whether these behaviors can emerge from a mass conserved reaction–diffusion system with instability ., In the following sections , we show that: ( 1 ) the model has one-peak stationary states , regardless of the system size ( if not too small ) ; ( 2 ) multiple-peak stationary states are unstable; and ( 3 ) the polarized peak moves depending on the gradient of the parameter value and the sensitivity is localized ., Finally , ( 4 ) , we verified our analyses by comparing analytical results with the values obtained by numerical simulations ., In this study , we used a mathematical model to clarify the role of the interaction between the Rho GTPases in cell polarity and developed a conceptual model of cell polarity to glean a theoretical understanding of the unique behaviors of cell polarity ., The Rho GTPases regulate the remodeling of the actin cytoskeleton via actin polymerization , depolymerization , and myosin activity 15 , 17 , 26 , which ultimately establishes cell polarity ., Then , what regulates the spatial activity of the Rho GTPases ?, The model proposed by Sakumura et al . indicates that the interaction of the Rho GTPases can regulate their own temporal activities 25 ., We demonstrated that the interaction of the Rho GTPases can regulate their own spatial activities ., In reality , the interaction of the Rho GTPases can provide more complicated temporal and spatial regulation of their activities ., Further study , including the determination of kinetic parameters of the interaction , is necessary to develop a more realistic model of the Rho GTPases ., The activator–inhibitor model for pattern formation 38 and the local excitation and global inhibition model for directional sensing in chemotaxis 6 belong to conceptual models , rather than to detailed biological models ., Such conceptual models with reduction of equations and parameters make analysis simpler and clearer ., We identified a mass conserved reaction–diffusion system with instability as common properties between the cell polarity models ., The model belonging to this system can sufficiently reproduce the important behaviors of cell polarity , such as uniqueness of axis and localization of sensitivity , and enabled us to theoretically understand such behaviors , which are difficult to examine without the models ., When interleukin-8 , a chemoattractant , is applied simultaneously from two directions at a 45° angle , normal neutrophils choose one direction for migration instead of responding to both sources 11 ., Neutrophils with multiple leading edges are rarely observed under normal conditions 39 ., When HL60 cells are transfected with a dominant-negative Rho construct or treated with Rho-kinase inhibitors , many cells exhibit the multiple pseudopods , where , in some cells , protrusions gradually withdraw , leaving a single , prominent pseudopod 10 ., In addition , inhibitions of PI3Ks cause HL-60 cells to form multiple pseudopods , which are weak and transient 39 ., These results suggest the instability of multiple leading edges , which may make the front of a migrating cell single and stable ., Chemotactic cells must have only one front–back axis because multiple fronts would prevent fine migration ., Subramanian and Narang investigated the response of their model to two almost identical stimulations 23 and showed that only one of the two peaks that arise persists , which agrees with our results ., Here we show that uniqueness of axis emerges from instability of multiple peak solutions in a mass conserved reaction–diffusion system ( Figures 4C and 5 ) ., In neutrophils 40 , HL-60 cells 10 , and Dictyostelium cells 41 , the polarized cells respond to changes in direction of a gradient by performing U-turns rather than by simply reversing polarity ., In addition , polarized migrating cells move forward without responding to the chemoattractant source near their rears 12 ., These experimental findings indicate that the sensitivity to chemoattractants is localized at the leading edge of polarized cells ., The localized sensitivity focuses the activity of the actin cytoskeleton at the leading edge , resulting in faster movement toward a chemoattractant source 12 ., Few mathematical theories , however , have been proposed to explain the localization of sensitivity ., Here we show that localization of sensitivity depends on the specific localization of a sensing window at the polarized peak in a mass conserved reaction–diffusion system ( Figure 4D and 4E and Figure 6 ) ., It should be added that many other systems can also exhibit a localized sensing window ., Consider a molecule that satisfies the following conditions: ( 1 ) the molecule ( X ) has two states ( Xm and Xc ) ; ( 2 ) the total amount of X is conserved; and ( 3 ) the diffusion coefficient of Xc is larger than that of Xm ., Two states of this molecule can be treated as components of a mass conserved system ., Some kinds of small GTPases , such as those of the Rho family , have two forms , active and inactive forms; the Rho GTPases in the active forms are located in the membrane , and those in the inactive forms are in the cytosol 26 ., Some enzymes involved in the cell polarity of chemotactic cells , such as PI3K and PTEN , are also reported to show a relationship between their activity and membrane binding 42–46 ., Molecules in the cytosol may well diffuse faster than those in the plasma membrane ., Thus , these molecules can be considered as components of mass conserved systems ., Chemotactic cells , such as Dictyostelium cells and neutrophils , polarize within a few minutes ( 30 s to 3 min ) after they are exposed to chemoattractants 6 , 10 , 47 ., Because the time scale of cell polarity is likely to be much shorter than that of gene expression and protein synthesis , we can assume that the masses of molecules are constant during the polarization of chemotaxis ., We numerically and analytically show that multiple-peak solutions are unstable ., To facilitate an understanding of the physics of this instability , we attempt to give an intuitive physical explanation of the behavior of molecules in the case where there exist two peaks ( Figure 8 ) , just as in Figure 4C ., We simplify the situation as follows ., ( 1 ) There are two spaces ( each space has one peak ) ., ( 2 ) The molecules have two forms , u and v , which have small and large diffusivities , respectively ., ( 3 ) No molecule is generated or degraded in the spaces ., ( 4 ) The molecules move between spaces , mainly in v-form , depending on the concentration gradient of v-form molecules ., ( 5 ) The u-form molecules convert v-form molecules into u-form , and this positive feedback is so strong that infusion of molecules into the space causes a decrease in v-form molecules ., Here , consider that a few molecules move from one space ( S1 ) to the other ( S2 ) ., According to ( 5 ) , because of the nature of the positive feedback , the number of v-form molecules in S2 declines as the total number of molecules in S2 increases ., In turn , according to ( 4 ) , the declining of the number of v-form molecules in S2 successively facilitates further transfer of v-form molecules from S1 to S2 , resulting in the further increase of the total number of molecules in S2 ., This flux is never disrupted by the generation or degradation of molecules because of the mass conservation ., Therefore , two peaks in such a system are unstable ., The condition ( 5 ) seems critical for instability , and we analyzed such conditions mathematically elsewhere 48 ., One of the most extensively studied reaction–diffusion models is the Turing model , in which robust spatial patterns , such as stripes or spots , emerge via a diffusion-driven instability 38 , 49 ., An ordinary Turing pattern in 1-D space is stripes with an intrinsic scale length 50 ., Mass conserved models also generate multiple peaks from the homogenous state during the early phase , which is explained by diffusion-driven instability ., However , they exhibit characteristic transitions after the initial peaks arise: most peaks become smaller and eventually disappear , and only one peak persists ( Figure 4A and 4B ) ., Why is the behavior of the mass conserved system so different from that of ordinary Turing models ?, Consider a reaction–diffusion system with vast size ( L → ∞ ) and interval I x1 , x2 within the system , where x1 and x2 are arbitrary but far apart ., Can we predict what will happen to interval I ?, For an ordinary Turing model , the linearization analysis around the homogenous solution gives us sufficient information 50 ., The mass conserved system is more complex , however , because the behavior differs between the case where the components flow into interval I versus the case where they flow out , and we cannot predict which will occur ., This difference in predictability seems to be fundamentally linked to the different behavior and the specificity of the mass conserved system ., Mass conserved models have multiple stationary states that are spatially homogenous or periodic , including a one-peak state and multiple-peak states ., We show that the multiple-peak stationary states are unstable , resulting in a single stable peak ., In some reaction–diffusion systems , such as the activator–inhibitor model ( or substrate–depleted model ) , high diffusivity of inhibitor ( or substrate ) make multiple peaks unstable , resulting in a single stable peak 51 ., An intuitive explanation for this instability is that an inhibitor , which is generated at the peak , rapidly diffuses throughout the cell ., In the cell where the inhibitor can be generated and degraded , the spread of the inhibitor requires the large diffusivity to overwhelm the inhibitor degradation ., On the other hand , in a mass conserved reaction–diffusion system , where no molecule is generated or degraded , a peak takes up molecules from its surroundings to grow ., That is , the growing peak inhibits the system not by spread of inhibitor but by deprivation of molecules ., In this case , large diffusivity is not required because there is no competitor to overcome , such as generation of molecules , and the inhibition can eventually spread throughout the cell ., Indeed , Equation 16b clearly indicates that any Dv can make multiple peaks unstable ( μ > 0 ) , at least in Model II ., It may be counterintuitive that any mass conserved system finally exhibits a one-peak pattern ., For Model II , the final steady state was a one-peak solution regardless of the system size , even when L was infinitely large ., But for Model I , the final steady state had two peaks when we set L = 80 ( unpublished data ) ., Some mass conserved models probably have a maximum size to have a unique peak ., However , this maximum size is independent of the linearization analysis ., The conditions for the uniqueness of concentration peak will be elucidated in future analyses ., Because properties observed in simple models are expected to be conserved in more detailed models , we assume that movement of molecules follows a simple diffusion equation in our conceptual model ., Active transport systems , such as actomyosin- and microtubule-based active transports , regulate cell polarity in various cellular processes 52 ., Such active transports are likely to follow the formation of intracellular asymmetry , which takes place under the resting condition where the cell polarity is yet to be generated ., Under such conditions , the diffusion of the Rho small GTPases has been measured and shown to be approximated by an apparent simple diffusion , if viewed on the order of seconds or tens of seconds 37 ., Because our concern in this study is the earlier asymmetry formation rather than the completion of the polarity , which includes active transport , we here assumed the simple diffusion of the Rho GTPases ., However , we will readily incorporate the detailed mechanism of the transport system of the Rho–GTPases in a future model ., In this study , we focused on the stationary state , but not on the transient state , which involves adaptation in response to a transient signal 44 , 53–55 ., Such properties , as well as high dimensionality and multiple components , should be incorporated into a future model ., Although our model is rather simple , it shows the important properties of cell polarity such as switch-like reversible response , uniqueness of axis , and localization of sensitivity ., We further demonstrated that the instability of multiple-peak solutions and the specific localization of a sensing window at the polarized peak in a mass conserved reaction–diffusion system are responsible for uniqueness of axis and localization of sensitivity , respectively ., One remarkable feature of a mass conserved reaction–diffusion system compared with other models is that a mass conserved reaction–diffusion system does not require strict assumptions for diffusion coefficients , such as smaller and much larger diffusivities of excitatory and inhibitory molecules , respectively , in the local excitation and global inhibition models 6 , 18 , or an extraordinary large diffusivity of the inhibitor in the Gierer–Meinhardt model 51 ., Since the Rho GTPases , PI3K , or PTEN have thus far not been demonstrated to involve such ad hoc assumptions of diffusivity , a mass conserved reaction–diffusion system is more likely to explain cell polarity where these molecules are involved , and to be adapted to a wide range of cell polarity ., Taking into consideration that the Rho GTPases system satisfies conditions of a mass conserved reaction–diffusion system , it is likely that this system is one of the fundamental principles of cell polarity ., We considered a one-dimensional circular system with circumference L . The position is represented by x ( −L/2 ≤ x ≤ L/2 ) ., We applied the periodic boundary condition , which is used in some models that explain cell polarity 23 , 24 ., We used explicit difference methods to perform simulations ., The difference intervals for calculations are shown in the following text ., The parameter values were set as follows: L = 10 , Dmi = 0 . 04 , Dci = 3 ( i = 1 , 2 , 3 ) , ks1= 1 , ks2 = 1 , ks3 = 1 , ka1 = 0 . 2 , ka2 = 0 . 2 , ka3 = 0 . 2 , ki1 = 0 . 4 , ki2 = 0 . 2 , ki3 = 0 . 2 , k11 = 4 , k12 = 3 , k13 = 5 , k23 = 6 , k31 = 4 , and k32 = 2 ., The difference intervals for calculations were taken to be Δt = 0 . 01 and Δx = 0 . 33 ., We set Xm ( x ) = 0 . 3 , Xc ( x ) = 0 . 7 ( X = Rac , Cdc , Rho ) , unless specified ., The parameter values were set as follows: a1 = 2 . 5 , a2 = 0 . 7 , Du = 0 . 01 , Dv = 1 , and L = 10 ., The difference intervals for calculations were taken to be Δt = 0 . 01 and Δx = 0 . 2 ., We set u = 1 and v = 1 , unless specified ., The parameter values were set as follows: a1 = 0 . 5 , a2 = 2 . 2 , Du = 0 . 1 , Dv = 1 or 2 ,, N̄ = 2 ., The difference intervals for calculations were taken to be Δt = 0 . 005 and Δx = 0 . 2 ., Approximation of a one-peak solution ., The computation was performed by setting L = 10 and Dv = 1 and taking the initial state as u = 1 and v = 1 with small perturbation ( ±0 . 01 ) ., The final profile ( t = 200 ) is shown in the left panel of Figure 7A ., The solid line indicates the profile of N , and the dashed line indicates P . Instability of a two-peak solution ., We examined the instability of two-peak solutions ., First , we obtained a stable one-peak pattern in Model II ( Equation 3 ) by taking the size to be L/2 , where L = 20 , 30 , 40 , and taking the initial state as u = 1 and v = 1 with small perturbation ( ±0 . 01 ) ., Because we applied the periodic boundary condition to this system , we could set the center of the concentration peak at x = 0 by translation ., Next , by duplicating and coupling this profile ( L/2 ) , we obtained a new profile ( L ) with two peaks ., We used this profile ( L ) with small perturbations ( ±0 . 01 ) as the initial state of the following simulation ., All trials ( Dv = 1 , 2 and L = 20 , 30 , 40 ) showed instabilities of two-peak profiles , and we obtained the growth rate , μsml , from the change in peak height ., Movement of a polarized peak in response to the parameter gradient ., We examined the response to the parameter gradient ., First , we obtained a stable one-peak pattern in Model II ( Equation 3 ) by setting L = 10 and taking the initial state as u =1 and v = 1 ., We set the center of the concentration peak at x = 0 by translation ., Next , we substituted, ( x ) = a2{1 + ( ɛ/2 ) sin2π ( x/L ) } for a2 in Equation 3 ., All trials ( Dv = 1 , 2 and ɛ = 0 . 02 , 0 . 04 , 0 . 06 ) showed movement of the polarized peaks , and we obtained the velocities , vsml , from the results ., In the homogenous stationary state , the Jacobian matrix for the reaction terms is given by, where fu and fv denotes the partial derivatives of f by u and v , respectively , at a homogenous stationary state of Equations 1a and Equations 1b ., We obtained a condition for instability of the homogenous solution: ( Dufv − Dvfu ) / ( DuDv ) > 0 ., For example , the homogenous solution is stable with S = 0 . 2 in Model I , whereas unstable with S = 1 , under the following conditions: a1 = 2 . 5 , a2 = 0 . 7 , Du = 0 . 01 , Dv = 1 , u + v = 2 ., The homogenous solution in Model II is always unstable ., Through the stability analysis using J , the range of wave numbers ( kh ) that have positive eigenvalues is obtained as 0 < kh < ( Dufv − Dvfu ) / ( DuDv ) 1/2 , and the wave number that has the largest eigenvalue and grows most rapidly from the homogenous state ,, , is obtained as follows:, For Model I ( Equation 2 ) , we obtain, = 1 . 32 under the following conditions: a1 = 2 . 5 , a2 = 0 . 7 , Du = 0 . 01 , Dv = 1 , S = 1 , u + v = 2 ., Equation 6b has the same formulation as classical Newton mechanics ., We define V ( Ne ) as, and Equation 6b implies, where E is a constant value , corresponding to period and total mass of Ne ( x ) ., The period λ and the average mass, N̄ = ( 1/λ ), Ne ( x ) dx satisfy the following equations:, where Nmin and Nmax are minimum and maximum levels of Ne ( x ) , respectively , and are derived from V ( Nmin ) = V ( Nmax ) = E ( 0 < Nmin <, N̄ < Nmax ) ., Equation 26a and Equation 26b give the relationship among Pe , λ , and, N̄ , where, N̄ is straightforwardly derived from the initial condition of ( u , v ) ., For Model II ( Equation 3 ) ,, and E can range between E* < E < 0 for Ne ( x ) to be a periodic solution ., Here E* = −{ ( Dv − Du ) /DuDv (, ) /6, } ., As E becomes smaller ( E → E* ) , the period λ converges to λmin , which is the shortest wavelength in the periodic solutions ., As E becomes larger ( E → 0 ) , the period λ diverges ., The solution of Ne ( x ) at E = 0 corresponds to the separatrix of Equation 6b , indicating an infinite period ( λ → ∞ ) ., The explicit form of Ne ( x ) for E = 0 can be obtained by, which has the sole peak at x = xp and decays to zero as x → ±∞ ., For a sufficiently large system , Equation 28 is a good approximation of the solution for −L/2 < x < L/2 ., Equations 28 and 7 lead to Equations 8a and 8b , and 9a–9c ., If there is nontrivial ( nμ ( x ) , pμ ( x ) ) that satisfies Equations 11a and 11b for μ with a positive real part , the solution is unstable ., Note that ( nμ0 , pμ0 ) = ( n0sech2 ( bx ) , − n0Pe/N0 ) satisfies Equations 11a and 11b for μ = 0 under periodic boundary conditions ., Here n0 is an arbitrary factor , originated from the linearity of equations , and we can set n0 = 1 ., For μ with an absolute value near zero , we can obtain ( nμ , pμ ) by the expansion from ( nμ0 , pμ0 ) with regard to μ ., To do this , we take nμ = nμ0 + μnμ1 + … and pμ = pμ0 + μpμ1 + … ., In the first order of the expansion , ( nμ1 , pμ1 ) obeys the following equations:, Thus , pμ is immediately derived from Equation 29a as:, where C1 and C2 are integral constants ., We can obtain nμ by solving by solving Equation 29b with substitution of pμ1 ., C2 is determined by the mathematical condition that nμ1 should be orthogonal to nμ0 ., In practice , C2 has little influence on ( nμ , pu ) , and we set C2 = 0 in the analysis ., The linearized approximations of Equations 19a and Equations 19b are given as follows:, where hN ( z ) = ∂f* ( Ne ( z ) , Pe , a2 ) /∂N , hP ( z ) = ∂f* ( Ne ( z ) , Pe , a2 ) /∂P , ha ( z ) = ∂f* ( Ne ( z ) , Pe , a2 ) /∂, ., Because t is no longer a variable , we set t = 0 without loss of generality and replace z with x in the following analysis ., Equation 31a under the periodic boundary condition leads to pε as follows:, where C3 is an integral constant ., By substituting , Equation 32 into Equation 31b , we obtain the following:, where Gn ( x ) and Ga ( x ) are defined by:, By solving Equation 33 , we obtain nɛ:, where W1 ( x ) and W2 ( x ) are defined by, Considering the periodic boundary condition for nɛ ( x ) , we obtain the following equation for sufficiently large L ( solvable condition ) :, This leads to the velocity of the peak:, where Z is given as follows:

Introduction, Results, Discussion, Materials and Methods

Cell polarity is a general cellular process that can be seen in various cell types such as migrating neutrophils and Dictyostelium cells ., The Rho small GTP ( guanosine 5′-tri phosphate ) ases have been shown to regulate cell polarity; however , its mechanism of emergence has yet to be clarified ., We first developed a reaction–diffusion model of the Rho GTPases , which exhibits switch-like reversible response to a gradient of extracellular signals , exclusive accumulation of Cdc42 and Rac , or RhoA at the maximal or minimal intensity of the signal , respectively , and tracking of changes of a signal gradient by the polarized peak ., The previous cell polarity models proposed by Subramanian and Narang show similar behaviors to our Rho GTPase model , despite the difference in molecular networks ., This led us to compare these models , and we found that these models commonly share instability and a mass conservation of components ., Based on these common properties , we developed conceptual models of a mass conserved reaction–diffusion system with diffusion–driven instability ., These conceptual models retained similar behaviors of cell polarity in the Rho GTPase model ., Using these models , we numerically and analytically found that multiple polarized peaks are unstable , resulting in a single stable peak ( uniqueness of axis ) , and that sensitivity toward changes of a signal gradient is specifically restricted at the polarized peak ( localized sensitivity ) ., Although molecular networks may differ from one cell type to another , the behaviors of cell polarity in migrating cells seem similar , suggesting that there should be a fundamental principle ., Thus , we propose that a mass conserved reaction–diffusion system with diffusion-driven instability is one of such principles of cell polarity .

Eukaryotic cells such as neutrophils and Dictyostelium cells respond to temporal and spatial gradients of extracellular signals with directional movements ., In a migrating cell , specific molecular events take place at the front and back edges ., The spatially distinctive molecular accumulation inside cells is known as cell polarity ., Despite numerous experimental and theoretical studies , its mechanism of emergence has yet to be clarified ., We first developed a mathematical model of the Rho small GTP ( guanosine 5′-tri phosphate ) ases that cooperatively regulate cell polarity , and showed that the model generates specific spatial accumulation of the molecules ., Based on our Rho GTPases model and other models , we further established a conceptual model , a mass conserved reaction–diffusion system , and showed that diffusion-driven instability and a mass conservation of molecules that have active and inactive states are sufficient for polarity formation ., We numerically and analytically found that molecular accumulations at multiple sites are unstable , resulting in a single stable front–back axis , and that sensitivity toward changes of a signal gradient is specifically restricted at the front of a polarized cell ., We propose that a mass conserved reaction–diffusion system is one of the fundamental principles of cell polarity .

biophysics, cell biology, none, computational biology

journal.pcbi.1005261

Functionality and Robustness of Injured Connectomic Dynamics in C. elegans: Linking Behavioral Deficits to Neural Circuit Damage

Understanding networked and dynamic systems is of growing importance across the engineering , physical and biological sciences ., Such systems are often composed of a diverse set of dynamic elements whose connectivity are prescribed by sparse and/or dense connections that are local and/or long-range in nature ., Indeed , for many systems of interest , the diversity in connectivity and dynamics make it extremely challenging to characterize dynamics on a macroscopic network level ., Of great interest in biological settings is the fact that such complex networks often produce robust and low-dimensional functional responses to dynamic inputs ., Indeed , the structure of their large connectivity graph can determine how the system operates as a whole 1 , 2 ., Neuronal networks , in particular , may encode key behavioral responses with low-dimensional patterns of activity , or population codes , as they generate functionality 3–8 ., Unfortunately , all biological networks are susceptible to pathological and/or traumatic events that might compromise their performance ., In neuronal settings , this may be induced by neurodegenerative diseases 9–11 , concussions , traumatic brain injuries ( TBI ) 12–14 or aging ., In this work , we extend a computational model to investigate behavioral impairments in the nematode C . elegans when the underlying neuronal network is damaged ., Specifically , we consider how the low-dimensional population codes are compromised under the impact of an injury ., Characterizing the resulting cognitive and behavioral deficits is a critical step in understanding the role of network architecture in producing robust functionality ., A hallmark feature of damaged neuronal networks is the extensive presence of Focal Axonal Swellings ( FAS ) ., FAS has been implicated in cognitive deficits arising from TBI and a variety of leading neurological disorders and neurodegenerative diseases ., For instance , FAS is extensively observed in Alzheimer’s disease 10 , 11 , Creutzfeldt-Jakob’s disease 15 , HIV dementia 16 , Multiple Sclerosis 17 , 18 and Parkinson’s disease 19 ., Most concussions and traumatic brain injuries also lead to FAS or other morphological changes in axons 20–25 ., Such dramatic changes in axon geometry may disrupt axonal transport 26 , 27 , and can potentially hinder the information encoded in neural spike train activity 28–30 ., Injured axons thus provide an important diagnostic marker for the overwhelming variety of cognitive and behavioral deficits 9 , 28 , 31 , in animals and humans 23 , 32–34 ., The massive size of human neuronal networks and their complex activity patterns make it difficult to directly relate neuronal network damage to specific behavioral deficits ., C . elegans , in contrast , has only 302 neurons , and its stereotyped connectivity ( i . e . the worm’s “Connectome” ) is known 35 ., This relatively small neuronal network generates a limited and tractable set of functional behaviors ( see Table 1 of 36 ) , with much of its locomotion/crawling behavior approximately confined to five observable motor states related to forward and backward crawling , omega turns , head sweeps and brief pause states ., Furthermore , these behaviors are well described as a superposition of only a few principal component body-shape modes 37 ., The combination of a fully-resolved neuronal network and a tractable low-dimensional output space makes C . elegans an ideal model organism for studying the impact of network damage on behavioral deficits ., Indeed , it is the only such neuronal network model currently available allowing for such a direct translational study of network damage ( injury ) to behavioral responses ., More precisely , computational models of C . elegans nervous system dynamics for the full or partial connectome successfully generate motorneuron outputs that can be related to behavior 38 , allowing for interpretable outputs even without accounting for muscular , mechanical or environmental factors , e . g . 39 ., We consider the model in 39 , which applies a single-compartment membrane model to the full somatic connectome; neurons are approximated as passive linear units connected by linear gap junctions and nonlinear chemical synapses ., Synaptic activation depends sigmoidally upon pre-synaptic voltage in equilibrium , and approaches this equilibrium value linearly in time ., All neurons are approximated as identical , with order-of-magnitude parameter assignments , except for their connectivity data ., Fig 1 ( a ) demonstrates a simulation of the putative forward crawling behavior identified in 39 within this model of C . elegans neural dynamics along with its projection onto principal component body-shape modes 37 ., In this perspective , we understand forward crawling as corresponding to a limit cycle ( i . e . a closed periodic trajectory ) in the principal component space of simulated neural recordings ., Extending this framework to damaged networks as in Fig 1 ( c ) allow us to explore how axonal pathologies lead to impaired functionality and behavioral deficits ., Even in our idealized injury simulations , the network’s impaired activity displayed significant variability ., This highlights one of the most challenging aspects of the field: the need for effective metrics to distinguish different types of behavioral deficits ., We propose such a criterium by using techniques borrowed from statistical shape analysis to quantify distortions in the main features of dynamical activity ., This metric is shown to be related to the functional outcome of an injury ., We further apply classification trees to our results to relate functional deficits to specific patterns of FAS ., This leads to experimentally-testable predictions about the effects of neuronal network-damage to the crawling motion of C . elegans and potentially new avenues for clinical diagnostics ., Indeed , our studies show that network damage leads to a diversity of dynamical/behavioral deficits ., We investigate how network distributed FAS as illustrated in Fig 1, ( c ) may affect its ability to generate desired responses to an input ., Network features associated with behavioral outcomes are best understood in model organisms such as the C . elegans since it has a limited repertoire of functional responses that include forward and backward crawling , omega turns , head sweeps and brief pause states ., Our focus in these studies will be on the behavior of forward crawling since a variety of experimental ablation studies have identified key neurons associated this functionality ., For instance , stimulation of PLM neurons excites densely-connected interneurons , which in turn , activate motorneurons responsible for forward body motion 40 ., Experimentally , optogenetic stimulation of the PLM neurons directly induces a forward motion response 41 , 42 ., Details of the underlying neurocircuitry were found by a series of ablation studies , where the functional role of a neuron is evaluated by disconnecting it from the network and observing behavioral deficits 39 , 43 ., The coordinated body motion of a crawling worm is well documented in videos and its postural dynamics were revealed by principal component analysis to consist of only a few dominant modes 37 ., Specifically , the sinusoidal body-shape undulations which describe the worm’s forward motion is well-described by circular trajectories ( limit cycles ) on the phase-space of its first two principal components ., An analogous mathematical form is present in the collective motorneuron dynamics following PLM stimulation 39 ., This commonality suggests that observed behaviors do retain fundamental signatures of the underlying network dynamics ., We show such a trajectory for ( simulated ) motorneuron responses to PLM excitation in Fig 1 ( a ) ., This low-dimensional representation captures 99 . 3% of the total energy of the system , and can be artificially mapped to crawling body-shape modes ., Although this mapping is still far from a mechanistic description of the worm’s coordinated body movement , we believe it captures important aspects of the crawling behavior ., See the Methods section for details ., Importantly , functional deficits of the C . elegans dynamics are understood as excursions/perturbations from the ideal limit cycle trajectory ., Damaged networks will be shown to fail to produce the low-dimensional output codes necessary for generating the optimal forward crawling limit cycle ., The robustness of the dynamical signatures ( population codes ) associated with behavior are investigated in injured neuronal networks ., Our injury statistics and FAS models are drawn from state-of-the-art biophysical experiments and observations of the distribution and size of FAS ., Fig 2 shows prototypical FAS injuries from stretching 26 and TBI in the optic nerve of mice 25 ., Fig 2 ( d ) shows a histogram of the probability of injury and size of the FAS ., These are used in our computational model 39 ., In a simulated injury , we assign to each affected neuron an axonal swelling from the distribution in Fig 1, ( b ) ., Values are scaled by an ( overall ) injury intensity parameter μ , such that, 1 + μ ∝ E swollen axon area healthy axon area ( 1 ), Fig 1, ( c ) exemplifies different injury settings: μ = 0 reproduces the original ( uninjured ) network , and lower/higher values of μ correspond to mild/severe injuries ., The presence of axonal swellings ultimately distorts the forward-motion limit cycle dynamics ., Fig 3 shows dynamical anomalies for different connectome injuries ., Notice how they induce qualitatively different changes to the closed orbit regarding location , size and shape ., Fig 3, ( c ) reproduces the specific simulated ablations from 39 , leading again to different dynamical effects ., A much larger ensemble of simulations ( 1 , 447 randomly-chosen injuries , as well as the code necessary to generate more ) and their corresponding effects to fundamental low-dimensional structures are included in the Supporting Materials ., Increasing values of μ typically shrink and shift the limit cycles within the plane ., In all simulations , there was always a sufficiently high injury level in which, μ * = {injured\xa0limit\xa0cycle\xa0collapses\xa0into\xa0a\xa0stable\xa0fixed\xa0point } ( 2 ) This occurs for instance , in Fig 3, ( b ) when μ = 3 . 80 ., Recent blast injury studies on C . elegans show that many of the nematodes display temporary paralysis before recovering to crawling behaviors 45 ., We would suggest that during the peak of the FAS , the injury levels on many of the nematodes are above μ* , thus leading to a collapse of a limit cycle to a fixed point where no motion is possible , i . e . it is in a paralyzed state ., Despite their common statistical distribution , randomly drawn injuries induce qualitatively different changes in the shape of the limit cycle ., Additional distorted sets are shown in the rows of Fig 4 ( along with 1 , 447 random-injury simulation sets in the Supporting Materials ) ., Thus , random injuries of equitable strength can lead to significantly different behavioral deficits ., Importantly , the deformation of the two-dimensional limit cycle can be used to characterize such functional differences ., To distinguish dynamical signatures of potentially different functional deficits , we evaluate the Procrustes Distance ( PD ) between healthy and injured limit cycles ., The PD is an important tool from statistical shape analysis to measure the similarity between two shapes after discounting effects due to translation , uniform scaling , or rotation ., Fig 4 depicts PD values for pairs of healthy/injured limit cycles as a function of injury level μ ., All curves are plotted until the injured limit cycle collapses into a fixed point ( μ = μ* ) , and the colored dots in the rightmost plots correspond to the same-colored limit cycles on the left plots ., Recent experimental work which induced mild TBI in C . elegans found that increasing the number of shock waves to which the worm was exposed reduced the worm’s average speed and , in many cases , led to temporary paralysis 45 ., The results of our simulations can be compared to these results: In Fig 5 we plot the location of the fixed points into which limit cycles collapse ( the “endpoints” , occurring at injury level μ = μ* ) ., We consider the following question: does the location of this endpoint ( and thus the behavioral outcome of the injury ) relate to the PD curve , and does it relate to the structure of the injury itself ?, Towards this end , we construct two simple classes of behavioral outcomes: endpoints which end in either the “upper” or “lower” part of the distribution ( for which we label the endpoints as red and green , respectively ) ., Panel, ( b ) of Fig 5 shows the average PD curve for the two classes ., They are qualitatively different: the average PD curve of “upper” endpoints is smoothly rising , whereas the average PD curve of “lower” endpoints has an extended declining region ., Shown also are the average scaling factor and translation distance of the distorted cycles ., Unlike the average PD curves , these change monotonically and are not distinct between classes ., This suggests that the shape of the PD curve carries information about the functional outcome of the injury ., We quantify this by fitting a classification tree to predict the endpoint class from the shape of the PD curve: this was found to predict endpoint class with a cross-validation error of 22 . 0% ., By comparison , randomly shuffling the labels leads to nearly double the cross-validation error , with an average of ( 44 . 6 ± 1 . 4 ) % ., Of even greater interest is any possible relationship between injury structure and behavioral output which could , given a specific pattern of distorted dynamics , make predictions about the class of neural injury ., To this end , we fit a classification tree to predict the endpoint class from the injury ., Fig 6 shows a classification tree which predicts endpoint class with a cross-validation error of only 14 . 6% ., This is much less than the error from a random class , suggesting that we can meaningfully relate the structure of a specific injury to a specific behavioral outcome ., Classification trees provide a highly interpretable and predictive method for making this connection , and make specific experimental predictions for the injuries corresponding to functional deficits ., The dynamic model for the C . elegans connectome simulates its neuronal responses to stimuli with a number of simplifications aimed at keeping the number of parameters at a minimum: we use a fairly standard and simple single-compartment membrane equation , and treat all neurons as identical save for their connectivity ., Many neurons in the network are nearly isopotential 46 , 47 , and it is a common and reasonable simplification to model neurons via single-compartment membrane equations , with membrane voltages as the state variables for each neuron ., Given this , Wicks et al . constructed a single-compartment membrane model for neuron dynamics 48 , which we later extended to incorporate connection data for the full somatic connectome 39 ., We assume that the membrane voltage dynamics of neuron i is governed by:, C V i ˙ = - G c ( V i - E c e l l ) - I i G a p ( V → ) - I i S y n ( V → ) + I i E x t ( 3 ) The parameter C represents the whole-cell membrane capacitance , Gc the membrane leakage conductance and Ecell the leakage potential of neuron i ., The external input current is given by I i E x t ., Note that this is , essentially , a fairly standard single-compartment membrane equation 49 , and its governing equations are formally identical to that used by Wicks et al . 48 except for our use of the full somatic connectome , our simplifying parameter assumptions , and minor differences in the treatment of synaptic dynamics taken from 50 ., In all simulations within this paper , we set I i E x t to be constant for the PLM neuron pair and zero for all other neurons ., This assures that densely connected interneurons will stimulate the motorneuron subcircuits responsible for forward crawling behavior ., Neural interaction via gap junctions and synapses are modeled by the input currents I i G a p ( V → ) ( gap ) and I i S y n ( V → ) ( synaptic ) ., Their equations are given by:, I i G a p = ∑ j G i j g ( V i - V j ) ( 4 ) I i S y n = ∑ j G i j s s j ( V i - E j ) ( 5 ) We treat gap junctions between neurons i and j as ohmic resistances with total conductivity G i j g ., We assume that I i S y n is also modulated by a synaptic activity variable si , which represents the conductivity of synapses from neuron i as a fraction of their maximum conductivity ., This is governed by:, s i ˙ = a r ϕ ( v i ; β , V t h ) ( 1 - s i ) - a d s i ( 6 ) Here ar and ad correspond to rise and decay time , and ϕ is the sigmoid function ϕ ( vi; β , Vth ) = 1/ ( 1 + exp ( −β ( Vi − Vth ) ) ) ., This form of sigmoidal activation is taken from 50 ., Note that it can be shown ( by setting s ˙ = 0 ) that , as in 48 , the equilibrium value of si depends sigmoidally upon Vi ., We keep all parameter values from 39 ( see Table 1 . The Connectome data , consisting of the number of gap junctions N i j g and number of synaptic connections N i j s , are taken from Varshney et al . 35 ( as available on WormAtlas 51 ) ., As in that study , we consider only the 279 somatic neurons which make synaptic connections ( excluding 20 pharyngeal neurons , and 3 neurons which make no synaptic connections ) ., Each individual synapse and gap junction is assigned an equal conductivity of g = 100pS ( such that G i j g = g · N i j g and G i j s = g · N i j s ) ., The values of cell membrane conductance and capacitance are affected by injuries , but in the uninjured case are set as equal for all neurons with values of Gc = 10pS and C = 1pF ., Note that in uninjured simulations , all neurons are modeled as identical except for their connectivity and the assignment of them as excitatory or inhibitory ( where Ej will have one of two values corresponding to these classes ) ., The model is valuable because it generates a low-dimensional neural proxy for behavioral responses ., Specifically , constant stimulation of the tail-touch mechanosensory pair PLM creates a two-mode oscillatory limit cycle in the forward motion motorneurons 39 ., This same dynamical signature was revealed in video analysis of the body-shape of the crawling worm 37 ., Thus the model is consistent with the observed biophysics ., Specifically , we calculate this plane by first simulating the forward-motion motorneuron membrane voltages ( class DB , VB , DD , VD ) in response to a PLM Input of IPLML , IPLMR = 2 × 104 Arb ., Units for the uninjured model ., We take time snapshots these membrane voltages V → M ( t ) , collect them into a matrix V , and take that matrix’s singular value decomposition ., That is:, V = V → M ( t 0 ) , V → M ( t 1 ) … = P · Σ · Q T ( 7 ), where P and Q are unitary and Σ is diagonal ., The columns of P are the principal orthogonal modes ., As in 39 , the first two of these modes ( the first two columns of P ) almost entirely capture the dynamics of the system within this subspace under constant PLM stimulation ., Projection of the full-system dynamics onto this plane consists of projecting the system’s motorneuron dynamics onto these modes ., Note that the single-compartment model which we employ ignores the spatial extent of neurons and specific location of each connection ., Our simplified injury model therefore must treat injury as a whole-cell effect ., Focal Axonal Swellings ( FAS ) increase the volume of an axon , which in turn , should alter the cell’s capacitance and leakage conductance within our model ., If we approximate a neuron by a single cable of length l and constant cross-section a , we may assume that the circuit parameters will scale with the axonal volume , i . e . ,, C ∝ a · l ( 8a ) G c ∝ a · l ( 8b ) When an axon swells , its healthy cross-sectional area aH will increase to some swollen value ai > aH ., Thus we assume that the healthy values for capacitance C and conductance Gc will also change according to, C i = C · ( a i / a H ) = C · ( 1 + μ · m i ) ( 9a ) G i c = G c · ( a i / a H ) = G c · ( 1 + μ · m i ) ( 9b ) We define the individual damage mi to neuron i as proportional to the relative excess area from swelling , i . e . , mi ∝ ( ai − aH ) /aH ., Values of mi are computed from the experimentally derived distributions in Fig 2 . Specifically , we construct FAS from the axonal swelling data of Wang et al . 25 , which used confocal microscopy to measure injury-induced swellings in the optic nerve of Thy1-YFP-16 mice ., Taken together , these define an “injury vector” m → , which we then normalize to | | m → | | 2 = 1 . After normalizing , the injury vector is then scaled by a global injury intensity defined as follows:, μ = ⟨ a i / a H ⟩ - 1 ⟨ m i ⟩ ( 10 ) Mild traumatic brain injuries yield small values of μ indicating that the average area of swollen axons is small ., Severe brain injuries yield high values of μ , indicating that large swellings are more common ., We leave the PLM pair of neurons receiving input uninjured ., All other neurons have their mi values assigned from the experimental statistical distributions ., The governing equation for an injured neuron is now, C V i ˙ = - G c ( V i - E c e l l ) - ( I i G a p ( V → ) + I i S y n ( V → ) ) / ( 1 + μ · m i ) ( 11 ) We can readily interpret the limiting cases: when μ ⋅ mi = 0 , the original governing equation is recovered , and thus μ = 0 corresponds to the healthy case ., When μ ⋅ mi is large , gap junction and synaptic currents have no effect ., The neuron’s voltage decays exponentially to its leakage potential , effectively isolating it from the network ., Note that our random assignment of swelling values neglects any spatial structure of the injury ., This could be easily modified by using a distribution which depends on the spatial location of the neuron ., Furthermore , this is a very simple model for neuronal swelling , in keeping with our simple model for neurons ., It necessarily neglects the actual geometry of swelling ., The use of a multi-compartment model would enable this in future studies ., Ultimately , there is currently limited biophysical evidence for making more sophisticated models ., As such , we have tried to capitalize on as many biophysical observations as possible so as to make a model that is consistent with many of the key experimental observations ., We use MATLAB ( version R2013a ) to solve the system of neuronal dynamical equations via Euler’s method , using a timestep of 10−4s ., We consider an ensemble of 1 , 447 different types of injury ( set of targeted neurons ) , for which the global intensity μ may vary from 0 ( uninjured ) to a critical value μ* ., When the intensity exceeds μ* ( found by a bisection algorithm ) , the limit cycle collapses to a fixed point ., To obtain intermediate values , we perform five simulations linearly spaced throughout ( 0 , 0 . 9μ* ) and ten additional simulations throughout ( 0 . 9μ* , μ* ) ., We classify the resulting injured trajectories as a Fixed Point or a Periodic Orbit according to the following criteria: Note that these criteria classify very small periodic orbits as fixed points , since their behaviors are very similar ., The method may also classify sufficiently slow , long-timescale oscillatory transients as periodic ., These tests ignore the first 5 seconds of simulation time ( 50 , 000 timesteps ) , chosen heuristically as a typical timescale of transient decay ., After this initial wait , we check the criteria at the end of each subsequent 5 seconds of simulation time until convergence is detected ., The results were not observed to be sensitive to the length of this interval ., Stephens et al . 37 found that the forward crawling motion of C . elegans is well described by two principal component body-shape modes called eigenworm modes ., When moving forward , the modes alternate within its phase space forming a limit cycle ., Kunert et al . 39 also found a two-dimensional limit cycle , but for the collective motorneuron activity after PLM stimulation ., They interpret this similar dynamical signature as a neuronal analog to the observed behavioral pattern ., To interpret the distorted neural activity caused by our simulated injuries , we construct a map from the neuronal activity plane onto the eigenworm plane ., The body-shape modes were extracted from Figure 2 ( c ) of 37 ., We first calculate the optimal linear mapping of the healthy trajectory onto a circle ( see Fig 3a ) ., We then use this calibration for all other trajectories ., This artificially translates anomalous neuronal responses to anomalous body motions ., Our procedure has a number of limitations , for which we list a few: The lack of direct neuronal analogs for injured network modes limits our ability to interpret arbitrary impaired behavioral responses ., Further computational work could also find neuronal proxies for additional behavioral modes so as to enable a more complete mapping ., Recent work on blast injuries of worms 45 could potentially help extend the analysis by providing injured eigenworm mode projections ., Procrustes Distance ( PD ) measures the dissimilarity between shapes , and in our context , we wish to compare the shape of the trajectories of the healthy neural responses ( circular orbits in the phase plane ) with the distorted ones produced after simulated injuries ., For that , we use the function procrustes . m from MATLAB’s Statistics and Machine Learning Toolbox ., We collect N data points from each trajectory and annotate their ( x , y ) coordinates in a ( 2 × N ) shape matrix S . The PD between two distinct shapes SA and SB is given by, P D = min b , R , c ∥ S B - b · S A · R + c → ∥ 2 ( 12 ) In other words , it finds the optimal ( 2D ) rotation matrix R , scaling factor b > 0 , and translation vector c → to minimize the sum of the squares of the distances between all points ., Intuitively , it compares shapes discounting translation , rotation , or scaling ., To calculate the PD curves as in Fig 4 , we use the uninjured ( μ = 0 ) limit cycle as our first shape SA ., The second shape SB is the limit cycle calculated for each injury at the indicated value of μ ., We pre-process the trajectories to extract data points only within a single period ., Since injuries usually distort the trajectory length , we use MATLAB’s spline . m function to interpolate them and collect the same number of data points ., Both limit cycles must also be phase-aligned , which we achieve by finding the phase that minimizes the Procrustes Distance ., We hypothesize that both the injury itself and the PD curves contain meaningful signatures of behavioral outcomes of a given injury ., For example , there is always a critical injury level μ = μ* in which the injured response collapses into a fixed point ., Our artificial map suggests that this endpoint location corresponds to the shape of a paralyzed worm ., We thus wish to relate endpoint location to ( 1 ) the shape of the PD curve , and to ( 2 ) the injury vector m → ., For these purposes , we classified the endpoints simply by dividing the endpoint distribution along its major axis ., Specifically , we take the distribution of endpoints in Fig 5 , calculate the leading principal orthogonal mode ( via taking the Singular Value Decomposition , as mentioned earlier ) , and classify the points by the value of their projection onto this mode ( where we arbitrarily classify projection values ≥ −0 . 01 as the “upper plane” and < −0 . 01 as the “lower” plane ) ., Given this definition , 63 . 2% of the points lie within the upper plane , and 36 . 8% lie in the lower plane ., Note that all of the forthcoming analysis could be equally well applied to any other output feature , and so we choose this classification for its relative simplicity ., We calculate the average PD curve within each class ., Since the PD curves may have a different number of points , we first pre-process them ., Specifically , we normalize the maximum μ and Procrustes Distance to 1 for all curves , and then interpolate them using MATLAB’s spline . m such that all curves have the same number of points ., We then simply take the average and standard deviation to obtain the average curves shown within Fig 5 ., This figure also plots the average scaling and translation curves as a function of injury level , for each class ., Scaling factors ( i . e . the factor by which the size of the distorted limit cycle has decreased from the original cycle ) are given as an output of MATLAB’s procrustes . m as used above ., Translation distance is found by calculating the location of the mean of each distorted cycle , and then calculating the distance by which this mean is displaced from the origin ., These curves are then normalized , interpolated and averaged , yielding the average curves in Fig 5 ., Note that , unlike the PD curves , translation and scaling are monotonic and not distinct between classes , and thus they do not carry the same information about the functional outcome of the injury ., We use the ClassificationTree class from MATLAB’s Statistics Toolbox ( version R2013a ) ., Fitting and cross-validation are performed using the included methods ClassificationTree . fit and kfoldLoss with default settings ( 10 folds ) ., The minimum leaf size was set by calculating cross-validation error over a range of minimum leaf sizes ( see Fig 6b ) ., For both PD curves and Injuries , cross-validation errors are optimal at a minimum leaf size of around 40 ., We use this minimum leaf size for all fits ., The classification tree that uses normalized PD Curve Shapes to predict the endpoint class yield a cross-validation error of 22 . 0% ., We can compare this to the random case ( i . e . the case where PD Curve Shape has no relationship to the class ) by repeating this analysis with randomly shuffled class labels ., For 100 trials with randomly-shuffled labels , the observed cross-validation error was 43 . 8 ± 1 . 4% ., Injury vectors were also used to fit classification trees for predicting endpoint classes ( see Fig 6 ) ., The cross-validation error of 14 . 6% was significantly lower in this case , while the randomly-shuffled labels analysis returned a error of 44 . 6 ± 1 . 3% ( consistent with the random error above ) ., In both cases we observe that the cross-validation error is far below what we would expect if the data had no relation to the classes ., Thus we can predict ( with cross-validated accuracy exceeding 85% ) the region into which the endpoint will fall given a specific injury ., Moreover , the classification tree in Fig 6 is very simple to interpret and depends on only three neurons: ALML , AVM and SDQL ., As per WormAtlas 51 , all three of these neurons have sensory functions ( ALML and AVM are mechanosensory; SDQL is an interneuron but is oxygen-sensing ) ., Simulated injuries distort dynamical signatures in the network’s activity , such as limit cycles ., Our Procrustes Distance metric quantifies how much the shape of the limit cycle is distorted , compared to the healthy cycle ., Our results indicate that as different injuries evolve , this metric follows qualitatively different trends ( as in Fig 4 ) ., In all trials , a sufficiently high injury level μ = μ* collapses the limit cycle into a stable fixed point ., The shape of the PD curve helps inform the location of this fixed point ( as in Fig 5 ) ., This suggests that the shape of the PD curve , as the injury evolves , may help predict the eventual behavioral outcome ( e . g . , the body shape the worm will assume during temporary paralysis ) ., Thus we have prescribed a method to monitor the dynamics of the injured worm and the implications of the injury as it evolves ., Finally , our classification trees divides neural injuries into two distinct classes of functio

Introduction, Results, Methods, Discussion

Using a model for the dynamics of the full somatic nervous system of the nematode C . elegans , we address how biological network architectures and their functionality are degraded in the presence of focal axonal swellings ( FAS ) arising from neurodegenerative disease and/or traumatic brain injury ., Using biophysically measured FAS distributions and swelling sizes , we are able to simulate the effects of injuries on the neural dynamics of C . elegans , showing how damaging the network degrades its low-dimensional dynamical responses ., We visualize these injured neural dynamics by mapping them onto the worm’s low-dimensional postures , i . e . eigenworm modes ., We show that a diversity of functional deficits arise from the same level of injury on a connectomic network ., Functional deficits are quantified using a statistical shape analysis , a procrustes analysis , for deformations of the limit cycles that characterize key behaviors such as forward crawling ., This procrustes metric carries information on the functional outcome of injuries in the model ., Furthermore , we apply classification trees to relate injury structure to the behavioral outcome ., This makes testable predictions for the structure of an injury given a defined functional deficit ., More critically , this study demonstrates the potential role of computational simulation studies in understanding how neuronal networks process biological signals , and how this processing is impacted by network injury .

Neurodegenerative diseases such as Alzheimer’s disease , Creutzfeldt-Jakob’s disease , HIV dementia , Multiple Sclerosis and Parkinson’s disease are leading causes of cognitive impairment and death worldwide ., Similarly , traumatic brain injury , the signature injury of the Iraq and Afghanistan wars , affects an estimated 57 million people ., All of these conditions are characterized by the presence of focal axonal swellings ( FAS ) throughout the brain ., On a network level , however , the effects of FAS remain unexplored ., With the emergence of models which simulate an organism’s full neuronal network , we are poised to address how neuronal network performance is degraded by FAS-related damage ., Using a model for the full-brain dynamics of the nematode Caenorhabditis elegans , we are able to explore the loss of network functionality as a function of increased neuronal swelling ., The relatively small neuronal network generates a limited and tractable set of functional behaviors , and we develop metrics which characterize how these behaviors are impaired by network injuries ., These metrics quantify the severity of TBI and/or neurodegenerative disease , and could potentially be used to construct diagnostic tools capable of identifying various cognitive deficits ., Additionally , we apply classification trees to our results to make predictions about the structure of an injury from specific cognitive deficits .

invertebrates, traumatic injury, medicine and health sciences, cognitive neurology, neural networks, engineering and technology, caenorhabditis, nervous system, neuroscience, animals, biomechanics, biological locomotion, animal models, decision analysis, caenorhabditis elegans, cognitive neuroscience, model organisms, management engineering, brain mapping, research and analysis methods, computer and information sciences, decision trees, animal cells, traumatic brain injury, cognitive impairment, critical care and emergency medicine, trauma medicine, connectomics, cellular neuroscience, neuroanatomy, cell biology, anatomy, physiology, neurons, neurology, nematoda, biology and life sciences, cellular types, crawling, cognitive science, organisms

journal.pntd.0003234

Impact of Schistosoma mansoni on Malaria Transmission in Sub-Saharan Africa

Malaria is highly endemic throughout sub-Saharan Africa in which 85% of global malaria cases and 90% of malaria deaths occur 1 ., Schistosoma mansoni ( the causative agent of intestinal schistosomiasis ) is likewise prevalent in many sub-Saharan African countries 2 , 3 , accounting for approximately one-third of the total cases of schistosomiasis in the region 4 ., Both malaria and intestinal schistosomiasis share similar epidemiological distributions and present challenges to public health and socio-economic development throughout these regions 5 ., Due to their coendemicity , there has been increased investigation into the interactive pathology between malaria and S . mansoni 6–9 ., Heavy S . mansoni infections have been found to be associated with a significant increase in the incidence of malaria among school-age children 6 ., While the mechanism responsible for the exacerbation of malaria in individuals infected with S . mansoni is not yet fully understood 7 , 9 , the interactions between the two diseases are possibly driven by countering effects the parasites have on immunological cytokines 10 , 11; that is , S . mansoni may alter the balance between Th1 and Th2 type immune responses 12–14 which reduces immunological control of malaria , although other mechanisms are possible ., Artemisinin-based combination therapies ( ACT ) are increasingly used as first-line treatment against malaria in sub-Saharan Africa 15 , 16 ., ACT is an efficacious drug regimen that reduces the risk of malaria-induced morbidity and mortality as well as malaria transmission from humans to vectors 17 , 18 ., For the control of schistosomiasis , current World Health Organization ( WHO ) guidelines recommend frequent mass administration of praziquantel , a highly effective and relatively inexpensive anti-schistosomal agent 19 , to school-age children or to entire communities depending on schistosomiasis prevalence and available resources 20 ., However , the adoption of mass praziquantel administration remains suboptimal in sub-Saharan Africa mainly due to limited drug availability , even as the schistosomiasis disease burden continues to rise 21 ., Mass praziquantel treatment coverage and compliance may vary substantially from one schistosomiasis endemic area to another ., We evaluated how S . mansoni infection and mass praziquantel administration may affect the dynamics and control of malaria in co-endemic communities ., In the absence of field studies that directly measure the effect of schistosomiasis control on malaria transmission and progression 6 , 22–25 , we address this question by using epidemiological and clinical findings that estimate the elevation in relative risk of malaria attributable to S . mansoni infection 6 to develop a mathematical model of the joint dynamics of malaria and S . mansoni among children ., We use this model to evaluate the inter-dependent impact of S . mansoni on malaria infection and the potential impact of schistosomiasis and malaria treatment for reducing malaria transmission ., We developed a mathematical model of the interplay between malaria and S . mansoni ., Malaria transmission was modeled as follows 26: At each point in time people can be in one of six infectious states – susceptible ( S ) , treated symptomatic disease ( T ) , untreated symptomatic disease ( D ) , asymptomatic patent infection ( A ) , asymptomatic sub-patent infection ( U ) and protected by a period of prophylaxis from treatment ( P ) ., We assumed that individuals entered the model susceptible and become infected at a rate determined by the force of infection in the population given by , where represents the biting rate on humans by a female mosquito , is the density of mosquitoes per human , is the probability of successful human inoculation upon an infectious bite , and the proportion of infectious mosquitoes in the vector population ., Upon infection , individuals either develop symptomatic disease ( with a probability Φ ) or develop patent asymptomatic infection ( 1−Φ ) ., Those who develop symptomatic disease have a fixed probability ( fT ) of being treated successfully with an ACT ( T ) , in which case they clear infection at a rate rT and enter a period of prophylactic protection ( P ) before returning ( rP ) to being susceptible to new infection ., Those who fail treatment ( 1−fT ) are assumed to eventually clear disease ( D ) and become patently asymptomatic ( A ) at rate rD ., From patent asymptomatic infection , individuals will move to a sub-patent stage ( U ) at a rate rA and then clear infection at rate rU and individuals return to being susceptible ., The force of infection of malaria on the mosquito population , , was given by the product of host biting rate per mosquito , probability of mosquito infection upon biting an infectious human , and the proportion of infected individuals at each infectious stage ( D , A , U ) ., The intensity of malaria transmission is represented as the annual entomological inoculation rate ( AEIR ) , defined as the product of the human biting rate of mosquitoes and the proportion of mosquitoes that are infectious ., AEIR is measured in the number of infective bites per person per year ( ibpy ) ., Here malaria prevalence refers to any level of parasitaemia rather than symptomatic disease alone ., For S . mansoni transmission , we assumed that at each point in time people can be in one of three states – susceptible ( S ) , infected with low egg output ( IL ) , and infected with high egg output ( IH ) 27 ., Likelihood of schistosomiasis transmission from humans to snails depends on worm burden and mean egg production per worm ., For the sake of simplicity , egg production was not explicitly modeled ., However consistent with previous schistosomiasis modeling studies , we used transmission rates that implicitly account for egg production rate per worm , contact with infested waters , and probability of worm establishment per contact 27 , 28 ., We assumed that individuals entered the model susceptible and become infected with an initially low egg output at a transmission rate ., Individuals with low egg output may then transition to high egg output at a transmission rate , where determines the rate of transition to a high egg output from a low egg output relative to ., We assumed the individuals infected with low egg output infect susceptible snails at a transmission rate , and individuals with high egg output infect snails at a transmission rate , where is the relative increase of transmission rate to snails for high egg output individuals relative to low egg output individuals ., Because rates of schistosomiasis reinfection are very high in endemic areas , we assumed that there is no natural recovery for S . mansoni infected individuals , and that without treatment infected individuals with a high egg output will not transition to a low egg output ., We incorporated annual praziquantel treatment into the model by assuming that treatment has an efficacy of 70% 29 , 30 ., We assumed that upon treatment , 70% of individuals with low egg output will recover from infection , and 70% of individuals with high egg output will either recover from infection or have their egg output reduced to a low level , such that 40% of treated high egg output will recover from infection , while 30% will have their egg output reduced to a low level 29 , 30 ., We evaluated the potential impact of deworming through mass drug administration with praziquantel on malaria prevalence by considering different levels of treatment coverage ranging from 30–80% ., Individuals can be infected with malaria only , S . mansoni only , or dually infected with malaria and S . mansoni ., The model captures the epidemiological interaction between the two diseases in terms of S . mansoni enhancing susceptibility to malaria denoted as and parameterized from epidemiological field data 6 ( Table 1 ) ., We focused on communities in which malaria and S . mansoni are co-endemic , and considered variation in malaria transmission intensity by varying the AEIR from 1 ibpy to 500 ibpy 31 ., We present results obtained at endemic equilibrium ., A detailed description of the model is given in the Supplement Material and an overview of parameters and values used to generate the model outcomes are given in Table 1 ., By comparing malaria prevalence in the presence and absence of S . mansoni co-endemicity , we showed that the impact of schistosomiasis co-infection on increasing malaria prevalence was higher in areas of low malaria transmission than areas of high malaria transmission ( Figures 1 & S1 ) ., For example , disease interaction was shown to increase malaria prevalence by 3 . 0–4 . 5% for an AEIR of 10 ibpy and by 0 . 6–1 . 5% for an AEIR of 100 ibpy , depending on malaria treatment coverage , ranging from 30–90% ( Figure 1 ) ., The effect of S . mansoni co-infection on malaria prevalence plateaued from 100 ibpy upwards ( Figure 1 ) ., We also found that the interaction between malaria and S . mansoni may reduce the effectiveness of malaria treatment for decreasing malaria prevalence ( Figure 2 ) ., For an AEIR of 100 ibpy , S . mansoni co-infection was shown to decrease the proportional reduction of malaria prevalence due to treatment by 1 . 3% for 90% treatment coverage , 1% for 60% treatment coverage , and 0 . 5% for 30% treatment coverage ( Figure 2A ) ., For 90% malaria treatment coverage , S . mansoni co-infection increases symptomatic malaria episodes by 29 episodes per 100 people annually , by 45 episodes per 100 people annually for 60% treatment coverage , and 93 episodes per 100 people annually for 30% treatment coverage ( Figure 2B ) ., For an AEIR of 10 ibpy , disease interaction was shown to decrease the proportional reduction of malaria prevalence due to treatment by 2 . 5% for 90% treatment coverage , 1 . 4% for 60% treatment coverage , and by 0 . 6% for 30% treatment coverage ( Figure 2A ) ., For 90% malaria treatment coverage S . mansoni co-infection increases symptomatic malaria episodes by 11 episodes per 100 people annually , by 16 episodes per 100 people annually for 60% treatment coverage , and 21 episodes per 100 people annually for 30% treatment coverage ( Figure 2B ) ., When ACT is used in combination with annual mass praziquantel administration , we showed that the intervention was more effective in reducing malaria prevalence and that this effectiveness increases both with the coverage of praziquantel and with the increased susceptibility to malaria infection due to S . mansoni ( Figure 3 ) ., This increase in effectiveness was more pronounced in areas of low malaria transmission intensity ( Figure 3A ) than in areas of high transmission intensity ( Figure 3B ) ., The interaction between S . mansoni and malaria generated an additional indirect benefit for mass praziquantel administration by reducing malaria prevalence ( Figure 3 ) ., We developed a co-epidemic model of malaria and S . mansoni transmission dynamics to take into account elevated susceptibility to malaria mediated by S . mansoni infection ., We used this model to investigate the potential effect of malaria-S ., mansoni interaction on the effectiveness of ACT and mass praziquantel administration for schistosomiasis for reducing malaria prevalence in co-endemic communities ., Our results suggested that co-infection with schistosomiasis in low malaria transmission settings increases malaria prevalence ., We further showed that in the absence of mass praziquantel administration , the interaction between S . mansoni and malaria may have contributed to reductions in population-level effectiveness of malaria treatment in areas of stable malaria transmission ., In regions of low malaria treatment coverage , co-infection with schistosomiasis led to the greatest increase in per person malaria episodes , independent of whether malaria transmission was high or low ., Our finding is consistent with epidemiological observations and laboratory studies that have suggested that presence of S . mansoni infections may affect the efficacy of malaria control measures , including a potential vaccine in co-endemic communities 9 , 32 ., The interaction between the two diseases may increase the health benefits of mass praziquantel administration by generating the additional indirect benefit of reducing malaria transmission in co-endemic communities ., Our results indicated that this benefit was particularly strong in low malaria transmission regions that experienced increased malaria susceptibility due to schistosomiasis co-infection ., Malaria is associated with a Th1 immune response 12 , while S . mansoni infection is associated with a Th2 response and had been demonstrated to impair immune responses to malaria 11 , 13 ., By reducing S . mansoni worm burden of infected individuals , praziquantel treatment may reduce the Th2 immune response associated with S . mansoni infection which may in turn result in a shift in the Th1/Th2 immune balance 14 , 33 towards the Th1 response that protects against malaria parasite ., Though our study focused on Plasmodium falciparum , our results may be applicable to other forms of malaria such as Plasmodium ovale and Plasmodium vivax , which may also interact with S . mansoni ., Prototype vaccines for both malaria 34 and S . mansoni intestinal schistosomiasis 35 are under development , such that the two vaccines could be co-formulated or combined 36 ., Our results suggest that a co-formulated or combined vaccine may be more efficacious in reducing malaria transmission in S . mansoni endemic communities than a vaccine targeting malaria alone ., In addition to increasing malaria incidence , clinical studies have shown that malaria–S ., mansoni co-infection may exacerbate clinical manifestations of both diseases 14 , 33 , 37 ., These additional impacts were not factored into our model , making our predictions of the effectiveness of joint programs of ACT and praziquantel conservative ., Our model also did not account for malaria age-dependent immunity 26 , 38 ., Age-dependent malaria immunity is less important among children than adults , however , and it is even less relevant in areas of low malaria transmission 26 , 38 , 39 ., We anticipate that accounting for age-dependent malaria immunity would only have a marginal quantitative effect on our results , such that the findings would remain qualitatively unchanged ., Malaria and S . mansoni may differ in their distribution of disease intensity , prevalence , and morbidity , with some portion of the population being at higher risk than others 33 ., Therefore , the magnitude of the interaction between malaria and S . mansoni on malaria transmission dynamics may vary from one risk group to another ., Given that data on risk group specific interaction between malaria and S . mansoni are not available , our model only accounted for elevated malaria susceptibility from S . mansoni high egg output ., Future studies could account for heterogeneity in malaria intensity and prevalence ., Currently , there is debate surrounding the extent and direction of the effects of malaria and co-infection with different helminth species on human hosts 7 , 33 , 40 ., Apparent contradictions arising from clinical and epidemiological studies may be resolved by the possibility of species-specific effects of helminth infections on malaria 7 , 40 , 41 ., As well as qualitatively different interactions for different worm burdens For example , Ascaris has been associated with protection from severe malaria complications 7 ., Conversely , epidemiological studies have suggested that hookworm elevates malaria prevalence 42 and exacerbates malaria-induced anaemia 22 , 43 ., Similarly , S . mansoni has been shown to be associated with increased malaria incidence 6 and exacerbation of hepatosplenomegaly 37 , 44 and anemia 45 in individuals co-infected with malaria ., It has also been reported that children with low ( but not high ) S . haematobium infection intensity co-infected with malaria have significantly lower P . falciparum parasitemia than worm-free individuals 46 ., This observation implies that the interaction between P . falciparum and S . haematobium may have contributed to lower malaria prevalence in S . haematobium low risk endemic communities , but that the reverse could be the case in S . haematobium high risk communities ., Additionally , malaria-schistosomiasis coinfection may have opposite effect on malaria transmission in S . haematobium compare to S . mansoni endemic communities ., Therefore , in S . mansoni - S . haematobium co-endemic communities 47 , 48 , schistosomiasis control may have a very complex impact on malaria transmission ., Further studies are needed on the interaction of S . mansoni and S . haematobium and their potential impact on malaria transmission ., Future transmission models on this topic could also account for worm mating probability and density dependent effects on egg output per worm , which can affect schistosomiasis transmission 49 , 50 ., Polyparasite helminth infections and malaria co-infection are widespread throughout sub-Saharan Africa 22 , 51 , 52 ., Therefore , studies investigating how co-infection affects the course of each infection , as well as immune responses , are fundamental to understand the potential additional benefits or perverse effects of mass drug administration and control programmes for tropical diseases ., There are myriad examples of parasitic co-endemicity and co-infections affecting health outcomes in sub-Saharan Africa ., For example malaria and hookworm co-infections 22 , 53 as well as and S . mansoni and hookworm co-infections 54 can lead to severe anemia ., A new modeling study on the interaction between lymphatic filariasis and malaria that takes into account increase in vector mortality due to lymphatic filariasis prevalence in mosquito and antagonistic Th1/Th2 immune response in co-infected host has shown that control strategies that reduce lymphatic filariasis transmission could potentially increase malaria prevalence 55 ., Similarly , some studies have indicated that antimalarial bednets may reduce transmission from lymphatic filariasis transmitted by anopheles mosquitoes 56 , 57 ., In addition , S . haematobium is interacting with HIV by increasing susceptibility to HIV infection through lesions and inflammation of genital track and immunomodulation effects 58 ., Two large studies in Zimbabwe and Tanzania found that women with genital schistosomiasis have a 3–4 fold increased odds of having HIV compared to women without genital schistosomiasis 59 , 60 ., Subsequent models have shown that female genital schistosomiasis ( caused by S . haematobium ) control strategies could reduce HIV transmission 61 , 62 , in co-endemic communities ., One of the limitations of our study was that we did not examine the relationship between S . haematobium and malaria ., Future studies could investigate the interaction between malaria and S . haematobium , as well as other helminths including hookworm ., Such studies could also investigate low risk schistosomiasis communities where , because of the low rate of schistosomiasis reinfection , the sequential order of infection between malaria and schistosomiasis may impact the co-infections of schistosomiasis on malaria transmission ., Clinical studies have shown that ACT used in combination with praziquantel may reduce both the malaria and the schistosomiasis health burden in co-infected individuals 63–66 , and that artemisinin-based therapy may have indirect benefits for reducing schistosomiasis health burden 63 ., Additional drug interaction studies may be required if ACT and praziquantel are combined for purposes of mass drug administration ., In an experimental rat model of clonorchiasis , combinations of praziquantel and artemisinins produced both synergistic and antagonistic effects depending on the doses administered 67 ., In humans infected with S . japonicum in China , it was noted that the combination of artemether and praziquantel chemotherapy did not improve treatment efficacy relative to praziquantel alone 68 , while in Africa ( Cote dIvoire ) the addition of mefloquine-artesunate did not increase the efficacy of praziquantel against S . haematobium infection 69 ., Integrating mass screening and treatment for malaria using ACT with mass drug administration of praziquantel could contribute to reducing both malaria and schistosomiaisis transmission in sub-Saharan Africa ., Therefore , future studies would investigate the complementary effects of ACTs and mass praziquantel administration for reducing both malaria and schistosomiasis transmission in co-endemic communities ., Immunological studies have suggested that praziquantel treatment in malaria-schistosomiasis co-endemic communities may alter the immune response of treated individuals , making them less susceptible to malaria infection 70 ., However , more studies are needed to confirm this impact of praziquantel treatment ., Our results suggest that in S . mansoni endemic areas , mass treatment of schistosomiasis may not only have a direct benefit of reducing schistosomiasis infection , it may also reduce malaria prevalence and disease burden ., This reduction of malaria prevalence was higher in areas of low malaria transmission intensity , but less pronounced in areas of high transmission intensity ( AEIR greater than 100 ibpy ) ., Additional epidemiological and clinical data on malaria–S ., mansoni co-infection to determine influence on immune responses and duration of malaria infection are needed to fully evaluate the potential effects of S . mansoni and schistosomiasis control strategies on malaria .

Introduction, Methods, Results, Discussion

Sub-Saharan Africa harbors the majority of the global burden of malaria and schistosomiasis infections ., The co-endemicity of these two tropical diseases has prompted investigation into the mechanisms of coinfection , particularly the competing immunological responses associated with each disease ., Epidemiological studies have shown that infection with Schistosoma mansoni is associated with a greater malaria incidence among school-age children ., We developed a co-epidemic model of malaria and S . mansoni transmission dynamics which takes into account key epidemiological interaction between the two diseases in terms of elevated malaria incidence among individuals with S . mansoni high egg output ., The model was parameterized for S . mansoni high-risk endemic communities , using epidemiological and clinical data of the interaction between S . mansoni and malaria among children in sub-Saharan Africa ., We evaluated the potential impact of the S . mansoni–malaria interaction and mass treatment of schistosomiasis on malaria prevalence in co-endemic communities ., Our results suggest that in the absence of mass drug administration of praziquantel , the interaction between S . mansoni and malaria may reduce the effectiveness of malaria treatment for curtailing malaria transmission , in S . mansoni high-risk endemic communities ., However , when malaria treatment is used in combination with praziquantel , mass praziquantel administration may increase the effectiveness of malaria control intervention strategy for reducing malaria prevalence in malaria- S . mansoni co-endemic communities ., Schistosomiasis treatment and control programmes in regions where S . mansoni and malaria are highly prevalent may have indirect benefits on reducing malaria transmission as a result of disease interactions ., In particular , mass praziquantel administration may not only have the direct benefit of reducing schistosomiasis infection , it may also reduce malaria transmission and disease burden .

Malaria and Schistosoma mansoni are co-endemic in many regions of sub-Saharan Africa ., Evidence from clinical and epidemiological studies support the hypothesis that concurrent infection with S . mansoni is associated with greater malaria incidence among school-age children ., We use mathematical modeling to evaluate the epidemiological impact of S . mansoni infection on malaria transmission in sub-Saharan Africa ., Using epidemiological data on the increased risk of malaria incidence in S . mansoni endemic communities from Senegal , we developed a co-epidemic model of malaria and S . mansoni transmission dynamics to address key epidemiological interactions between the two diseases ., Parameterizing our model for S . mansoni high-risk endemic communities , we show that the interaction between S . mansoni and malaria may reduce the effectiveness of malaria treatment for curtailing malaria transmission ., Moreover , we show that in addition to reducing schistosomiasis health burden , mass praziquantel administration will generate indirect benefit in terms of reducing malaria transmission and disease burden in S . mansoni–malaria co-endemic communities ., Our findings indicate the possible benefit of scaling up schistosomiasis control efforts in sub-Saharan Africa , and especially in areas were S . mansoni and malaria are highly prevalent .

infectious diseases, helminth infections, medicine and health sciences, schistosomiasis, population modeling, biology and life sciences, infectious disease modeling, computational biology, infectious disease control, malaria, parasitic diseases

journal.pcbi.1000868

A Bayesian Approach to the Evolution of Metabolic Networks on a Phylogeny

Biological networks are under continuous evolution and their evolution is one of the major areas of research today 1–6 ., The evolution of biological networks can be studied using various approaches such as maximum likelihood and parsimony 7 , 8 ., The maximum likelihood approach calculates the likelihood of evolution of one network into another by summing over all possible networks that can occur during the course of evolution under the given model ., Parsimony , on the other hand , assumes minimum evolution and only considers those networks that correspond to the minimum number of changes between the two networks ., However , the problem with these approaches is that enumeration of networks potentially occurring during evolution becomes impractical in the case of biological networks as the number of networks grows exponentially with the network size ., Recently , the evolution of biological networks has been studied using stochastic approaches where efficient sampling techniques makes the problem computationally tractable ., For example , Wiuf et al . 5 used importance sampling to approximate the likelihood and estimate parameters for the growth of protein networks under a duplicate attachment model ., Similarly , Ratmann et al . 6 used approximate Bayesian computation to summarize key features of protein networks ., The authors also approximated the posterior distribution of the model parameters for network growth using a Markov Chain Monte Carlo algorithm ., In this work , we focus on metabolic networks ., The evolution of metabolic networks is characterized by gain and loss of reactions ( or enzymes ) connecting two or more metabolites and can be described as a discrete space continuous time Markov process where at each step of the network evolution a reaction is either added or deleted until the desired network is obtained 9 ., To give a biologically relevant picture of evolution some reactions may be defined as core ( reactions that cannot be deleted during the course of evolution ) or prohibited ( reactions that cannot be added ) in the given networks ., The evolution of metabolic networks can then be studied using simple ( independent loss/gain of reactions ) or complex ( incorporating dependencies among reactions ) stochastic models of metabolic evolution ., We previously presented a neighbor-dependent model for the insertion and deletion of edges from a network where the rates with which reactions are added or removed from a network depend on the fraction of neighboring reactions present in the network 9 ., In this model , two reactions were considered to be neighbors if they shared at least one metabolite ., The model is summarized in Section ‘Neighbor-dependent model’ below ., The neighbor-dependent model depicts a biologically relevant picture of metabolic evolution by taking the network structure into account when calculating the rates of insertion and deletion of reactions from a network ., The model is , however , limited in the sense that it does not allow one to measure the strength of the neighborhood structure affecting network evolution ., Here , we present an extended model called the hybrid model that combines an independent edge model , where edges are gained or lost independently , and a neighbor-dependent model of network evolution 9 such that the rate of going from one network to another is a sum of the rates under the two models based on a parameter , which measures the probability of being in the neighbor dependent model ., This allows estimation of the neighborhood effect during metabolic evolution ., When modeling network evolution , we represent metabolic networks as directed hypergraphs 9–11 , where an edge called a hyperedge represents a reaction and may connect any number of vertices or metabolites ., Representing metabolic networks as hypergraphs not only captures the relationship between multiple metabolites involved in a reaction but also provides an intuitive approach to study evolution since loss or gain of reactions can be regarded as loss or gain of hyperedges ., We use the hybrid model to study the evolution of a set of metabolic networks connected over a phylogeny ., Previous attempts to study the evolution of metabolic networks in a phylogenetic context include Dandekar et al . 12 and Peregrin et al . 13 ., However , to our knowledge , the stochastic treatment of metabolic evolution over a phylogeny is an unexplored area ., Here , the phylogenetic relationship between the networks is established using sequence data since the metabolic annotations available for the majority of genome-sequenced organisms are generated using automated annotation tools based on the similarity of predicted genes to genes of known function and , therefore , contain a huge amount of noise ., In addition , we treat the branch lengths obtained using the sequence data as certain ., The advantage of fixing branch lengths is that the calculations do not require summing over all branch lengths for the given tree ., Calculating the likelihood over a phylogeny then requires a sum , over all possible networks that may have existed at the interior nodes of the tree , of the probabilities of each scenario of events ., This is similar to the idea introduced by Felsenstein 14 for observing DNA sequences over a phylogeny ., To sample the networks at internal nodes of the tree a Gibbs sampler 15 , 16 is presented that samples a network conditioned on its three neighbors , including a parent and two children networks , for given parameter values ., A Gibbs sampler for estimating the parameters of evolution that encases the Gibbs sampler for internal networks sampling is also presented ., The sampler estimates the evolution parameters without exploring the whole search space by iteratively sampling from the conditional distributions of the trees and parameters ., We demonstrate the Gibbs sampler by estimating and comparing the evolution parameters for the metabolic networks of bacteria belonging to the genus Pseudomonas ., The Gibbs sampler can also be used to infer the ancestral networks of a given phylogeny ., This is shown by inferring the metabolic networks of Pseudomonas spp ., ancestors ., In the neighbor-dependent for the evolution of metabolic networks 9 hyperedges are inserted or deleted from a network depending on the fraction of neighboring hyperedges present in the network ., Two hyperedges are considered as neighbors if they share a node ., The model assumes that the number of nodes in a network remains fixed and there is a set such that of hyperedges connecting these nodes ., The model also assumes the existence of a network called Reference Network which contains all these hyperedges ., If the hyperedges in the reference network are labeled 1 to then any given network can be represented as a sequence of 0s and 1s such that the -th entry in the sequence is 1 if and only if the hyperedge labeled is present in the network , and 0 otherwise ., Let the rate matrix describing the evolution under the neighbor-dependent model be denoted by ., An entry in this rate matrix corresponds to the rate of going from a network to a network , which differs from at position ., In the neighbor-dependent model , the rate of going from to depends on , and the neighboring hyperedges present in the network , and is given as follows: ( 1 ) where the function corresponds to the neighborhood component and is the appropriate entry from the rate matrix for the hyperedge ., The rate matrix is given as ( 2 ) where is the insertion rate and is the deletion rate ., The neighborhood component weights the insertion and deletion rates by the proportion of neighbors present in the network and is given as follows: ( 3 ) The denominator in Equation 3 gives the number of hyperedges present in the current network ., Although the neighbor-dependent model summarized above produces a biologically relevant behavior whereby highly connected reactions are toggled more frequently than the poorly connected counterparts , it does not allow one to determine the strength of the neighborhood structure effecting the evolution of metabolic networks ., To overcome this limitation , a parameter can be introduced in the model that corresponds to the neighborhood effect during the course of metabolic network evolution ., Consider two networks and which differ at position ., The hybrid model combines the independent edge model where edges are added or deleted independently , and the neighbor-dependent model summarized above such that the rate of going from to is the sum of the rates under the two models based on a parameter , which specifies the probability of being in the neighbor-dependent model ., The rate from to is given aswhere the term is the rate under the neighbor-dependent model given by Equation 1 and the term is the rate under the independent edge model corresponding to the appropriate entry from the rate matrix Q given by Equation, 2 . Substituting the value of from Equation 1 , the above equation can be simplified as follows ., ( 4 ) where the term corresponds to the neighborhood component given by Equation, 3 . It can been seen from ( 4 ) that the model behaves under the independent edge model when equals 0 and under the neighbor-dependent model described in the previous section when equals 1 ., For example , consider the toy network shown in Figure 1A ., The reference network containing all allowed hyperedges for this example system is also shown in the figure ., The system behavior for different values of is illustrated in Figure S1 for the toy network when simulated under the hybrid model along with the number of neighbors for each hyperedge ., The rates were calculated at each step using ( 4 ) ., An edge was then selected based on these rates and was inserted if absent from the current network and deleted otherwise ., As expected , hyperedges evolve independently when , resulting in similar insertion frequencies for all hyperedges and increasingly reflecting their neighborhood as the value of goes up to unity ., The fitness of the model is discussed in the Section ‘Fitness of the hybrid model’ below ., Biological networks are connected over a phylogenetic tree which is known through sequence analysis ., Calculating the likelihood over a phylogeny requires a sum , over all possible networks that may have existed at the interior nodes of the tree , of the probabilities of each scenario of events ., For example , Figure 1A shows an example system containing three networks , and with a phylogeny connecting the three networks shown in Figure 1B ., Let the phylogenetic tree be denoted by ., The likelihood of the tree is given as follows ., ( 5 ) Here denotes the parameters of the model , which is in the case of the neighbor-dependent model and in the case of the hybrid model ., is the marginal probability of observing the root and denotes the pairwise likelihood of evolving from the network to the network conditioned on in time for the given parameters ., In general , the likelihood of a tree with more than three networks can be calculated using the recursion described by Felsenstein 17 ., The likelihood at an internal node of the tree is given by the following recurrence relation ( 6 ) where and are left and right descendants of the node ., The likelihood of the complete tree is then given as ( 7 ) where is the marginal probability of observing the root and is given by Equation 6 ., Evaluating Equations 5 and 7 requires an algorithm to systematically and efficiently sample networks at the internal nodes of a tree and a method to calculate the pairwise likelihood of network evolution ., A Metropolis-Hastings algorithm to calculate the pairwise likelihood based on sampling paths between network pairs was described by Mithani et al . 9 , which calculates the likelihood by summing over paths between the given network pairs ., To sample networks at the internal node of a tree , a Markov chain can be constructed where states correspond to networks at the internal nodes ., The networks can then be sampled using a Gibbs sampler 15 , 16 as described in the next section ., Given a set of networks related by a phylogenetic tree , the networks at the internal nodes of the tree can be sampled using a Gibbs sampler ., The general idea is to sample each internal network by conditioning on its three neighbors ( one parent and two children ) ., This approach for sampling internal networks is similar to the one used by Holmes and Bruno 18 for DNA sequence alignment ., However , instead of using linear sequences , the sampler takes into account the network structure when calculating the new state ., The procedure is described below ., Consider a network with its three neighbors with branch lengths , ., The new network is selected as follows ., Example Consider the network in Figure 2 for which new state is to be calculated ., Denote the network by ., The three neighboring networks of the network are the networks , and labeled as , and respectively ., If denotes the neighborhood component for hyperedge then for the given rate parameters ( insertion ) and ( deletion ) , and the neighbor-dependence probability the rate matrix is written asFor simplicity , assume that ., The system then behaves under the neighbor-dependent model and the rate matrix simplifies toThe transition probability matrix of transforming to is then given asThe transition probability matrices and can be calculated in the similar fashion ., Once the transition probability matrices have been obtained , the sample for the new network can be drawn using Equation 8 ., For example , if the current configuration of the networks are taken as shown in Figure 2 , then the sample for the new state , for hyperedge 1 is drawn from the following distribution: The samples for hyperedges labeled 2 to 10 can be drawn in a similar fashion to obtain the new network ., The Gibbs sampler described above samples the internal networks on a phylogenetic tree for given parameter values ., This can be extended to estimate the parameters of evolution where equals ( ) in case of the neighbor-dependent model and ( ) in case of the hybrid model ., One way is to nest it within another Gibbs Sampler which iteratively samples internal networks and parameters from the distributions and respectively ., The general outline of the Gibbs sampler is as follows: The samples for parameters can be drawn using a Metropolis-Hastings algorithm 19 , 20 as described next ., Since the Metropolis-Hastings algorithm is a well-established method , it suffices here to give details about how a proposal for new parameters can be generated ., Readers interested in the general details of the algorithm are referred to Chapter 1 of Gilks et al . 21 ., The performance of the Gibbs sampler is discussed in Text S1 ., The Metropolis-Hastings procedure described above to sample parameters requires the likelihood of the tree when moving in the parameter space ., The likelihood can be calculated using Equation 5 which in turn requires calculation of the pairwise likelihood between network pairs ., The pairwise likelihood can be calculated using the Metropolis-Hastings algorithm described in Mithani et al . 9 which calculates the likelihood by summing over all paths between the given network pair ., However , for the Gibbs sampler described above in Section ‘Estimation of parameters’ this seems impractical since it will require running the Metropolis-Hastings sampler for all network pairs ., An alternate way is to use a pseudo-likelihood value when calculating the acceptance probability for parameters ., We calculate the pseudo-likelihood for a given network pair by dividing the network into smaller sub-networks and multiplying the pairwise likelihoods of the individual sub-networks ., Let denote the pseudo-likelihood from the network to the network in time for the given parameter values ., This is given aswhere is the pairwise likelihood of evolving sub-network into calculated by solving the exponential ., The procedure to obtain sub-networks containing at most hyperedges is outlined below ., An example is given in Figure S2 , which shows the sub-networks for the toy network shown in Figure 1 for different values of ., The above procedure was used to calculate the pseudo-likelihood of evolution of the toy network to the network ( Figure 1A ) for different subnetwork sizes , and the results were compared against the likelihood obtained by the MCMC approach described in Mithani et al . 9 and the true likelihood values obtained by evaluating ., All likelihood values were conditioned on the starting network ., The average CPU time taken by different approaches is shown in Figure 3 and the pseudo-likelihood values are listed in Table S1 ., The sub-network approach provides a reasonable approximation of the likelihood with a significant time advantage over the MCMC approach ., To see if the hybrid model fitted the metabolic network data better than the neighbor-dependent model , a likelihood ratio test was performed using the metabolic data for the bacteria belonging to the genus Pseudomonas ., The results show that the hybrid model fits the metabolic data better than the neighbor-dependent model ., For example , consider the metabolic networks in Pseudomonas fluorescens Pf0-1 ., The maximum likelihood estimates ( MLEs ) for the evolution of glycolysis/gluconeogenesis map 22 from Pseudomonas fluorescens Pf-5 to P . fluorescens Pf0-1 obtained using the Gibbs sampler described by Mithani et al . 9 were under the neighbor-dependent model and under the hybrid model ., Using the MLEs , the likelihood of observing the data under each model was calculated ., Assuming that evolution has been taking place for a long time , it is reasonable to use the equilibrium probability of a network to approximate the probability of observing the network ., The equilibrium probabilities were calculated using the procedure described by Mithani et al . 9 ., The maximum log likelihood obtained under the neighbor-dependent model equaled −76 . 53 whereas the maximum log likelihood obtained under the hybrid model equaled −63 . 47 ., The likelihood ratio test statistic was calculated as under degree of freedom ., The -value on 1 degree of freedom suggests that the hybrid model fits the data better than the neighbor-dependent model ., The MLEs , maximum log-likelihoods and the -values for different pathway maps in P . fluorescens Pf0-1 used in this analysis are listed in Table, 1 . The low -values for all the pathway maps suggest a better fit for the hybrid model compared to the neighbor-dependent model ., Likelihood ratio tests for other genome-sequenced Pseudomonas strains used in this analysiss showed similar results ( data not shown ) ., The fit of the data was further tested by comparing the degree distributions of the nodes obtained by simulating network evolution under the neighbor-dependent and hybrid models ., The MLEs for the evolution of networks obtained under the two models were used as the simulation parameters ., For example , when evolving the pathway maps in P . fluorescens Pf0-1 , the parameter values listed in Table 1 were used ., A total of 60 , 000 iterations were run with the first 10 , 000 iteration regarded as burn-in period ., Samples were collected every iteration and degree distributions were calculated ., The results for the six pathway maps used in this analysis are shown in Figure 4 for P . fluorescens Pf0-1 as an example which suggest a better fit for the hybrid model than the neighbor-dependent model ., Similar results ( data not shown ) were obtained for the other genome sequenced Pseudomonas strains used in this analysis ., To test the Gibbs sampler described in Section ‘Sampling internal nodes’ , the three network phylogeny shown in Figure 1 was used ., The networks were sampled at the internal nodes for different rate combinations with the neighbor-dependence probability kept constant at, 1 . The likelihood value was then calculated using Equation 5 by summing over the networks visited by the sampler at each internal node for each rate combination ., When calculating the likelihood over the phylogeny , the pairwise likelihood was calculated using matrix exponentiation ., A total of 25 , 000 iterations were run for each rate combination with the first 10 , 000 iterations regarded as burn-in period ., The exact likelihood of the phylogeny was also calculated by matrix exponentiation using all networks at each internal node ., The likelihood values estimated using the networks visited by the Gibbs sampler were comparable to those obtained by summing over all 1024 networks ., The true and estimated likelihood surfaces for a range of parameter values are shown in Figure S3 ., We also ran the Gibbs sampler for parameter estimation for the toy networks ., The sampler was run from a random starting value for 60 , 000 iterations with the first 10 , 000 iterations regarded as burn-in period ., The samples were collected every iteration to reduce computational overhead relating to storage as well as the correlation between samples ., A sample MCMC trace for the first 1 , 000 iterations of the sampler for the rate parameters is shown in Figure S4 ., The autocorrelation of parameters is plotted in Figure 5 suggesting an exponential decrease in the correlation as the lag between the samples increases ., To test the performance of the sampler , the likelihood of evolution for different rate combinations visited by the sampler was also calculated using Equation 5 by summing over networks visited by the sampler with ., As before , the pairwise likelihood was evaluated by calculating the exponential of the rate matrix ., The maximum likelihood averaged over three runs was found to be for parameters which is very close to the true likelihood obtained by matrix exponentiation ( Figure S3 ) ., To study the metabolic evolution in bacteria , we used the Gibbs sampler to estimate the evolution parameters for the metabolic networks of bacteria belonging to the genus Pseudomonas ., The diversity of pseudomonads , and the availability of genome-sequence data for multiple plant-associated Pseudomonas fluorescens , Pseudomonas mendocina , Pseudomonas putida , Pseudomonas stutzeri and Pseudomonas syringae strains , along with genome data for clinical isolates of Pseudomonas aeruginosa and for the insect pathogen Pseudomonas entomophila provide an excellent opportunity to use comparative genomic approaches to develop insight into the evolution of metabolic networks ., The phylogeny connecting the seventeen genome-sequenced strains of Pseudomonas is shown in Figure 6A ., The phylogeny was generated using multilocus sequencing analysis of conserved housekeeping genes ( gltA , gapA , rpoD , gyrB ) 23 ., The metabolic network data was extracted from the KEGG database 22 on January 2010 for pathway maps across the seventeen Pseudomonas strains shown in Figure 6A using the Rahnuma tool 24 ., The evolution parameters were also compared between two Pseudomonas species: P . fluorescens , a saprotroph that colonizes the soil environment , and P . syringae , a plant-pathogen that is found on leaf surfaces and in plant tissues ., The phylogenetic relationships between these species is shown in Figures 6B and C . The results are discussed here for the six pathway maps listed in Table 2 as they provide a representative set of different neighborhood characteristics observed across the Pseudomonas strains used in this analysis ., The basic information for each network across the seventeen Pseudomonas strains is given in Table S2 ., When estimating the parameters , the hyperedges corresponding to the reactions that were common to all seventeen Pseudomonas strains were defined as core edges and the hyperedges corresponding to the reactions not present in any of these seventeen species were defined as prohibited edges ., Three independent replicates of the sampler were run from random starting values for 60 , 000 iterations for P . fluorescens and P . syringae phylogenies , and 110 , 000 iterations for the phylogeny connecting the seventeen Pseudomonas strains with the first 10 , 000 iterations regarded as burn-in period in each case ., The samples were collected every iteration to calculate the posterior expectations and variances of the parameters ., These are listed in Table 2 and the ESS used for parameter estimation are listed in Table S3 ., The convergence of the algorithm was tested by checking the trace of the MCMC runs initiated from different starting values ., An example is shown in Figure S5 , which shows the trace for the sampler run on P . fluorescens phylogeny ( Figure 6B ) ., The running times and the acceptance percentages of the algorithm are listed in Table S4 for all three phylogenies ., We also calculated the number of insertion and deletion events for each reaction as well as at each branch of the Pseudomonas phylogeny for all six pathway maps ., These are shown in Figures S6 and S7 ., The high insertion to deletion ratio ( Table 2 ) for all three phylogenies for the glycolysis/gluconeogenesis map , pentose phosphate pathway map and pyruvate metabolism map , which are defined as a part of the carbohydrate metabolism of the bacteria in KEGG 22 and for the histidine metabolism map , which is a part of amino acid metabolism , suggests that very few reactions are missing from these networks in one or more Pseudomonas strains used in the analysis , resulting in a highly conserved network ., Lysine and phenylalanine pathway maps , on the other hand , have higher deletion rates compared to the insertion rates suggesting a variable reaction distribution across the Pseudomonas phylogeny and instability of these functionalities ., The results obtained in this study are consistent with the previous observation that the histidine metabolism map shows conservation of reactions across pseudomonads ( Mithani , Hein and Preston , submitted ) and that many Pseudomonas strains are able to use histidine as sole carbon and nitrogen source 25 whereas lysine and phenylalanine pathway maps have few conserved reactions across pseudomonads ( Mithani , Hein and Preston , submitted ) and are poor nutrient sources for these bacteria 25 ., The results also indicate that the pathway maps which are highly conserved across the seventeen Pseudomonas strains , i . e . glycolysis/gluconeogenesis map , pentose phosphate pathway map , pyruvate metabolism map and histidine metabolism map , also have higher neighbor dependence probabilities compared to the other two pathway maps , which have variable reaction distribution across the Pseudomonas phylogeny ., This might suggest a relationship between the neighborhood structure and the conservation of networks ., The comparison of the evolution parameters between P . fluorescens and P . syringae provides interesting insights into the evolution of the metabolic networks of these bacteria ., For example , the insertion and deletion rates are generally higher in P . fluorescens than those in P . syringae suggesting a higher number of insertion and deletion events in P . fluorescens networks compared to P . syringae networks ., This was expected since the evolutionary distance between the P . fluorescens strains is greater as compared to P . syringae strains ( Figure 6 ) allowing more time for the networks in P . fluorescens to evolve ., A higher deletion rate for lysine and phenylalanine pathway maps in P . syringae compared to P . fluorescens , however , suggests that P . syringae have had a higher number of deletion events than P . fluorescens during the course of evolution ., This supports the finding that P . syringae have gone through a high number of deletion events than expected based on the comparison between observed and expected distribution of reactions across the Pseudomonas phylogeny , and the identification of reactions that are uniquely present or absent from a single lineage ( Mithani , Hein and Preston , submitted ) ., In addition , a very low insertion to deletion ratio ( ) for lysine metabolism in P . syringae suggests a high number of deletion events in the lineage and consequently the loss of the ability of these bacteria to assimilate lysine ., This is in agreement with nutrient utilization assays , which have reported that bacteria belonging to the species P . syringae do not assimilate lysine as a nutrient source 25 ., Phenylalanine metabolism also has a higher deletion rate as compared to insertion rate in both P . fluorescens and P . syringae lineages ., This in conjunction with experimental data reporting the weak ability of these bacteria to utilize phenylalanine as a nutrient source might lead to a hypothesis that both P . fluorescens and P . syringae are drifting towards losing their ability to assimilate phenylalanine ., Overall , the results show that genome reduction is taking place in plant pathogenic bacteria belonging to the species P . syringae at a higher rate than their non-pathogenic counterparts in the species P . fluorescens ., The final aim of this study was to infer reactions present in the common ancestor of Pseudomonas spp ., and of individual species of Pseudomonas ., One way to address this is to predict that the common ancestor contained all the reactions that are common to existing Pseudomonas ., The variable reactions can then be assigned using a parsimonious approach which generates a conservative model of network evolution in which a minimum number of events occur ., However , the results above suggest that some lineages , particularly P . syringae , have undergone deletion events relative to the common ancestor and that some reactions absent in one or more modern pseudomonads might be present in the ancestral strain ., To take this into account , stochastic approaches such as the Gibbs sampler described in Section ‘Estimation of parameters’ can be used to sample ancestral networks from the posterior distribution of networks and the likelihood of reactions being present at various levels of the phylogeny can be calculated ., To demonstrate this , the Gibbs sampler was run on the pathway maps listed in Table, 2 . The Gibbs sampler was run with the same settings that were used for parameter estimation and samples for the networks at internal nodes of the Pseudomonas phylogeny ( Figure 6A ) were collected ., The degree distributions of nodes at the ancestral levels of the phylogeny are given in Figures S8 , S9 , S10 , S11 , S12 , S13 ., The likelihood of reactions being present at each level was obtained by calculating the proportion of times each hyperedge was present in the sampled networks ., The results are shown in Figures 7A–12A ., Only alterable reactions , that is the reactions which were neither defined as core nor were defined as prohibited in the networks , are shown ., The ancestral network reconstruction using the Gibbs sampler reported high likelihood values for reactions which are present in all the networks down a lineage and low likelihood values for reactions which show variable distributions across the Pseudomonas phylogeny ., For example , in the pentose phosphate pathway map ( Figure 8A ) , the reaction R01066 , which is present only in the three P . syringae strains , was assigned a very high likelihood of being present in the common ancestor of P . syringae pv ., phaseolicola 1448A and P . syringae pv ., syringae B728a as well as in the common ancestor for all the tree P . syringae strains but a very low likelihood of being present for all other internal networks ., In contrast , R06836 , which is present in sixteen out of the seventeen Pseudomonas strains ( absent in P . fluorescens Pf-5 ) , is reported to have high likelihood values of being present in all internal networks of the phylogeny ., Ancestral predictions were also generated under the parsimony model for these networks using the Fitch Algorithm 26 ., When assigning the reactions at the ancestral nodes the ties were resolved in favor of presence of reactions ., The results are shown in Figures 7B–12B ., Reactions for which parsimony failed to resolve ancestral predictions at the root are

Introduction, Methods, Results, Discussion

The availability of genomes of many closely related bacteria with diverse metabolic capabilities offers the possibility of tracing metabolic evolution on a phylogeny relating the genomes to understand the evolutionary processes and constraints that affect the evolution of metabolic networks ., Using simple ( independent loss/gain of reactions ) or complex ( incorporating dependencies among reactions ) stochastic models of metabolic evolution , it is possible to study how metabolic networks evolve over time ., Here , we describe a model that takes the reaction neighborhood into account when modeling metabolic evolution ., The model also allows estimation of the strength of the neighborhood effect during the course of evolution ., We present Gibbs samplers for sampling networks at the internal node of a phylogeny and for estimating the parameters of evolution over a phylogeny without exploring the whole search space by iteratively sampling from the conditional distributions of the internal networks and parameters ., The samplers are used to estimate the parameters of evolution of metabolic networks of bacteria in the genus Pseudomonas and to infer the metabolic networks of the ancestral pseudomonads ., The results suggest that pathway maps that are conserved across the Pseudomonas phylogeny have a stronger neighborhood structure than those which have a variable distribution of reactions across the phylogeny , and that some Pseudomonas lineages are going through genome reduction resulting in the loss of a number of reactions from their metabolic networks .

Metabolic networks correspond to one of the most complex cellular processes ., Most organisms have a common set of reactions as a part of their metabolic networks that relate to essential processes such as generation of energy and the synthesis of important biological molecules , which are required for their survival ., However , a large proportion of the reactions present in different organisms are specific to the needs of individual organisms ., The regions of metabolic networks corresponding to these non-essential reactions are under continuous evolution ., Using different models of evolution , we can ask important biological questions about the ways in which the metabolic networks of different organisms enable them to be well-adapted to the environments in which they live , and how these metabolic adaptations have evolved ., We use a stochastic approach to study the evolution of metabolic networks and show that evolutionary inferences can be made using the structure of these networks ., Our results indicate that plant pathogenic Pseudomonas are going through genome reduction resulting in the loss of metabolic functionalities ., We also show the potential of stochastic approaches to infer the networks present at ancestral levels of a given phylogeny compared to deterministic methods such as parsimony .

computational biology/evolutionary modeling, computational biology/metabolic networks, computational biology/systems biology

journal.ppat.0030049

M. tuberculosis Ser/Thr Protein Kinase D Phosphorylates an Anti-Anti–Sigma Factor Homolog

Mycobacterium tuberculosis ( Mtb ) is among the worlds most harmful pathogens , causing approximately two million deaths annually 1 ., In addition to the emergence of multi-drug–resistant strains , Mtb evades current therapeutics by shifting from active infection to a persistent , metabolically dormant state 2 ., This transition exemplifies the distinctive Mtb life cycle , which encompasses unique developmental adaptations to distinct environments 3 ., Little is known about the signaling mechanisms that mediate the biochemical changes that initiate and maintain the stages of Mtb development ., Candidate regulators of Mtb development include receptor Ser/Thr protein kinases ( STPKs ) that modulate intracellular events in response to external stimuli ., In eukaryotes , homologous STPKs sense environmental cues and transduce signals that regulate virtually all aspects of cell physiology ., The Mtb genome encodes 11 such Hanks-type ( also called “eukaryotic-like” ) STPKs , including nine putative transmembrane receptor kinases 4 ., Although the activating stimuli for these kinases have not been identified , the extracellular C-terminal sensor domains include a β-propeller interaction motif , a PASTA repeat thought to bind cell wall structures , and a redox-sensitive DsbG homolog 5–8 ., The intracellular , N-terminal kinase domains structurally resemble eukaryotic homologs , and similar receptor STPKs are widely distributed in bacterial genera ., The first reported bacterial STPK substrates include pThr-binding forkhead-associated ( FHA ) domains 9 , metabolic enzymes 10 , and apparent regulators of cell division 11 , 12 , but the mechanisms of signaling in vivo are not established ., Genetic studies suggest that two of the 11 Mtb STPKs are essential for growth 13 and that the STPKs regulate characteristics such as cell shape 11 , virulence 14 , and nitrogen balance 15 ., Identifying the intracellular targets of Mtb STPKs is essential to understanding their mechanistic roles in Mtb biology ., A second class of bacterial Ser/Thr kinases , the anti–sigma factors , regulates gene expression by controlling alternative sigma factors 16 ., Alternative sigma factors , such as sigma B ( SigB ) and sigma F ( SigF ) in Bacillus subtilis , mediate transcriptional responses to environmental cues by binding RNA polymerase and mediating promoter recognition ., Work on B . subtilis has established the paradigm by which complex regulatory cascades influence alternative sigma factor activity ( reviewed by Hughes and Mathee 16 ) ., Anti–sigma factor proteins ( e . g . , RsbW ) directly sequester cognate alternative sigma factors and prevent RNA polymerase binding ., Anti-anti–sigma factors ( e . g . , RsbV ) relieve this transcriptional repression by binding the anti–sigma factor ., The anti–sigma factors phosphorylate anti-anti–sigma factors on a conserved Ser or Thr , and this modification promotes dissociation of the complex ., This basic regulatory organization is recapitulated for multiple layers in which paralogs of anti–sigma factors and anti-anti–sigma factors switch partners and ultimately determine the concentration of the active sigma factor 17 ., In this “partner switching” mechanism , anti–sigma factor paralogs play two distinct roles ., Some anti–sigma factors ( e . g . , RsbW ) antagonize transcription by directly sequestering alternative sigma factors ., In contrast , other anti–sigma factors ( e . g . , RsbT ) act upstream to stimulate transcription by binding and activating the master “environmental sensing” phosphatase ( RsbU in B . subtilis 18 ) ., This phosphatase reactivates anti-anti–sigma factors , which bind the cognate anti–sigma factor , thus increasing the concentration of free sigma factor ., Environmental cues affect the phosphorylation state of upstream anti-anti–sigma factor paralogs ( such as RsbS and the RsbRA-D proteins ) , and these proteins form a complex ( termed the “stressosome” ) that binds the positive regulator of the phosphatase 19 ., The central role of Ser/Thr phosphorylation in anti-anti–sigma factor regulation and the established role of eukaryotic kinases in gene regulation led us to test the hypothesis that the eukaryotic-like STPKs may impinge on transcription regulated by alternative sigma factors ., Here , we demonstrate that increasing the activity of the PknD STPK in Mtb resulted in specific phosphorylation of a single anti-anti–sigma factor homolog , Rv0516c ., Simultaneously , the Rv0516c gene was activated and transcription of genes regulated by the SigF alternative sigma factor was coordinately altered ., PknD phosphorylated Rv0516c at a novel site , Thr2 , distinct from conserved Ser/Thr phosphorylation sites in the anti-anti–sigma factor family ., Thr2 phosphorylation abolished binding to another anti-anti–sigma factor ., These results demonstrate that PknD phosphorylates a putative sigma factor regulator in Mtb , alters binding of a cognate regulator , and , by a mechanism that has not been determined , changes the expression of SigF-dependent genes ., To investigate the pathways regulated by receptor STPK signaling , we constructed Mtb strains expressing either wild-type ( WT ) or kinase-dead ( Asp138Asn ) PknD under the control of an acetamide-inducible promoter 20 ., The Asp138Asn mutation in the catalytic site reduced the in vitro activity of kinase domain ∼2 , 600-fold ( Figure S1 ) ., In this approach , excess kinase substituted for an activating signal to stimulate downstream pathways ., Western blotting with anti-PknD and anti-pThr antibodies showed that the WT or mutant kinases accumulated after induction and produced a concomitant increase in Thr phosphorylation ( Figure 1A ) ., Expression of the attenuated Asp138Asn mutant produced a much smaller increase in phosphorylation of cellular targets ., Consistent with the idea that the expressed PknD ( directly or indirectly ) mediated the observed phosphorylation in vivo , cellular–protein phosphorylation was blocked when the PknD variants were induced in the presence of SP600125 ( Figure S1B ) , a c-Jun N-terminal kinase ( JNK ) inhibitor that shows specificity for PknD over other Mtb STPKs ( C . Mieczkowski and T . Alber , unpublished data ) ., Transcriptional profiling using microarrays confirmed the induction of PknD transcripts and revealed a set of genes regulated by PknD overexpression in a kinase-dependent manner ( Figure 1B; Table S1 ) ., Remarkably , the transcripts most differentially expressed in the strain expressing WT PknD ( Figure 1B ) included the genes with the largest reductions in transcription during log phase growth of an Mtb mutant harboring a deletion of sigF 21 ., Moreover , the Rv0516c gene , which is homologous to anti-anti–sigma factors , was dramatically induced by PknD overexpression ., The established role of phosphorylation in anti-anti–sigma factor regulation supported the hypothesis that PknD specifically phosphorylates Rv0516c ., To test this idea , we measured the phosphorylation by the purified PknD kinase domain of all predicted Mtb homologs of the B . subtilis alternative sigma factor regulators ., Potential regulators were identified using iterative PSI-BLAST searches for homologs of Rv0516c , SpoIIAA , and SpoIIAB , and putative homologs were confirmed using 3D-PSSM 22 to verify that the predicted fold resembled anti– or anti-anti–sigma factors ( Figure 2 ) ., All of the identified sigma factor regulators were cloned , expressed in Escherichia coli , and purified ., Using a γ-32PATP transfer assay , we found that the PknD kinase domain ( PknD1–378 ) efficiently phosphorylated Rv0516c , but not any of the other sigma factor regulator homologs ( Figure 3A ) ., The SP600125 inhibitor blocked this Rv0516c phosphorylation ( Figure 3B ) in a dose-dependent manner , indicating that PknD catalyzed the observed phosphorylation ., Rv0516c phosphorylation was reversed by PstP ( Figure 3C ) , the Mtb protein Ser/Thr phosphatase that dephosphorylates all Mtb STPK substrates tested to date 23 ., These results showed that PknD and PstP act on the putative anti-anti–sigma factor Rv0516c in vitro ., To determine if Mtb UsfX ( Rv3287c ) , the anti–sigma factor Ser kinase that controls SigF 24 , also phosphorylates Rv0516c , we incubated these proteins under conditions in which UsfX phosphorylated the model substrate , myelin basic protein ( MyBP ) ., In contrast to PknD , UsfX failed to phosphorylate Rv0516c ( Figure S2 ) ., The anti–sigma factor paralogs RshA 25 and Rv0941c also failed to phosphorylate Rv0516c ( unpublished data ) ., Thus , unlike anti-anti–sigma factors that are phosphorylated by anti–sigma factors , Rv0516c is phosphorylated by a eukaryotic-like STPK , PknD ., To determine if sigma factor regulator phosphorylation is a general function of Mtb STPKs , we assayed the ability of four other Mtb kinase domains ( PknA , PknB , PknE , and PknK ) to phosphorylate Rv0516c and the eight other purified Mtb sigma factor regulators ., At concentrations of each kinase domain equally active in phosphorylating the nonspecific substrate , MyBP , Rv0516c was phosphorylated most efficiently by PknD , and to a lesser extent by PknB and PknE ( Figures 3D and S3 ) ., Neither PknA nor PknK phosphorylated any of the sigma factor regulators in vitro ., PknB and PknE also phosphorylated the anti–sigma factor RshA ( Rv3221A ) ., The role of this phosphorylation remains to be determined , as the phosphorylation of an anti–sigma factor has not been described previously ., PknE weakly phosphorylated Rv1904 and RsfA ( Rv1365c ) ., The five kinase domains tested failed to phosphorylate any of the other sigma factor regulators ., The specific in vitro phosphorylation of Rv0516c by the PknD kinase domain correlated with the transcriptional stimulation of Rv0516c by PknD in vivo ., To explore the mechanism of PknD regulation of Rv0516c , we determined the site of Rv0516c phosphorylation by mass spectrometry and protein sequencing ( Figure 4 ) ., Electrospray ion-trap mass spectrometry revealed a molecular mass of 17 , 392 . 4 ± 1 for purified , recombinant Rv0516c phosphorylated to completion in vitro using the PknD kinase domain ., This mass corresponded to mono-phosphorylated Rv0516c ( the expected MR of the unphosphorylated protein is 17 , 312 . 7 ) ., Matrix-assisted laser desorption ionization ( MALDI ) tandem time-of-flight ( TOF ) analysis of trypsin-digested phospho-Rv0516c indicated that the peptide consisting of the N-terminal nine residues was the only phosphopeptide ( Figure 4A ) ., Tandem MS and N-terminal sequencing showed that Thr2 accounted for all of the Rv0516c phosphorylation ( Figure 4B ) ., Eight Rv0516c N-terminal mutants—individual Ala substitutions of the five Ser or Thr residues in the N-terminal segment as well as deletions of up to eight residues—were created to confirm phosphorylation at Thr2 ( Figure 4C ) ., As expected , PknD failed to phosphorylate the Thr2Ala Rv0516c and the two deletion mutants lacking Thr2 ., The Thr7Ala variant showed a reduction in phosphorylation , suggesting that Thr7 plays a role in kinase binding and recognition ., These data demonstrated that PknD phosphorylates Rv0516c on a unique N-terminal site , Thr2 ., To determine if PknD phosphorylates Rv0516c at Thr2 in vivo , we constructed Mtb strains that co-expressed Rv0516c ( WT or Thr2Ala fused to a C-terminal 3XFLAG tag and expressed using a GroEL promoter ) and full-length PknD ( WT or kinase-dead with no tag expressed using an acetamide-inducible promoter ) ., Western blots confirmed equivalent Rv0516c and PknD overexpression in each strain ( Figure 5A ) ., Using anti–phospho-Thr antibodies , we found that Rv0516c was efficiently phosphorylated only in the strain overexpressing WT PknD and WT Rv0516c ., Mutations that inhibited PknD ( Asp138Asn ) or eliminated the Rv0516c in vitro phosphorylation site ( Thr2Ala ) abolished this phosphorylation of recombinant Rv0516c in Mtb ., These data indicated that PknD phosphorylated Rv0516c on Thr2 in vivo when both proteins were overexpressed ., Although some anti-anti–sigma factor homologs of Rv0516c are known to be regulated by phosphorylation , Thr2 differs from the consensus phosphorylation sites in the anti-anti–sigma factor family ., In Rv0516c , Gly80 occupies the position of the consensus Ser or Thr phosphorylation site , which occurs , for example , at Ser58 of SpoIIAA ( Figure S4A ) ., This consensus Ser/Thr phosphorylation site is conserved in only two of the six Mtb anti-anti–sigma factor domains ( Figure S4B ) , but oxidation of cysteine at this position plays a key regulatory role for at least Rv1365c 24 ., The Rv0516c peptide containing the corresponding segment was identified exclusively in an unphosphorylated form ( unpublished data ) in our MS analysis of PknD-phosphorylated Rv0516c ., The absence of a Ser or Thr at the consensus phosphorylation site and our failure to observe Rv0516c phosphorylation by any anti–sigma factor in vitro are consistent with PknD phosphorylation at the distinct site , Thr2 ., A yeast two-hybrid analysis of interactions among Mtb sigma factor regulators has suggested that Rv0516c can bind the homologous predicted anti-anti–sigma factor , Rv2638 18 ., To test this association and investigate the role of Thr2 phosphorylation in regulating the interaction , we used affinity chromatography to compare binding of purified Rv2638 to Rv0516c before and after PknD phosphorylation ., Rv2638 bound Rv0516c , and Rv0516c phosphorylation by PknD abolished this association ( Figure 5B ) ., These results showed that Thr2 phosphorylation regulates the interaction in vitro between Rv0516c and the anti-anti–sigma factor paralog , Rv2638 ., Because the identity of activating environmental signals remains unknown , we stimulated PknD receptor kinase activity in Mtb by overexpressing the protein ( Figure 1A ) ., Overexpression was expected to stimulate phosphorylation of PknD substrates directly by increasing the concentration of the kinase and indirectly by favoring dimerization ( by mass action ) , which leads to allosteric activation 26 ., PknD activity produced a transcriptional response that altered genes activated by SigF during log phase growth ( Figure 1B ) , including the anti-anti–sigma factor homolog Rv0516c 21 ., Strikingly , the PknD kinase domain also directly phosphorylated the Rv0516c protein ( but none of the other Mtb sigma factor regulators ) in vitro and upon overexpression in vivo ., In contrast to the conserved internal phosphorylation sites found in many anti-anti–sigma factors 16 , PknD phosphorylated Rv0516c on Thr2 in an N-terminal extension similar to that found in two additional mycobacterial anti-anti–sigma factors , Rv1904 and Rv2638 ., Phosphorylation directly blocked Rv0516c binding to Rv2638 , indicating that Thr2 phosphorylation has a direct functional consequence ., Although the roles of the Rv0516c:Rv2638 complex are unknown , alternative phosphorylation sites and functional interactions between upstream anti-anti–sigma factor homologs have been demonstrated in B . subtilis for the RsbS and RsbRA–RsbRD regulators 17 , 27–29 ., These B . subtilis regulators form a large complex that controls the environmental sensing phosphatase RsbU , which dephosphorylates anti-anti–sigma factors 17 , 27–29 ., The correlation between SigF-responsive genes 21 and genes that are transcriptionally sensitive to PknD activity ( Figure 1 ) is specific to PknD; overexpression of Mtb PknB produces a distinct transcriptional response ( T . Lombana , J . MacGurn , J . Cox , and T . Alber , unpublished data ) ., Nonetheless , these data do not demonstrate a direct mechanistic link between Rv0516c and SigF ., To the contrary , the lack of Rv0516c phosphorylation by the Mtb anti-SigF ( UsfX , Rv3287c ) or any other anti–sigma factor hints that PknD indirectly influences SigF-mediated transcription by phosphorylating Rv0516c or other substrates ., The present data do not distinguish models in which repression of the SigF response is caused by Rv0516c phosphorylation or by a distinct signal generated by phosphorylation of one or more other proteins in vivo ., The substrates of the Mtb STPKs are not restricted to transcriptional regulators ., Rough estimates suggest that the number of Ser/Thr phosphoproteins in Mtb may exceed 100 30 , and proposed Mtb substrates include metabolic enzymes 31 , regulatory proteins 12 , 32 , and membrane channels 33 ., Using proteomic methods to analyze lysates phosphorylated in vitro by the PknD kinase domain , Perez and coworkers recently reported that PknD phosphorylates MmpL7 , the transporter for phthiocerol dimycocerosate ( PDIM ) , a lipid essential for virulence 34 ., These studies , however , did not test whether MmpL7 is phosphorylated in vivo , whether phosphorylation altered the function of MmpL7 , or whether MmpL7 is a better substrate of other STPKs ., Perez and coworkers did not detect PknD phosphorylation of Rv0516c , perhaps because this regulatory protein may not be sufficiently abundant in the Mtb lysates or because proteins <20 kDa ( such as Rv0516c ) were run off the two-dimensional gels used to identify potential substrates 34 ., Similar reasons may explain the failure of Perez and coworkers to detect phosphorylation of small proteins containing FHA domains previously found to be in vitro substrates of PknD 9 ., In contrast , the complementary approach used here , based on assaying the biochemical and transcriptional effects of kinase activation in vivo , is sensitive to changes in the activity of regulatory factors , even proteins present in small amounts ., By assaying in vitro PknD phosphorylation of all the Mtb homologs of the B . subtilis SpoIIAA and SpoIIAB sigma factor regulators , we found that only Rv0516c was efficiently phosphorylated ( Figure 3 ) ., Nonetheless , overexpressing PknD may cause abnormal phosphorylation or physiological changes that result in indirect transcriptional changes unrelated to normal kinase functions ., The striking correlation between PknD phosphorylation of Rv0516c and activation of the Rv0516c gene , however , suggests a potential autoregulatory loop and sets the stage to explore the biological roles of this sigma factor regulator and the consequences of Rv0516c phosphorylation in vivo ., Although bacterial STPKs phosphorylate many types of proteins 30 , alternative sigma factor regulators may be substrates of STPKs in diverse genera ., In addition to the activity of PknD , the PknB and PknE kinase domains phosphorylated sigma factor regulators in vitro ( Figures 3D and S3 ) ., In contrast , some kinase domains ( e . g . , PknA and PknK; Figure 3D ) apparently do not phosphorylate these sigma factor regulators ., With up to 12 candidate alternative sigma factors in the Mtb genome , it is unlikely that each kinase controls a completely autonomous pathway ., Instead , our data suggest that phosphorylation pathways may converge on overlapping sets of regulators ( Figure 3D ) ., The specific phosphorylation of Rv0516c on a novel functional site by PknD suggests that STPK phosphorylation of sigma factor regulators goes beyond the paradigm established to date in B . subtilis ., The strains and plasmids used in this study are listed in Table 1 ., M . tuberculosis ( Erdman ) cultures were grown in 7H9 medium and transformed as previously described 35 ., Plasmids were maintained episomally by growth in medium containing antibiotics ., Strains were grown to mid-log phase in 7H9 media before induction of PknD by addition of acetamide ( 0 . 2% ) ., RNA was isolated from cultures at indicated time points as previously described 36 and quantified by measuring OD260 ., RNA was random-primed and reverse transcribed in the presence of amino-allyl dUTP ., Residual RNA was hydrolyzed by addition of 0 . 2 N NaOH , 0 . 1 M EDTA , and incubation at 65 °C for 15 min , followed by addition of 0 . 2 N HCl to neutralize ., The cDNA was purified with Zymo binding columns ( Zymo Research , http://www . zymoresearch . com ) and conjugated to either Cy3 ( individual cDNA samples ) or Cy5 ( common reference pool cDNA ) ., An equal quantity of each RNA sample within an experiment ( representing both Mtb strains at all time points ) was used to make a common cDNA reference pool ., Dye-conjugated cDNA from each individual sample was mixed and co-hybridized with dye-conjugated cDNA from the common reference pool on microarray slides containing oligonucleotide spots representing every gene in M . tuberculosis ( Qiagen , http://www . qiagen . com ) ., After 2 d of hybridization at 63 °C , arrays were washed and scanned using a GenePix 3000B scanner ( Axon Instruments , http://www . moleculardevices . com ) ., Array gridding was performed in GenePix Pro 4 . 1 , and Nomad 2 . 0 was used to select high quality spots ., For each spot , the ratio of medians ( Rm ) was averaged from repeat hybridizations and normalized to t = 0 ( uninduced ) ., Cluster analysis was performed using Cluster 3 . 0 ., Two biological replicates were performed , and each biological replicate was averaged over two hybridizations ., Using Mtb H37Rv genomic DNA as a template for PCR amplification , gene segments encoding PknB1–308 , PknE1–286 , and PknK1–289 were cloned into pET-28b vectors ( Novagen , http://www . emdbiosciences . com ) ., PknD1–378 was cloned into pET-24b ( Novagen ) ., PknA1–337 and full-length clones of each anti–sigma factor or anti-anti–sigma factor were inserted into the Gateway vector pHMGWA 37 , which included NH2-terminal 6X-His and maltose binding protein ( MBP ) tags , followed by a tobacco etch virus ( TEV ) protease site ., All constructs were confirmed by DNA sequencing ., Proteins were expressed in E . coli BL21 CodonPlus ( Stratagene , http://www . stratagene . com ) at 18 °C ., The kinase-domain constructs and Rv0516c were purified to homogeneity ( as assayed by SDS-PAGE ) by immobilized metal affinity chromatography ( IMAC ) using nickel-equilibrated HiTrap chelating Sepharose ( Amersham , http://www . amershambiosciences . com ) , size-exclusion chromatography using HiLoad 26/60 Superdex 75 ( Amersham ) , and anion-exchange chromatography using HiTrap Q Sepharose ( Amersham ) ., The sigma factor regulators prepared for kinase-activity screens were purified by nickel-IMAC ( Rv0516c was purified by IMAC only for these assays as well ) ., The molecular weight of each sigma factor regulator , as assayed by SDS-PAGE , corresponded to the mass predicted by the gene sequence ., Because the kinase-domain constructs autophosphorylated during expression , migration on SDS-PAGE was slightly retarded ., The sigma factor regulators were dialyzed into the reaction buffer ( 80 mM NaCl , 20 mM Tris pH 7 . 5 , 0 . 5 mM Tris ( 2-carboxyethyl ) phosphine hydrochloride TCEP , 250 μM MnCl2 ) ., In a total reaction volume of 19 μl , the final concentration of each IMAC-purified 6X-His-MBP–tagged sigma factor regulator was 20 μM , and the final concentration of kinase was 1 . 2 μM ., The reaction was initiated with the simultaneous addition of 1 μL of γ-32PATP ( 800 Ci/mmol and 10 mCi/ml; ICN , http://www . mpbio . com ) and ATP ( Sigma , http://www . sigmaaldrich . com ) to final concentrations of 250 nCi/μl and 50 μM , respectively ., The reaction was allowed to proceed for 2 h at room temperature and quenched by the simultaneous addition of EDTA to 20 mM and 7 . 2 μg of TEV protease ., The TEV cleavage was allowed to proceed 2 h or overnight at room temperature , resulting in efficient separation of each sigma factor regulator from the tag ., The sequence GlyHisMet was left at the NH2-terminus after TEV cleavage of the tag ., The cleavage reactions were separated by SDS-PAGE on 4%–12% NuPage Novex BisTris gels ( Invitrogen , http://www . invitrogen . com ) , and the gels were dried ., Radioactivity was quantified with a Molecular Dynamics Typhoon 8600 phosphoimager ., To assess the activity of the Asp138Asn mutant of PknD , we incubated 0 . 036 nM WT kinase or 3 . 6 nM Asp138Asn kinase with 0 . 5 mg/ml MyBP or 0 . 5 mg/ml Rv0516c ., The reaction was carried out for 30 min under buffer , metal , and ATP concentrations similar to those described above , and then quenched with either 5X SDS-PAGE loading dye ( for MyBP ) or the TEV/EDTA mixture described above ., Phosphorylation was quantified with ImageQuant ( GE Healthcare , http://www . gehealthcare . com ) after electrophoresis , drying , and phosphoimager data collection ., Untagged PknD1–378 was purified by IMAC , cleaved with TEV , purified on HiLoad 26/60 Superdex 75 ( Amersham ) , and concentrated from the flow-through fraction of a second IMAC column ., Each reaction was set up with or without 38 nM kinase , 20 μM Rv0516c , 250 nCi/μl γ-32PATP , and 25 μM unlabeled ATP in 50 mM NaCl , 50 mM HEPES ( pH 7 . 5 ) , 0 . 5 mM TCEP , 10 mM MnCl2 , and 10 mM MgCl2 ., SP600125 was diluted into water and added to a concentration of 20 nM to 20 μM ., Reactions were carried out and analyzed as described above ., PknD , Rv0516c , and PstP were purified to homogeneity 30 ., Heat-inactivated PknD was prepared by incubation at 95 °C for 1 h ., Phospho-Rv0516c prepared in a 2-h incubation with PknD and γ-32PATP was treated with 2 . 3 μg of PstP ., The reaction was quenched with EDTA and TEV after zero or 30 min ., Separation and quantification were carried out as described above ., Purified Rv0516c was phosphorylated using 6X-His-PknD1–378 and 2 mM ATP ., The reaction proceeded overnight , and the kinase was removed by IMAC ., The flow-through fraction was diluted with water and rocked at room temperature for 2 d to induce precipitation ., Supernatant was removed , and the resulting pellet was dissolved in 6 M guanidinium hydrochloride ., The mass of the intact protein was determined by electrospray ionization–ion-trap mass spectrometry ., Rv0516c was digested with trypsin; the resulting digest mixture was separated on a reversed-phase C-18 column ( 0 . 15 × 150 mm ) , and fractions were collected ., The MALDI TOF spectrum of each fraction was obtained , and the phosphorylated peptide was identified using MALDI-tandem TOF ( MS/MS ) ., The MS/MS spectrum was used , along with Edman sequencing , to identify the phosphorylation site ., Mutations to the N-terminus of Rv0516c were created with QuikChange ( Stratagene ) ., Proteins were purified and phosphorylated as described , except that the Rv0516c variants were treated with TEV protease and quenched with the protease inhibitor , aminoethyl-benzene sulfonyl fluoride ( AEBSF ) ( MP Biomedicals , http://www . mpbio . com ) , prior to phosphorylation ., Full-length PknD was cloned into an acetamide-inducible M . tuberculosis expression vector ( pGWdest3 . kan ) 38 ., Full-length Rv0516c ( WT or mutant ) was cloned into a tuberculosis expression vector under the control of the constitutive GroEL promoter ( pGWdest1 . hyg ) ., The vector also encoded a C-terminal antigenic FLAG ( DDDDK ) tag ., Mutants were made using Quikchange ( Stratagene ) on the Rv0516c gene in the entry vector ., Mtb Erdman was transformed by electroporation 35 , grown for 3–6 wk on solid rich medium , and single colonies were picked and grown to mid-log phase ., Large ( 50 mL ) cultures were inoculated and adjusted to an optical density ( OD ) at 600 nm of 0 . 3 after 5 d ., To induce PknD expression , acetamide was added to a final concentration of 0 . 2% at 24 , 8 , 4 , or 2 h before harvesting ., Induced and uninduced cultures were grown to a final OD600 of ∼0 . 6 ., Then , 10 mL of each culture were harvested by centrifugation , and resuspended in 200 μL of extraction buffer ( 1% SDS , 20 mM EDTA , 50 mM HEPES ( pH 7 . 5 ) , 1 mM AEBSF ) ., Samples were immediately boiled for 25 min ., Next , 200 μL of 100-μm glass beads were added , and samples were bead-beaten twice for 5 min ., Samples were boiled a second time for 10 min and centrifuged ., The soluble fraction was removed and diluted with SDS-PAGE loading dye ., After electrophoresis in 10%–20% Tris-glycine gels , proteins were transferred to PVDF membrane , and detected with anti-phosphoThreonine ( Invitrogen ) , anti-DDDDK ( AbCam ) , anti-KatG , or anti-PknD antibodies ( Pacific Immunology , http://www . pacificimmunology . com ) ., HRP-conjugated secondary antibody was used with Kodak BioMax MR film to develop the Western blots ., In cases of multiple antibody detection , blots were stripped in 2% SDS/100 mM DTT , 62 mM Tris ( pH 7 ) , for 30 min at 50 °C ., Rv2638 was expressed in E . coli in the pHxGWA ( N-terminal His6-thioredoxin ) vector 37 ., Phosphorylated Rv0516c was prepared by incubation of purified His6-MBP–tagged protein with 5 mM MnCl2 , 2 mM ATP , and 10:1 ( w:w ) PknD1–378 ., Phospho-Rv0516c was purified by IMAC , and both phosphorylated and unphosphorylated Rv0516c were dialyzed into the pull-down buffer ( 70 mM NaCl , 20 mM HEPES 7 . 5 , 0 . 5 mM TCEP ) ., Pull-downs were performed by lysing ( by sonication ) E . coli that expressed Rv2638 in the presence of 200 μg MBP-tagged Rv0516c ( phosphorylated or unphosphorylated ) or His6-MBP control in 70 mM NaCl , 20 mM HEPES 7 . 5 , 0 . 5 mM TCEP , 1 mM AEBSF , and 10 mM MnCl2 ., After rocking the lysate for 30 min at 4 °C , samples were centrifuged for 10 min at 14 , 000 rpm in a microcentrifuge at 4 °C ., A small amount of the supernatant was retained for analysis , and the majority was applied to 50 μL of amylose-Sepharose ( New England Biolabs , http://www . neb . com ) pre-equilibrated in the buffer ., After rocking for 10 min at 4 °C , resin was washed three times in buffer , resuspended in 40 μL of 2X SDS-PAGE loading dye , and boiled for 10 min ., Samples were separated on 12% Tris-glycine gels ( Invitrogen ) , transferred to PVDF , blocked in 4% non-fat dry milk , and incubated overnight at room temperature with 1:2000 monoclonal anti-His6 clone HIS-1 ( Sigma ) ., Blots were washed and imaged as described above ., For loading controls , the same reactions were separated on 12% Tris-glycine gels and stained with Coomassie blue ., Please see Table 2 for a list of genes described in this study .

Introduction, Results, Discussion, Materials and Methods, Supporting Information

Receptor Ser/Thr protein kinases are candidates for sensors that govern developmental changes and disease processes of Mycobacterium tuberculosis ( Mtb ) , but the functions of these kinases are not established ., Here , we show that Mtb protein kinase ( Pkn ) D overexpression alters transcription of numerous bacterial genes , including Rv0516c , a putative anti-anti–sigma factor , and genes regulated by sigma factor F . The PknD kinase domain directly phosphorylated Rv0516c , but no other sigma factor regulator , in vitro ., In contrast , the purified PknB and PknE kinase domains phosphorylated distinct sigma regulators ., Rather than modifying a consensus site , PknD phosphorylated Rv0516c in vitro and in vivo on Thr2 in a unique N-terminal extension ., This phosphorylation inhibited Rv0516c binding in vitro to a homologous anti-anti–sigma factor , Rv2638 ., These results support a model in which signals transmitted through PknD alter the transcriptional program of Mtb by stimulating phosphorylation of a sigma factor regulator at an unprecedented control site .

Many bacteria , including Mycobacterium tuberculosis ( Mtb ) , sense the environment using a family of signaling proteins called Ser/Thr protein kinases ( STPKs ) , but the functions of these sensors are not well understood ., This study shows that the Mtb protein kinase ( Pkn ) D STPK attaches a phosphate group to one and only one member of a family of regulators of “alternative” sigma factors , which activate sets of genes in numerous bacteria ., Phosphorylation of the regulator at an unprecedented position abolished binding in vitro to a putative partner ., Remarkably , increasing PknD activity in Mtb not only strongly activated the gene encoding the specific regulatory protein phosphorylated by PknD , but also altered the expression of genes controlled by an alternative sigma factor ., By providing evidence for a mechanistic link between PknD and gene regulation , this work supports a new model in which STPKs in numerous microorganisms transduce environmental signals by controlling expression of specific groups of genes ., Thus , certain bacterial STPKs may orchestrate aspects of the coordinate control of gene expression essential for adaptation in the environment and in host infections .

biochemistry, infectious diseases, in vitro, microbiology, molecular biology, genetics and genomics, eubacteria

journal.pcbi.1006806

Characteristics of measles epidemics in China (1951-2004) and implications for elimination: A case study of three key locations

Measles is a highly contagious viral disease ., Before the availability of an effective vaccine in the 1960s , it infected nearly all children under 15 yr of age ., Due to mass vaccination , the number of measles infections has declined dramatically worldwide and in the Americas , elimination of endemic measles transmission was declared in 2016 1 , 2 ., Given these encouraging outcomes , measles has been targeted for global eradication ., Under the Global Vaccine Action Plan , all six World Health Organization ( WHO ) regions have committed to eliminate measles and five have aimed to achieve this by 2020 1 ., However , as Year 2020 draws close , eradicating measles has proven challenging ., In particular , in China , where reported countrywide vaccination coverage for the last decade has been above 95%—the critical vaccination rate for measles elimination 1 , 3—measles continues to cause large epidemics every year 4 ., To identify the key factors contributing to this persistent transmission , we recently analyzed the geospatial pattern of measles transmission in China during 2005–2014 5 ., Our study showed that industrial cities tended to sustain endemic measles transmission and higher incidence rates and that the large migrant populations attracted to these cities may facilitate this persistence ., However , the mechanism controlling these epidemiologic dynamics remains unclear and longer-term dynamics—in particular , in the years leading to the recent decade—are needed for further diagnosis ., In China , routine disease surveillance has been in place since the 1950s ., During 1951–2004 , measles cases were recorded by the National Notifiable Diseases Reporting System ( NNDRS ) nationwide in monthly intervals 6 ., However , published data on measles at best reported incidence in yearly intervals and/or monthly numbers aggregated across multiple years ., Further , the dynamics of measles are complexly shaped by population susceptibility ( due to newborns , migration , natural infection , or vaccination ) and contact patterns ( e . g . , mixing in schools ) 7–17 , many of which have undergone dramatic changes in China since the establishment of P . R . China in 1949 ., In particular , the birthrate has changed from over 3% in the 1950s to ~1% in recent years due to changing population policies 18 ., Second , as school systems were gradually built and strengthened , the effect of mixing in schools on measles transmission evolved ., Third , measles vaccination was introduced in China in 1965 and since then the level of coverage and vaccination procedures ( e . g . , 1 versus 2 or more doses ) have varied over time 6 , 19 ., Lastly , several major social political changes ( in particular , the Great Chinese Famine in 1959–1961 20 , the Cultural Revolution in 1966–1976 21 , and Economic Reform since 1978 22 ) have profoundly affected population dynamics and in turn the epidemic dynamics of measles , as well as the quality of surveillance data ., These changes altogether create challenges for studying the long-term dynamics of measles epidemics in China ., In this study , we compiled measles incidence data during 1951–2004 , as reported in the literature for three locations—Beijing , Guangzhou , and Shandong—with relatively complete population and vaccination data during the same time period ( S1 Table ) ., Both Beijing and Guangzhou are highly developed cities in China , whereas Shandong is a province of median economic development ( ranked 18th by per capita GDP among 28 provincial level administrative units in mainland China in 1978 ) ., As such , comparison of the long-term dynamics between the two cities ( Beijing and Guangzhou ) and Shandong ( as a “control” ) provides further insight into the underlying mechanisms governing recent epidemic dynamics in China ., Given the sparsity of data , here we develop a model-inference system to infer measles transmission dynamics ., Our model-inference system uses demographic and vaccination data over 1951–2004 as model inputs to directly account for changing population dynamics ( including births , deaths , migration , and vaccination ) ; in addition , it estimates the unobserved time-varying epidemiological parameters , including the basic reproductive number ( R0 ) , population mixing patterns , local seasonality , and reporting rates , based on incidence data ., When fitted to yearly incidence data for the entire population , the model-inference system is able to accurately estimate held-out age-specific data ( i . e . out-of-sample data ) reported for Beijing and Shandong ., As such , we are able to recreate measles transmission dynamics in the three study locations during 1951–2004 in great detail ( i . e . weekly age-grouped estimates as opposed to the yearly data for the entire population ) ., Fig 1 shows the incidence rates in the three study locations—Beijing , Guangzhou , and Shandong—during 1951–2004 ., During 1951–1965 , measles caused large epidemics in all three sites ., Unlike the common biennial cycle observed in developed countries 14 , 23 , measles epidemics occurred almost every year , although a transient biennial cycle was evident in Guangzhou and Shandong in the late 1950s ( Fig 1A ) ., These frequent epidemics were likely fueled by the high birthrates at the time ( Fig 1C ) ., Vaccination programs started in the late 1960s , which dramatically reduced measles incidence; however , the level of vaccine coverage in the early phase ( 1966–1977 ) varied greatly among the three sites , with moderate coverage in Beijing but much lower coverage in Guangzhou and Shandong ., Following the implementation of nationwide mandatory vaccination ( 1-dose during 1978–1985 and 2-doses since 1986 ) , the gaps in vaccination started to close ., Accounting for effectiveness and dose of vaccination , we estimate that immunization rates ( i . e . effective vaccination ) surpassed 80% by 1986 and have increased steadily since ( Fig 1D ) ., This increasing immunization rate clearly contributed to the continuous decline in measles incidence from 1978 to the early 2000s ., Limited monthly incidence data aggregated over approximately a decade ( referred to as quasi-decadal monthly incidence hereafter ) indicate that , during our study period ( 1951–2004 ) , measles epidemics tended to start in Nov/Dec , peak in March/April , and last until July in all three locations ( Fig 1B ) ., However , epidemics peaked earliest and highest in Beijing , which is located northernmost among the three sites , followed by Shandong , and then ~1–2 months later in Guangzhou , which is southernmost ( Fig 1A inset ) ., In addition , measles in Shandong shows a slight shift to later in the year over time ( i . e . , from peaking in March during 1951–1989 to April during 1990–1994 ) as epidemic intensity declined due to mass vaccination ., To infer the transmission dynamics of measles over the five decades , we developed a model-inference system ( Fig 2 ) ., Our epidemic model uses demographic data ( birthrates , age-specific death rates , and migrations ) and vaccination data ( including vaccine coverage , doses and efficacy ) as inputs to account for the aforementioned societal changes during 1951–2004 ., However , some state variables ( e . g . population susceptibility ) and parameters ( e . g . reporting rate and population mixing pattern ) are not observed or documented ., To estimate these variables/parameters and changes over time , the model is run in conjunction with a modified particle filter 24 , 25 , a Bayesian inference method ., This combined model-inference system simultaneously estimates all state variables and parameters based on yearly incidence for the entire population , the most complete measles data for all three study sites ., We first tested the model-inference system using model-generated mock epidemics ( i . e . synthetic testing ) ., When selected by the fits to quasi-decadal monthly incidence , the model-inference system was not only able to reproduce the yearly incidence curve for the entire population ( used as observations in the filter ) and monthly quasi-decadal incidence ( used to select the priors ) , but also the detailed weekly epidemic curves for both the entire population and the key age group ( i . e . 1–14 yr olds; S3–S6 Figs ) ., The correlation between the latter two model-simulated time series ( 2817 weeks over 54 years ) and the truths ( not used in the filter or selection ) was >0 . 89 for the entire population and >0 . 87 for the 1–14 yr olds for all tests ( S2 Table ) ., In addition , the model-inference system was able to identify the optimal prior state-space that spans the true parameter values ( e . g . , S3I–S3R Fig for truth 1 ) , in particular , the amplitudes of school forcing and seasonality ., For the basic reproductive number ( R0 ) , reporting rate , and the mixing parameter m2 ( i . e . , the exponent of the infectious; see Eq 1 in Methods ) , collinearity among the three parameters could lead to compensation of one for another ( e . g . higher R0 with lower m2 , S3I and S3J Fig ) ; however , in general the posterior 95% credible intervals ( 95% CIs ) capture the true values ., Further , this issue was mostly seen in two of the tests ( i . e . , truths 1 and 3 ) and less severe for the other two tests ( S3–S6 Figs ) ., Taken together , these results indicate that our model-inference system is able to truthfully infer the underlying dynamics and key epidemiological parameters , using only the yearly incidence data for filtering and quasi-decadal monthly incidence for selection of parameter priors ., We then used the model-inference system to infer the transmission dynamics of measles during 1951–2004 for each site ., Fitted to yearly incidence for the entire population only , the model-inference system is able to recreate the observed epidemic curves for Beijing , Guangzhou , and Shandong ( Fig 3A ) as well as capture the seasonal dynamics as indicated by the quasi-decadal monthly incidence ( Fig 3B ) ., More importantly , the model-inference system is also able to accurately predict independent , out-of-sample incidence data reported for different age groups ., Compared to data reported for Shandong 26 , the correlations ( r ) between the model-estimates and observations during 1985–2004 are , respectively , 0 . 94 , 1 . 00 , and 0 . 82 for <1 , 1–14 , and 15–50 yr olds , the three most affected age groups ( Fig 3C ) ., Similarly , it accurately predicts out-of-sample yearly incidence for children 1–14 yrs in Beijing ( r = 0 . 99; Fig 3A , 1st panel , inset ) ., However , we note that estimates for infants ( <1 yr ) tended to be slightly lower than reported ( Fig 3C , 1st panel ) ., This is expected , as , for simplicity , our model assumed the same reporting rate for all age-groups whereas the reporting rate for infants was likely higher than average ., These accurate out-of-sample predictions suggest that the model-inference system is able to correctly capture intra-year transmission dynamics , infection age-structure , and observation errors ., With this validated model-inference system , we are thus able to provide detailed estimates of underlying measles transmission dynamics ., While the reported incidence rates in Beijing were about twice as high as in Guangzhou and Shandong during 1951–1965 ( Fig 1A ) , after accounting for reporting rate , the total incidence rates ( i . e . including unreported cases ) were comparable among the three locations during that period ( Fig 4A ) : 2811 ( range: 1619–4116 ) in Beijing , 2978 ( 2098–4218 ) in Guangzhou , and 2653 ( 260–4222 ) in Shandong per 100 , 000 population per year ., After the introduction of vaccine in the late 1960s , incidence in the two cities declined precipitously ., In comparison , due to lower vaccination coverage , in Shandong measles continued to cause large epidemics until the late 1970s ., In addition to inferring total incidence , our model-inference system is also able to estimate population susceptibility during the five-decade record ( Fig 4 ) ., Before 1966 , large epidemics led to similar low susceptibilities in all three locations ., The model-inference system estimates that average susceptibilities were 3 . 9% ( 2 . 3–5 . 1% ) in Beijing , 5 . 8% ( 4 . 2–6 . 7% ) in Guangzhou , and 6 . 1% ( 5 . 1–7 . 0% ) in Shandong during 1951–1965 ., With mass vaccination , susceptibility is determined by both natural infection and immunization ., Thanks to high vaccination coverage ( Fig 1D ) , population susceptibility in Beijing and Guangzhou remained at similar low levels despite much lower epidemic intensity ( Fig 4A ) ., In comparison , due to lower vaccination coverage and fewer infections , population susceptibility in Shandong increased substantially ., The model-inference system estimates that during 1995–2004 , population susceptibility in Shandong increased to 9 . 0% ( 9 . 0–9 . 1% ) , twice as high as in the other two locations ., This large difference in susceptibility was estimated for all ages >1 yr ( Fig 4C–4E ) , in particular for children ( 13 . 3% , 13 . 0–13 . 9% , Fig 4C ) and young adults ( 8 . 8% , 8 . 3–9 . 2% , Fig 4D ) ., Fig 5 shows the estimated spatial temporal variations in key epidemiological parameters over the five decades ., Key epidemiological parameters describe the underlying transmission characteristics of an infection ., For example , the basic reproductive number ( R0 ) , defined as the average number of secondary infections caused by a primary case in a naïve population , indicates the transmissibility of an infection ., For measles , R0 estimates are in the range of 12–18 3 , mostly based on epidemics in industrialized countries prior to mass-vaccination ., Here we estimate that the mean of R0 was near 16 for most of the years during 1951–2004 and stayed at similar levels after the implementation of mass-vaccination in all three sites in China ( see Fig 5A for the full range of R0 estimates ) ., The mean estimates over our study period are 16 . 0 ± 0 . 9 ( Mean ± SD ) in Beijing , 15 . 8 ± 0 . 9 in Guangzhou , and 15 . 9 ± 0 . 5 in Shandong , respectively ., These estimates account for under-reporting ., Estimated reporting rates , though highest in Beijing and lowest in Shandong , have increased substantially over the five decades , reaching 72% 42% , 100% ( Mean and 95%CI ) , 55% 30% , 79% , and 52% 34% , 70% in 2004 in the three sites , respectively ( Fig 5B ) ., In addition to the intrinsic transmissibility of the etiologic agent , the basic reproductive number is also determined by the contact pattern of the host population ., To disaggregate these two factors , our model-inference system explicitly accounts for the latter , which further includes differential contacts among different age groups ( Eq 4 ) as well as imperfect-mixing among the susceptibles and infectious ( represented by the mixing parameters m1 and m2 in Eq 1 ) ., During our study period , estimates for contacts among age groups ( β2 to β6 in the contact matrix; Eq 4 ) , while varied slightly from year to year , did not exhibit a clear secular trend and were similar among the three sites ( S1 Fig ) ., In contrast , we found that the mixing parameter m1 ( i . e . , the exponent of the susceptibles in Eq 1 ) decreased over time with increased vaccination coverage; and similar temporal pattern was estimated for all three sites ( Fig 5C ) ., Similarly , we found that the mixing parameter m2 ( the exponent of the infectious in Eq 1 ) decreased over time; this decrease was most substantial in Beijing ( Fig 5D ) ., In addition , m2 was estimated to be lower in Shandong than Beijing ., This may explain the milder outbreaks in Shandong in recent years despite the higher population susceptibility therein ( Fig 4A ) ., Previous studies have identified mixing in schools as a key factor driving the rise of measles cases following school opening in the fall during the pre-vaccine era 14 , 17 , 27 ., Here we modeled this effect using a school term-time forcing function controlled by the amplitude of forcing ( b1; see Eq 3 ) and school schedules ., Interestingly , the estimated forcing amplitude was higher for Beijing ( ~0 . 9 ) than Guangzhou and Shandong ( ~0 . 6 for both locations; Fig 5E ) , which is consistent with the greater school enrollment rates in Beijing 28 ., In addition , to recreate the observed seasonal patterns , a second seasonal forcing was needed ., The estimated seasonal amplitude ( b . season; see Eq 2 ) was highest in Beijing ( mean = 0 . 76 ) , moderate in Shandong ( 0 . 50 ) , and lowest in Guangzhou ( 0 . 30; Fig 5F ) ., Much of previous measles research has focused on industrialized countries ., To date , the transmission dynamics of measles in China , where the worlds largest population resides , remain largely unknown ., Here we have developed a comprehensive model-inference system that takes into account complex population demographics , contact patterns , mass vaccination , and under-reporting ., When fitted to only yearly incidence data for the entire population using parameter priors selected based on quasi-decadal monthly incidence , our model is able to accurately estimate out-of-sample age-specific epidemic data ., Using this validated model-inference system , we are thus able to reveal epidemiological and demographical characteristics key to measles transmission during 1951–2004 in 3 key locations in China ., These characteristics include age-specific population susceptibility and incidence rates , the basic reproductive number ( R0 ) , reporting rate , the importance of school mixing , and amplitude of seasonality ., The basic reproductive number ( R0 ) for measles is of importance as it is used to inform target vaccination levels for measles elimination ., However , a recent systematic review 29 revealed a large discrepancy in R0 estimates ( 58 estimates ranging from 1 . 43 to 77 . 38 ) and none were estimated for China ., Here we estimate that R0 for Beijing , Guangzhou , and Shandong—two major cities and one province in China—was around 16 , and stayed at similar levels in the pre- and post-vaccine eras ( Fig 5A ) ., Based on our estimate , to eliminate measles in these locations , a minimal population herd immunity of 93 . 8% , or with a vaccine efficacy ( VE ) of 95% 30 , a minimal vaccine coverage of 98 . 7% VT = ( 1–1/R0 ) ÷ VE = ( 1–1/16 ) ÷ 0 . 95 is needed ., Note the latter , more conservative estimate is above the targeted 95% vaccination threshold 1 ., The stable estimation of R0 here is due to careful control of changes in case reporting and population mixing pattern over time ., We found that reporting rates have increased substantially in all three locations over the five decades , consistent with the enhanced disease surveillance in China ., In addition , we also found that the intensity of mixing , as represented by the mixing parameters m1 and m2 , tended to decrease with increased vaccination coverage ., These changes conform with the intuition that vaccination can provide indirect protection to the entire population , i . e . herd immunity ., In contrast , our model-inference system did not identify significant changes in the contact parameters ( β2–6 ) over time ., Such changes could be masked by the wide ranges of our posterior estimates ( S1C–S1G Fig ) as age-specific data were not available for the entire study period to constrain these age-related model parameters ., Nonetheless , the accurate estimates of independent , out-of-sample age-grouped incidence over 20 years for both Beijing ( Fig 3A , 1st panel ) and Shandong ( Fig 3C ) indicate our β estimates are broadly accurate ., Population susceptibility , i . e . the complement of herd immunity , reflects the vulnerability of the population to infection ., This variable is commonly measured by serological surveys ., However , such studies are limited by the small number of people surveyed ( e . g . typically in the hundreds ) ., Here using a model-inference system , we are able to estimate population susceptibility by age group in weekly intervals ( Fig 4 ) ., The estimates reveal that population susceptibility has remained low in Beijing and Guangzhou due to high vaccination rates but has increased substantially in Shandong , particularly in children and young adults ., This differential increase in susceptibility across locations may have profound public health implications for current measles epidemic dynamics in China ., In a recent study 5 , we found that large industrial cities in China with large migrant populations supported endemic measles transmission during 2005–2014 ., Both Beijing and Guangzhou were among such cities; for instance , in 2010 , 35 . 7% ( 7 . 0/19 . 6 millions ) of Beijings population were migrants , of which 8 . 5% came from Shandong ( census data 31 ) ., These migrants , likely from regions with higher susceptibility as in Shandong , could be subject to greater risk of infection ., With an R0 of 16 , a city of 35 . 7% migrants ( assuming 10% and 5% susceptibility for migrants and local residents , respectively ) would have an effective reproductive number ( Re ) of 1 . 09 , slightly above unity and thus capable of sustaining an epidemic ., This simple assessment suggests that migrants may have been ( and continue to be ) a vulnerable subpopulation and contributed to the persistent transmission of measles in big cities despite high vaccination coverage therein ., In addition , this finding suggests that catch-up vaccination targeting migrant populations might be an efficient means of controlling current epidemics in these big cities ., Indeed , such targeted catch-up vaccination has been implemented in Beijing since 2005 ( e . g . , ~2 M migrant workers were vaccinated during 2005–2010 32 ) and substantially reduced the number of infections in migrant workers in recent years 32 ., Our study also reveals interesting differences in measles seasonality among the three study locations ., As found previously for industrialized countries 14 , 17 , increased mixing among school-age children during school terms can facilitate measles transmission ., Among the three sites , the estimated amplitude of school forcing was highest in Beijing , the capital and cultural center of China ., However , additional seasonal forcing was needed to reproduce observed seasonal epidemic patterns ., The estimated seasonal amplitude decreased with decreasing latitudes in the three sites ( Fig 1A ) ., This finding suggests that climate condition may also play a role in measles seasonality ., More specifically , winter indoor heating in cold climates ( e . g . , Beijing and Shandong in this study ) may increase crowding and/or reduce ventilation and hence increase the risk of infection during cold months ., In comparison , there is no indoor heating in Guangzhou due to its mild winters ., We recognize a number of limitations in our study ., First , due to a lack of long-term city-level data , we used Shandong province as a control for the two study cities ( i . e . Beijing and Guangzhou ) ., The aggregate data for Shandong from its many cities ( 17; as of 2018 ) may have masked some of the local characteristics ., In addition , we note that all three sites are located in the more developed coastal region of China and thus may be less representative of inland regions ., Second , synthetic testing indicates that parameter collinearity exists and may reduce the identifiability of certain parameters in our inference system ., In particular , we found that overestimation of R0 can be compensated by underestimation of the mixing parameter m2 or vice versa ., However , this issue is relatively mild ., In addition , out-of-sample data ( e . g . , age-specific data ) can be used to validate model estimates , which was done here ., Third , due to a lack of data , in the model we assumed that migrants have the same susceptibility as local residents ., This may lead to underestimation of the population susceptibility in Beijing and Guangzhou after the mid-1990s , when migrants started to account for >10% of the total population ., More in-depth analyses of this issue based on the findings presented here are underway ., Fourth , for simplicity , we assumed the same reporting rate for all age groups ., In reality , reporting rates for infants and young children are likely higher than older age groups ., As a result , our model-inference system tended to underestimate incidence in infants and overestimate incidence in adults >50 yrs ( Fig 3C ) ., Fifth , we did not include supplementary immunization activities ( SIAs ) in our model ., However , a recent study 26 has evaluated the impact of SIAs on reducing population susceptibility in six provinces in China ( including Shandong ) and found that efficiencies of SIAs prior to 2005 ( <50% ) were lower than later years ( 32–87% ) ., Lastly , for simplicity , we assumed the duration of maternal immunity follows an exponential distribution with a mean of 6 months ., Recent studies 32–34 have suggested that infants born to mothers immunized by vaccination may have weaker and shorter passive immunity relative to the pre-vaccine era , and thus are subject to risk of infection earlier in life ., Future work will test such an impact using more detailed recent data ., In summary , we have developed a model-inference system capable of inferring the underlying transmission dynamics of measles in China , based on sparse observations ., Fitted to highly aggregated incidence data , the model-inference system is able to estimate population susceptibility , the basic reproductive number ( R0 ) and other key epidemiological parameters during 1951–2004 , a period that spans the pre-vaccine and modern mass-vaccination eras ., Our findings reveal population and epidemiological characteristics crucial to understanding the current persistence of measles epidemics in China and for devising future elimination strategies ., Yearly data on demographics ( i . e . , birthrates , death rates , and migrations ) , vaccination coverage and doses , and measles incidence during 1951–2004 reported for Beijing , Guangzhou , and Shandong were used in our model-inference system ., These data ( Fig 1 and S1 Table ) were compiled from many sources or estimated in this study ., Compilation of reported data and estimation of missing variables are summarized in S1 Text ., The main measles transmission model represents susceptible-exposed-infectious-recovered ( SEIR ) dynamics with 4-age groups ( i . e . <1 , 1–14 , 15–50 , and >50 yr olds ) per Eq 1:, {dSidt=−∑j=14βij ( t ) Sim1Ijm2/NjdEidt=∑j=14βij ( t ) Sim1Ijm2/Nj−EiZ+αi ( t ) dIidt=EiZ−IiDdRidt=IiDdMdt=B−M180, ( Eq 1 ), where Si , Ei , Ii , Ri and Ni are , respectively , the numbers of susceptible , exposed ( i . e . latently infected ) , infectious , recovered people and population size in the i-th age group; B is the number of newborns with maternal immunity ( see calculation details and other demographic processes at the end of this section ) and M is the number of infants with maternal immunity ( note , we assume a mean maternal immunity period of 180 days ) ; t is time in days ., The exponents m1 and m2 describe the level of inhomogeneous mixing 35 , 36; Z and D are the latent and infectious period , respectively ., αi ( t ) is the number of travel-related infections ( i . e . seeding ) in the i-th group on day t; it was set to 1 case in Groups 1–3 during two major holidays in China: the national day on Oct 1 and the Chinese New Year in Jan/Feb ., This seeding allows reintroduction of measles after local epidemic extinction ., The transmission rate at time t ( day of the year here ) , βij ( t ) , varies with an annual cycle per:, βij ( t ) =βij{1+b . season⋅cos2π365 ( t−23 ) }, ( Eq 2 ), where βij is the annual mean transmission rate from the j-th to the i-th group and b . season is the amplitude of seasonality ., Note that we shift the phase by 23 days to better match observed seasonality ., In addition , to capture changes in mixing among school-age children , an additional school term-time forcing function is applied to Group 2 ( i . e . 1–14 yr olds ) , such that, β22 ( t ) =β22bTerm⋅1+b1Term ( t ) ⋅{1+b . season⋅cos2π365 ( t−23 ) }, ( Eq 3 ), where b1 is the amplitude of school forcing , and Term ( t ) is set to 1 for school terms , -1 for summer breaks , and 0 . 5 for winter breaks ( note that winter breaks in China last for 5 weeks spanning the Chinese New Year when mixing tends to be higher ) 17 ., Per 17 , we adjust β22 by dividing the mean forcing , i . e . bTerm , such that the school forcing averages to 1 over a year ., For a 4-age group model , the β matrix includes 16 elements ., To reduce the number of parameters , we formulate β using 6 parameters as follows:, β=β1β1β5β4β1β2β6β4β5β6β3β4β4β4β4β4, ( Eq 4 ), where , β1 to β4 represent within-group contact for the four age groups , respectively; β5 ( β6 ) represents mixing between infants ( children ) and parents ., As contact with the elderly tends to be less frequent than other age groups , we set all those related to Group 4 ( >50 yr ) to β4 ., Further , for better configuration of the priors , we set β1 to 1 and estimate the relative magnitude of β2–β6 ( see S1 Text for details ) ., The absolute values of β are then determined by the basic reproductive number ( R0 ) via the relationship 37:, R0=eigenmax ( nβD ), ( Eq 5 ), where eigenmax ( · ) denotes the function giving the maximum eigenvalue of a matrix , and n is a diagonal matrix with elements ni = Ni/∑Ni ( i = 1 , … , 4 ) , i . e . the fraction of population in Group-i ., We then superimposed demographic processes onto the transmission model ( Eq 1 ) ., These include birth , death , aging , migration , and vaccination based on population and vaccination data reported for each year during 1951–2004 ( S1 Table and S1 Text ) ., All processes were updated in daily time step ., Briefly , for the birth process , newborns with maternal immunity ( i . e . , B ) were added to compartment M ( Eq 1 , 5th line ) and the remainder were added to compartment S1 ( i . e . , susceptibles aged <1 yr ) ., For simplicity , we roughly estimated the proportion of newborns with maternal immunity as 1–1/R0 ≈ 1–1/15 = 93 . 3%; that is , the complement of long-term equilibrium susceptibility 3 , 37 , assuming R0 = 15 ( i . e . , the mid-point of the reported 12–18 range 3 ) ., Note that since maternal immunity wanes quickly ( here the mean sojourn time in compartment M was 6 months ) , for R0 values slightly different from 15 , this approximation would only cause a slight shift in timing for a small number of newborns to enter the susceptible pool ( i . e . , S1 ) ., Daily age-specific deaths were subtracted from the corresponding age groups; for simplicity , the same death rate was used for all disease classes from the same age group ., Aging was modeled as an exponential process ., To include the migrant population , daily net numbers of age-specific migrants , assumed to have the same susceptibility and infection rate as the locals , were added to the corresponding age groups ., For simplicity , vaccination was “administrated” in the model at 1-yr of age and with 1 effective dose ., Nevertheless , changes in number of vaccine doses and vaccine efficacy over time were accounted for in our immunization rate estimates , which were used as model input of the effective vaccine coverage ., We applied a modified particle filter 38 , 39 jointly with the model described above and the yearly incidence data to estimate the state variables ( i . e . Si , Ei , Ii , Ri , and M; i = 1 , … , 4 ) and parameters ( D , Z , m1 , m2 , b1 , b . season , R0 , and β2–β6 ) ., Briefly , we first initiated a model ensemble using a suite of random state variables and parameters ( n = 5000 model replicates , or particles ) and ran the model stochastically with a daily time step from 1851 for 100 years to reach equilibrium ., Beginning in 1951 ( when incidence data became available ) , we ran the model in conjunction with a particle filter 38 , 39 to incorporate the data and estima

Introduction, Results, Discussion, Methods and materials

Measles is a highly infectious , severe viral disease ., The disease is targeted for global eradication; however , this result has proven challenging ., In China , where countrywide vaccination coverage for the last decade has been above 95% ( the threshold for measles elimination ) , measles continues to cause large epidemics ., To diagnose factors contributing to the persistency of measles , here we develop a model-inference system to infer measles transmission dynamics in China ., The model-inference system uses demographic and vaccination data for each year as model inputs to directly account for changing population dynamics ( including births , deaths , migrations , and vaccination ) ., In addition , it simultaneously estimates unobserved model variables and parameters based on incidence data ., When fitted to yearly incidence data for the entire population , it is able to accurately estimate independent , out-of-sample age-specific incidence ., Using this validated model-inference system , we are thus able to estimate epidemiological and demographical characteristics key to measles transmission during 1951–2004 for three key locations in China , including its capital Beijing ., These characteristics include age-specific population susceptibility and incidence rates , the basic reproductive number ( R0 ) , reporting rate , population mixing intensity , and amplitude of seasonality ., Key differences among the three sites reveal population and epidemiological characteristics crucial for understanding the current persistence of measles epidemics in China ., We also discuss the implications our findings have for future elimination strategies .

Despite high vaccine coverage , measles continues to cause large epidemics in China , a country currently supporting 18% of the worlds population ., To improve understanding of this phenomenon , here we develop a comprehensive model-inference system; using this system , we are able to simulate measles epidemic dynamics and estimate key epidemiological characteristics in three key locations in China during 1951–2004 , a period that spans the pre-vaccine and modern mass-vaccination eras ., These estimates include spatiotemporal variations in population susceptibility and the basic reproductive number ( R0 ) , an epidemiological parameter commonly used to inform target vaccination levels for measles elimination ., Our findings reveal population and epidemiological characteristics crucial for understanding the current persistence of measles epidemics in China and for devising future elimination strategies .

medicine and health sciences, pathology and laboratory medicine, infectious disease epidemiology, china, pathogens, immunology, geographical locations, microbiology, vaccines, preventive medicine, age groups, viruses, rna viruses, infectious disease control, vaccination and immunization, measles virus, public and occupational health, infectious diseases, paramyxoviruses, medical microbiology, epidemiology, microbial pathogens, people and places, asia, immunity, measles, viral pathogens, biology and life sciences, population groupings, viral diseases, organisms

journal.pntd.0006381

Development and validation of four one-step real-time RT-LAMP assays for specific detection of each dengue virus serotype

Dengue is a worldwide public health concern annually affecting more than 100 million people in tropical and subtropical areas 1 , 2 ., It is caused by dengue virus ( DENV ) , the most common vector-borne viral pathogen of humans , transmitted by mosquitoes of the Aedes genus ( primarily A . aegypti and to a lesser extent A . albopictus ) , as previously reviewed 3 ., DENV infection in humans results in a broad spectrum of disease manifestations , ranging from self-limiting , acute febrile illness ( dengue fever ) to more severe forms of the disease ( dengue haemorrhagic fever and dengue shock syndrome ) , which may lead to death 4 ., In 2013 , the annual global incidence was estimated close to 390 million DENV infections , which was more than three times the dengue burden estimate of the World Health Organization 2 ., DENV is an enveloped virus ( genus Flavivirus , family Flaviviridae ) with a genome that consists of a single-stranded , positive-sense RNA molecule of about 11 kb in length ., The DENV genome encodes three structural proteins ( C , capsid; prM , pre-membrane , and E , envelope ) at the N terminus and seven non-structural ( NS ) proteins ( NS1 , NS2a , NS2b , NS3 , NS4a , NS4b and NS5 ) 5 , 6 ., This virus is classified into four phylogenetically related and loosely antigenically distinct serotypes ( DENV1 , DENV2 , DENV3 and DENV4 ) , each of which contains phylogenetically different genotypes 7–9 ., DENV outbreaks between 2006 and 2013 , in India , Vietnam , Solomon Islands , Myanmar , China , Singapore , Malaysia and Portugal 10–14 , highlight the necessity of rapid virus detection to identify DENV as the cause of an outbreak , in order to manage and control virus spread in infrastructure poor urban , peri-urban and rural settings ., Notably , routine detection of DENV in children who are often asymptomatic carriers could improve outbreak control 15 ., A first vaccine has recently been licensed for the prevention of dengue , which aims to reduce the number of hospitalizations per year , being approved for people aged between 9 to 45 years 16 ., Traditional virus isolation is time-consuming , requires experienced staff , costly facilities and equipment and needs more than seven days to complete the assay 17 , 18 ., IgM- and IgG-capture enzyme-linked immunosorbent assay ( ELISA ) are most widely used but some degree of cross-reactivity against other flaviviruses is usually observed and this method is not useful when antibody titers are not sufficiently high ( febrile viremic phase ) 19 ., Molecular amplification techniques to detect DENV RNA ( RT-PCR , quantitative RT-PCR—qRT-PCR ) , which have emerged as a new standard , have a quick turnaround time and can distinguish DENV serotypes 20–26 ., However , these techniques require sophisticated equipment and experienced staff , making them unpractical for laboratories with limited resources ., Loop-mediated isothermal amplification ( LAMP ) has the potential to substitute PCR-based methods because of its simplicity , rapidity , specificity , sensitivity and cost-effectiveness , as no special equipment is needed ( just a heating block or water bath capable to maintain a constant temperature between 60°C to 65°C ) 27–29 ., Reactions can be visualised by monitoring either the turbidity in a photometer or the fluorescence in a fluorimeter , by visual inspection under UV lamp when using an intercalating dye or by colour change 8 , 28–36 ., Previously reported reverse transcription LAMP ( RT-LAMP ) assays for DENV target the 3’ untranslated region ( UTR ) 8 , 30 , 32 , 34 , 37 , whilst other detect a fragment of the C-prM region 33 , a conserved region of the NS1 36 , or regions of NS2A ( DENV1 ) , NS4A ( DENV3 ) , NS4A ( DENV2 ) and the 3’ UTR ( DENV4 ) 38 ., In all cases information about the primer design is limited as only one sequence per serotype or reference sequences were considered or it is not clearly detailed how the sequence alignment was carried out or how many sequences were included in the design ., An initial screen of all published DENV RT-LAMP detection amplicons quickly revealed that all of them fail to cover the documented sequence variation ., To improve DENV RT-LAMP design we used the LAMP Assay Versatile Analysis ( LAVA ) algorithm 39 which solves the limitations of existing LAMP primer-designing programs by allowing designs based on large multiple sequence alignments ., Our LAMP design is based on 2 , 056 whole-genome DENV sequences covering DENV strains from 2004 to 2014 and yielded 4 one-step , real-time RT-LAMP assays to specifically detect each DENV serotype ., Ethical approval for retrospective use of the anonymized samples in diagnostic development research was available: Tanzania samples ( Ethikkommission Basel in Switzerland , Institutional Review Board of the Ifakara Health Institute and National Institute for Medical Research Review Board in Tanzania ) , IPD and IPC samples ( Ministry of Health of Senegal and National Ethics Committee for Health Research of Cambodia , respectively ) ., Virus material: DENV isolates were provided and tested at the Institut Pasteur in Paris ( Table 1 ) ., TriReagent extracts from flavivirus culture supernatants were provided by M . Weidmann ., Inactivated strains ATCC VR-344 ( DENV1 ) , ATCC VR-345 ( DENV2 ) , ATCC VR-1256 ( DENV3 ) and ATCC-1257 ( DENV4 ) were provided by ENIVD / Robert Koch Institute ., An inactivated Zika virus strain ( ZIKV , H/PF/2013 ) was provided by Prof . Xavier de Lamballerie ( Unité des Virus Emergents , Marseille , France ) ., An External Quality Assessment ( EQA ) 2015 panel was provided by QCMD ( Quality Control for Molecular Diagnostics , Glasgow , UK ) including ten unknown samples ( 15–01 to 15–10 ) ., Patient samples: We used RNA extracts of 31 blood samples collected during a fever study in Tanzania , 2013 ( Table, 2 ) provided by the Swiss Tropical and Public Health Institute in Basel , Switzerland ., These samples included 24 DENV qRT-PCR positive , 2 DENV positive ( not characterized by qRT-PCR ) and 5 negative samples ., In addition , a negative sample from MAST Diagnostica GmbH ( Reinfeld , Germany ) was included ., RNA extracts of 11 DENV qRT-PCR serum samples from Senegal , Sudan and Mauritania collected in November-December 2014 by the Institut Pasteur in Dakar ( IPD ) , Senegal ( Table, 3 ) were tested by qRT-PCR and LAMP in Dakar ., Additionally serum samples from Cambodia collected through the National Dengue Surveillance System 40 were tested ., RNA was extracted and air-dried using pre-dried RNAstable 1 . 5 mL microfuge tubes ( Biomatrica , USA ) from 13 DENV3 and 12 DENV4 samples , collected by the Institut Pasteur du Cambodge ( IPC ) in 2004–2006 and between 2008 and 2014 , respectively ., Samples were shipped at ambient temperature ., Moreover , samples were tested by qRT-PCR before shipment and after receipt and reconstitution in molecular grade water ., Overall the qRT-PCR CT deviation was in a range of 0 . 8 CT ., Five μL RNA of each sample were used per reaction ., RNA extractions were carried out using the RNeasy mini ( DENV strains from Robert Koch Institute , QCMD samples ) ( QIAGEN , Crawley , West Sussex , UK ) and the QIAamp Viral RNA mini ( DENV samples from IPD and IPC and ZIKV strain from Unité des Virus Emergents ) ( QIAGEN , Courtaboeuf , France ) kits ., TriReagent extracts were processed according to the manufacturer’s extraction protocol ( Sigma-Aldrich , Dorset , UK ) ., RNA extraction of the clinical samples from Tanzania was initially performed from 50 μL whole blood using a trial version of a nucleic acid isolation system equivalent to the protocol established for the MagSi-gDNA blood kit ( MagnaMedics , Geleen , The Netherlands ) ., RNA was eluted in 190 μL elution buffer , and 5 μL per sample were used for each RT-LAMP reaction ., Additionally , an improved trial version of the MagnaMedics system for nucleic acid isolation , starting from 100 μL whole blood and eluting the RNA in 100 μL elution buffer , using 5 μL per sample for each RT-LAMP reaction , was used ., RNA was extracted from the clinical samples from Senegal using the QIAamp Viral RNA mini kit ., A DENV RNA standard was transcribed from the DENV 3’ UTR region , ligated into pCRII and evaluated as previously described 41 ., DEN FP and DEN P were as described with the probe now tagged 5’-FAM / BBQ-3’ but an adapted reverse primer DEN RP2 ( 5’-CTGHRGAGACAGCAGGATCTCTG-3’ ) as described 42 ., DENV qRT-PCR was performed using the Light Cycler 480 Master Hydrolysis Probes ( Roche , Mannheim , Germany ) in a 20-μL reaction volume containing 1x LightCycler 480 RNA Master Hydrolysis Probes , 3 . 25 mM activator Mn ( OAc ) 2 , 500 nM primers DEN FP and DEN RP2 , 200 nM probe DEN P , and 1 μL RNA template on the LightCycler 2 . 0 ( Roche ) , as follows: reverse transcription for 3 min at 63°C , activation for 30 s at 95°C , followed by 45 cycles consisting of amplification for 5 s at 95°C and 15 s at 60°C and a final cooling step of 40 s at 40°C ., Analysis of the reactions was conducted using LightCycler software version 4 . 1 . 1 . 21 ( Roche ) ., The Institut Pasteur in Dakar performed a qRT-PCR 43 , using the ABI7500 Fast Real-time PCR System ( Applied Biosystems , Foster City , CA ) ., An RT-PCR assay , which simultaneously detects the 4 DENV serotypes , followed by a nested PCR , that specifically detects each DENV serotype , were used 20 ., A two-step approach was used ., First , all available sequences of DENV1 to 4 were downloaded from the NCBI database ., Searches were limited to the samples collected between 2004 and 2014 ., All sequences were then aligned ( for each serotype ) using GramAlign v3 . 0 44 , and diversity was assessed using the glPCA module of R/adegenet v1 . 4 . 1 45 ., Finally , based on the Principal Component Analysis ( PCA ) and phylogenetic tree ( Neighbor-Joining tree using the R/ape 3 . 2 package ) , the sequences were manually split into different clusters in order to maximise the region of sequence identity ., LAMP DNA signatures for each cluster ( and all combinations to minimise the number of primer sets ) were designed using a modified version https://github . com/pseudogene/lava-dna of LAVA 39 applying the loose parameters set for DENV1-3 and the standard parameter set for DENV4 ., Full scripts and methods are available on GitHub at https://github . com/pseudogene/lamp-denv ., All the designed sets of primers were first checked for primer dimerisation with the VisualOMP version 7 . 8 . 42 . 0 ( DNA Software , Ann Arbor , MI ) ., In addition , primer combinations for each of the DENV assays were tested for primer dimerisation by comparing amplification signals in positive and negative controls ., RT-LAMP reactions were run at 64°C using either an ESE-Quant TubeScanner ( QIAGEN Lake Constance GmbH , Stockach , Germany ) or Genie II ( Optigene , Horsham , UK ) , in a final reaction volume of 25 μL ., The Genie II device displays the annealing curve for specificity after the reaction has finished , by melting curve analysis from 98°C to 80°C ( 0 . 05°C/s ) ., Four RT-LAMP assays were developed , one for each DENV serotype ( S1 File ) ., Each reaction consisted of 1x RM Trehalose , 6 mM MgSO4 , 5% polyethylene glycol ( PEG ) , 1 μL fluorochrome dye ( FD ) , 8 U Bst 2 . 0 DNA Polymerase ( New England BioLabs , Hitchin , Herts , UK ) , 10 U Transcriptor Reverse Transcriptase ( Roche ) and 1 μL template ( DENV RNA or H2O as negative control ) ., For each primer set per RT-LAMP assay , the final concentrations was as follows: 50 nM F3 , 50 nM B3 , 400 nM FIP , 400 nM BIP , 200 nM FLOOP , 200 nM BLOOP ., Before adding the Bst 2 . 0 DNA Polymerase , Transcriptor Reverse Transcriptase and template , mixes were incubated at 95°C for 5 min to melt any primer multi-mers and cooled immediately on ice for 5 min ., Reaction times vary for each RT-LAMP protocol , running for 45 min ( DENV1 ) , 90 min ( DENV2 ) , 75 min ( DENV3 ) and 50 min ( DENV4 ) ., RM Trehalose , MgSO4 , PEG and FD were supplied by MAST Diagnostica GmbH ., Sensitivity analysis was performed in the ESE-Quant TubeScanner ( QIAGEN ) ., Ten-fold dilutions of viral DENV RNA samples ( ATCC VR-344 ( DENV1 ) , ATCC VR-345 ( DENV2 ) , ATCC VR-1256 ( DENV3 ) and ATCC VR-1257 ( DENV4 ) ) , quantified by qRT-PCR , were used to analyse the sensitivity of the developed RT-LAMP assays ( range from 104−105 to 10 molecules/μL ) and 1 μL per dilution was added to the RT-LAMP reaction ., The complete RNA standard was tested in eight separate runs ., The values obtained were subjected to probit analysis ( Statgraphics plus v5 . 1 , Statistical Graphics Corp . , Princeton , NJ ) and the limit of detection at 95% probability for each RT-LAMP assay was obtained ., Cross-specificity tests for the four RT-LAMP assays were carried out at the Institut Pasteur ( Paris ) using the QuantStudio 12K Flex Real-Time PCR System , and results were analysed with the software QuantStudio 12K Flex v1 . 2 . 2 ., ( Applied Biosystems , Carlsbad , CA ) ., Each of the RT-LAMP assays was tested using 1 μL RNA extracted from the DENV strains described in Table 1 ., Cross detection of other flaviviruses , ZIKV , Yellow fever virus ( YFV ) , West Nile virus ( WNV ) and Ntaya virus ( NTAV ) , was analysed using the Genie II ( Optigene ) at the University of Stirling ., The RT-LAMP assays were also tested against several DNA pathogens ( Salmonella Typhi , S . Paratyphi , Streptococcus pneumoniae and Plasmodium falciparum ) ., DNA samples were provided by MAST Diagnostica GmbH ., The performance of the RT-LAMP assays ( sensitivity and specificity ) was additionally evaluated using the 2015 DENV EQA panel provided by QCMD ., Results obtained from QCMD refer to 8 core and 2 educational samples ., Core samples are those needed to assess the performance from the regulatory point of view and educational samples are additional samples related to questions such as limit of detection or specificity ., We used 31 blood samples from a fever study in Tanzania , 2013 ( Table 2 ) ., Twenty-six samples had been confirmed as DENV2 positive by the Swiss Tropical and Public Health Institute ( Basel , Switzerland ) ( 2 of them were not tested by qRT-PCR ) ., Aliquots of these blood samples were sent to MAST Diagnostica GmbH and stored at -20°C until RNA extraction was performed using the Magnamedics kit trial version ., RNA samples were stored at -80°C ., RT-LAMP reactions were run in the TubeScanner TS2 ( QIAGEN ) , using 5 μL RNA of each sample per reaction ., The samples at IPD were analysed by both qRT-PCR 43 , and the DENV1 and DENV2 RT-LAMP assays ( in triplicates ) in an ABI7500 Fast Real-time PCR system ( Applied Biosystems ) , using 5 μL RNA of each sample per reaction ., Sensitivity , specificity , positive predictive value ( PPV ) and negative predictive value ( NPV ) were obtained for the DENV2 RT-LAMP developed when compared against the results obtained by qRT-PCR ., The RNA standard was tested 3 times and similar crossing point ( CP ) values were obtained for the different dilutions from 107 to 103 RNA molecules detected ( S1 Fig ) , showing an efficiency ( E = 10−1/slope—1 ) of 0 . 99 ± 0 . 04 ( mean ± standard deviation , SD ) ., Quantification of DENV1-4 RNA extracted from inactivated isolates ATCC VR-344 ( DENV1 ) , ATCC VR-345 ( DENV2 ) , ATCC VR-1256 ( DENV3 ) and ATCC VR-1257 ( DENV4 ) ( Table 1 ) ranged from 6 . 9x104–9 . 4x104 ( DENV1 ) , 4x105–5 . 3x105 ( DENV2 ) , 1 . 5x105 - 3x105 ( DENV3 ) , and 1 . 8x105–2 . 7x105 ( DENV4 ) RNA molecules/μL ., In total 1 , 145 , 477 , 376 and 58 genomic sequences were retrieved from the NCBI database for DENV1 , DENV2 , DENV3 and DENV4 , respectively ., Each serotype dataset was split into 4 to 21 clusters ( Fig 1A and S2–S4 Figs ) , allowing for the LAVA algorithm to design LAMP primer sets , and was executed for each group separately as well as for all possible combinations of the groups ., Sets of primers that showed dimerisation when analysed with VisualOMP ( DNA Software , Ann Arbor , MI ) were discarded ( Fig 2A ) ., Remaining sets where sequentially combined and tested by RT-LAMP to discard cases of primer dimerisation , visualised by the non-specific amplification signal ( intercalating dye ) in the no template control ( NTC ) ( Fig 2B ) ., The final primer sets are described in Fig 1B and S1–S4 Tables and consist of 84 ( 14 amplicons , DENV1 ) , 72 ( 12 amplicons , DENV2 ) , 48 ( 8 amplicons , DENV3 ) and 18 ( 3 amplicons , DENV4 ) primers ., When combining the amplicon primer sets for each RT-LAMP assay , amplification was not observed when using published standard LAMP primer concentrations for each primer set: 0 . 2 μM F3 , 0 . 2 μM B3 , 1 . 6 μM FIP , 1 . 6 μM BIP , 0 . 8 μM FLOOP and 0 . 8 μM BLOOP ., To determine the concentration window of the complicated primer mix , a 2-fold dilution series of the above primer mix was used ., Amplification yielding the best possible detection without amplification in the NTC was achieved at a dilution of 1:4 ( 50 nM F3 , 50 nM B3 , 400 nM FIP , 400 nM BIP , 200 nM FLOOP and 200 nM BLOOP , Fig 2C ) ., Table 1 and Fig 3 show the results of the cross-specificity and cross-detection tests ., All DENV cell culture RNA extracts were detected and no amplification was observed in the NTC ., The RT-LAMP protocols for DENV2 , DENV3 and DENV4 were specific for each respective serotype ., The RT-LAMP protocol for DENV1 detected all DENV1 RNA strains , but also scored positive in RNA extracts KDH0010A and VIMFH4 containing RNA extracts from DENV3 and DENV4 isolates , respectively ( Table 1 ) ., Additional testing of samples KDH0010A and VIMFH4 by nested RT-PCR ( Fig 4A and 4B ) indicated contamination of the cell cultures samples with DENV1 confirming the RT-LAMP results ., The RNA of other flaviviruses was not cross-detected ( Fig 3 and Table 1 ) ., Specific amplification was also indicated by a specific single peak temperature in the melting curve analysis ( Fig 3B , 3D , 3F and 3H ) , with mean values ± SD of 85 . 4 ± 1 . 1°C ( DENV1 ) , 83 . 1 ± 1 . 0°C ( DENV2 ) , 84 . 3 ± 0 . 9°C ( DENV3 ) and 86 . 4 ± 0 . 3°C ( DENV4 ) ., No amplification was observed when DNA from S . Typhi , S . Paratyphi , S . pneumoniae and P . falciparum was used as template in the different RT-LAMP assays ( Table 1 ) ., The 2015 DENV EQA panel analysis confirmed that the RT-LAMP assays developed passed 8 core and the 2 educational samples of that panel ., Concerning the core samples , 5 positive samples were scored 3/3 , and 1 positive sample was detected once ( the other 2 samples were negative ) ., Results obtained from the educational samples indicated that 1 sample was detected in the 3 repetitions whilst the other sample was detected in 1/3 repetitions ., DENV1-4 RNA samples , previously quantified by qRT-PCR , were used to analyse the sensitivity of the developed RT-LAMP assays ., RT-LAMP protocols for DENV1 , DENV2 and DENV4 detected as few as 10 molecules per reaction , although this amount was only obtained in 3 , 5 and 2 of 8 repetitions , respectively , with the following mean times: 28 . 8 ± 6 . 3 min ( DENV1 ) , 78 . 2 ± 5 . 8 min ( DENV2 ) and 44 . 6 ± 3 . 3 min ( DENV4 ) ., RT-LAMP for DENV3 detected as few as 102 molecules , but only in 4 of 8 reactions , at 44 . 9 ± 18 . 6 min ., The lowest amount of molecules detected in the 8 reactions , showing 100% reproducibility , were 102 ( DENV1 , mean time of 25 . 3 ± 2 . 6 min ) , and 103 ( DENV2 , DENV3 and DENV4 , mean times of 69 . 2 ± 11 . 6 min , 37 . 2 ± 11 . 6 min and 26 . 8 ± 2 . 7 min , respectively ) ( Fig 5 ) ., Considering 8 independent reactions per protocol developed , the probit analysis revealed that the limit of detection at 95% probability for each RT-LAMP was 22 RNA molecules ( DENV1 ) , 542 RNA molecules with a confidence interval from 92 to 3 . 2x1013 RNA molecules ( DENV2 ) , 197 RNA molecules ( DENV3 ) and 641 RNA molecules with a confidence interval from 172 to 1 . 2x105 RNA molecules ( DENV4 ) ., Tables 2 and 3 show the results of the blood and serum samples analyses when using both qRT-PCR and RT-LAMP ., Out of 26 DENV2-infected blood samples 24 scored positive in qRT-PCR with cycle threshold ( CT ) values ranging from 21 . 57–29 . 13 ( Table 2 , column 2 ) ., In a first test DENV2 RT-LAMP detected 17/24 ( 70 . 8% positive samples ) with initial time to positive ( TT ) values between 37 and 89 min ( Table 2 , column 3 ) ., RNA from 14 samples , including those with initial TT values over 60 min , negative in both RT-LAMP and qRT-PCR , and 6 DENV negative samples ( Table 2 ) , were extracted a second time using the optimized MagnaMedics extraction starting from 100 μL sample and yielding enhanced detection ., Five samples with initial TT values from 81–89 min , now tested positive with TT values from 55–77 min ., Six samples initially negative by RT-LAMP became positive with TT values of 61 . 7–72 . 2 min ., Three samples , 1 of which had scored positive in qRT-PCR , remained negative in RT-LAMP ., Most RNA samples extracted with the optimized method scored positive in all 3 replicates ., One sample was detected 2/3 times , and 2 were detected only once ., All negative samples included in these analyses scored negative ., Calculation of the clinical sensitivity and specificity yielded 100% specificity ( CI: 0 . 63–1 . 00 ) , as no false positives were detected , and a sensitivity of 95 . 8% ( CI: 0 . 79–1 . 00 ) with 23/24 positive samples , a PPV of 1 . 00 ( CI: 0 . 85–1 . 00 ) and NPV of 0 . 86 ( CI: 0 . 42–1 . 00 ) ., Table 3 summarises the results obtained with samples collected by the IPD and IPC ., All 11 RNA samples from IPD used in this study were analysed in parallel by qRT-PCR and with DENV1 and DENV2 RT-LAMP assays ., All scored positive in qRT-PCR ( CT 25 . 89–38 . 48 ) , 4 samples scored positive in the DENV1 RT-LAMP , and 7 scored positive in the DENV2 RT- LAMP ( TT values 20–45 min ) ., Samples 267175 , 267197 and 267174 were serotyped as DENV1 with the developed RT-LAMP ., Additionally , of 12 qRT-PCR positive DENV4 samples dried with RNAstable shipped by IPC , 10 tested positive by qRT-PCR after shipment , and 9 were detected by DENV4 LAMP ., Of 13 DENV3 samples qRT-PCR positive before shipment , only 1 tested positive by qRT-PCR on arrival and only 3 by RT-LAMP ., Dengue is now prevalent in more than 100 countries of the tropics and subtropics and as DENV continues to spread , all four serotypes co-circulate widely 46–48 ., The introduction of new DENV strains continues through travellers moving between dengue-endemic countries 11 and recently the capacity of individual mosquitoes to carry multiple DENV serotypes was described 49 , while elsewhere acute simultaneous infection with several DENV serotypes was observed 10 ., DENV detection methods include virus culture , which is time consuming 17 , 18 as well as ELISA or immunofluorescence methods to detect IgM and IgG which suffer from cross-reactivity to other flaviviruses antibodies and which are only considered valid when antibody titers are sufficiently high 19 ., The introduction of NS1 antigen detection has improved the situation and recent studies show a high sensitivity of NS1 detection 50 , with some concluding that the combination with IgM detection can outperform PCR 51 ., However , its use for routine screening in dengue epidemics is questioned in terms of clinical necessity 52 ., For molecular RNA detection , nested PCR 20 and real time PCR-assays 21–26 with high specificity and sensitivity are being used but need expensive and sophisticated thermocyclers and experienced staff ., In recent years , isothermal amplification assays have been described , such as RT-LAMP 8 , 30 , 32–38 and RT-RPA 53 , 54 ., These assays require less expensive equipment and can be delivered in dried pellet format , making handling easier and amenable to poor infrastructure settings ., Worldwide monitoring and the use of Next Generation Sequencing methods have increased the number of complete DENV genomes sequenced and deposited in GenBank to 2 , 988 ( as of June 2016 ) ., It is virtually impossible to use this amount of sequence information to manually align and design amplicons for molecular detection methods ., There have been several attempts to create algorithms to derive signature sequences for PCR techniques from sequence datasets or alignments 55 , 56 ., LAMP amplicons are inherently more challenging to design as they require a minimum of 4 and a maximum of 6 signature sequences ., LAVA software was developed to facilitate the determination of signature sequences for LAMP primer design using a set of aligned sequences 39 ., The original and modified version of LAVA take into consideration the limitations observed with other primer-design programs ( LAMP DESIGNER http://www . optigene . co . uk/lamp-designer/ and PRIMER EXPLORER https://primerexplorer . jp/e/ , such as preventing the use of extensive alignments or sequences longer than 2 , 000 nt ., We used this approach to design serotype-specific primers aiming to match all possible DENV strains circulating worldwide , by considering 2 , 056 available GenBank DENV sequences ( 2004–2014 ) ., This is the greatest difference compared to other previously published RT-LAMP assay designs in which primer design focused on the conserved 3’ UTR , NS1 or C-prM regions but detailed limited information about the DENV sequences used to develop the primers ., As the LAMP primers were designed from different clusters of each DENV serotype obtained after PCA and phylogenetic analyses , the individual LAMP amplicons locate to several regions across the DENV genome conserved in these clusters ( Fig 1 ) ., This allows an overall detection of DENV variability surpassing any other molecular amplification assay ., The final amplicons were selected through a combination of in silico primer dimer formation assessment ( Visual OMP ) and in vitro assessment by checking amplicons selected in the first step for unspecific amplification in the NTC ., A similar methodology has been used to design RT-LAMP primers to detect Chikungunya virus ( manuscript submitted to PLoS Neglected Tropical Diseases ) and we consider this approach would be suitable for the assay development of other infectious diseases ., The final DENV1-4 specific RT-LAMP assays contained 84 , 72 , 48 and 18 oligonucleotides respectively ., The challenge was to find a working concentration of these oligonucleotide mixes , which would allow for sensitive detection ., A 2-fold dilution series approach for the individual final primer mix allowed to identify a working concentration window in the dynamic range of these assays ., This however came at the cost of run time ., In order to increase the reaction speed without losing sensitivity , several combinations of enzymes were tested ., We tested the combination of AMV RT ( Promega , Southampton , UK ) and GspSSD DNA polymerase ( Optigene ) recommended by others who successfully developed rapid RT-LAMP assays with 10–15 minute run times 57 ( Manuguerra personal communication ) ., We also tested Bst 3 . 0 DNA polymerase ( New England BioLabs ) , but found that none offered an advantage over the enzyme combination we used ( Transcriptor Reverse Transcriptase and Bst 2 . 0 ) ., As a matter of fact , we saw an increased level of unspecific amplification with Bst 3 . 0 DNA polymerase ( data non-shown ) ., Thus currently reaction times range from 45 ( DENV1 ) to 90 minutes ( DENV2 ) ., This was not correlated with the number of oligonucleotides in the mixture but may reflect the efficiency of the individual primer sets in the mixture detecting the respective standard strains we used for the validation , and the low oligonucleotide concentration ., Alternative approaches to evaluate the sensitivity of each RT-LAMP would consist of having either a pool of RNA samples representative for each amplicon included or specific primer sets for each particular DENV strain that would be compared with the primer mixtures included in the developed assays ., We used an RNA standard evaluated by qRT-PCR to quantify viral RNA of DENV1-4 ., These quantified RNA were then used to test the analytical sensitivity of the 4 individual specific RT-LAMP assays for the detection of each serotype ., The analytical sensitivities of the DENV1-4 RT-LAMP assays , as estimated per probit analysis , ranged from 22 to 641 RNA molecules detected , and 100% reproducibility after 8 independent runs was achieved for 102−103 RNA molecules detected ., Therefore , results were in the range observed for previously described RT-LAMP methods detecting all four serotypes in a single reaction 8 , 33 , 37 with sensitivities between 10 and 100 RNA molecules detected , and RT-LAMP assays distinguishing the serotypes in individual reactions 30 , 38 ., For the latter assays the analytical sensitivities determined were 10 to 100 plaque-forming units ( PFU ) /mL and 10 RNA molecules detected respectively ., Our RT-LAMP assay for DENV1 showed a limit of detection as per probit analysis of 102 PFU/mL with a confidence interval from 20 to 7 . 8x103 PFU/mL ( data non-shown ) ., The assays developed were serotype-specific , and no cross-detection of other flaviviruses was observed ., Surprisingly , 2 viral preparations tested—KDH0010A ( DENV3 ) and VIMFH4 ( DENV4 ) —were also found positive for DENV1 ., Subsequent analysis by serotype-specific nested PCR 20 confirmed the presence of DENV1 RNA probably due to contamination during RNA extraction or virus culture , and indicating that the DENV RT-LAMP assays had picked up the contamination correctly ., EQA panels have been developed in order to evaluate the performance and reliability of current diagnostic methods in laboratories worldwide , by using different samples ( both negative and positive samples , including different concentrations ) that provide information about their specificity and sensitivity 58 , 59 ., The EQA panel used in this study , provided by QCMD , comprises strains for the 4 DENV serotypes , as well as negative samples ., The analysis showed that our RT-LAMP assays passed all the samples included in the 2015 DENV EQA panel , consisting of 8 core and 2 educational samples ., For evaluation with clinical material , RNA was extracted from whole blood samples collected in Tanzania , confirmed as DENV2 positive by qRT-PCR ., A bead-based extraction protocol was improved and , in addition , instead of using 50 μL whole blood and eluting in 200 μL RNA , the extraction commenced from 100 μL whole blood and RNA was eluted into 100 μL ., Due to this improved extraction protocol , time to positivity reduced from 81–89 min to 55–77 min ., In some cases , there were disparate results between RT-LAMP and qRT-PCR ., Sample 1232 , negative by RT-LAMP , had a CT value of 28 . 78 , and samples 1241 and 1473 , with CT values of 24 . 27 and 29 . 13 , showed current mean TT values of 70 and 73 . 9 min , respectively ., These differences in results observed may not be related to the sensitivity levels of the individual assay and we suggest that the performance of isothermal amplification reactions could be compromised when not using fresh samples , as previously described 53 ., All 11 serum samples collected by Institut Pasteur in Dakar ( 2014 ) , tested positive by qRT-PCR and the DENV1 and DENV2 RT-LAMP assays ., While 3 of the samples could not be characterised with the qRT-PCR protocol , they were successfully amplified by the DENV1 RT-LAMP , providing evidence that determination of serotype is possible when handling samples that have not been serotyped yet ., Based on the results obtained for the fever study in Tanzania , our DENV2 RT-LAMP scored a sensitivity of 95 . 8% ( CI: 0 . 79–1 . 00 ) and specificity of 100% ( CI: 0 . 63–1 . 00 ) in reference to the qRT-PCR used by the Swiss Tropical and Public Health Institute , indicating that all detected as positive by the LAMP assay were truly positive and no false positives were detected ., We used predried tubes of RNAstable for shipment of DENV4 and DENV3 RNA extracts from Institut Pasteur du

Introduction, Materials and methods, Results, Discussion

4 one-step , real-time , reverse transcription loop-mediated isothermal amplification ( RT-LAMP ) assays were developed for the detection of dengue virus ( DENV ) serotypes by considering 2 , 056 full genome DENV sequences ., DENV1 and DENV2 RT-LAMP assays were validated with 31 blood and 11 serum samples from Tanzania , Senegal , Sudan and Mauritania ., DENV3 and DENV4 RT-LAMP assays were validated with 25 serum samples from Cambodia 4 final reaction primer mixes were obtained by using a combination of Principal Component Analysis of the full DENV genome sequences , and LAMP primer design based on sequence alignments using the LAVA software ., These mixes contained 14 ( DENV1 ) , 12 ( DENV2 ) , 8 ( DENV3 ) and 3 ( DENV4 ) LAMP primer sets ., The assays were evaluated with an External Quality Assessment panel from Quality Control for Molecular Diagnostics ., The assays were serotype-specific and did not cross-detect with other flaviviruses ., The limits of detection , with 95% probability , were 22 ( DENV1 ) , 542 ( DENV2 ) , 197 ( DENV3 ) and 641 ( DENV4 ) RNA molecules , and 100% reproducibility in the assays was obtained with up to 102 ( DENV1 ) and 103 RNA molecules ( DENV2 , DENV3 and DENV4 ) ., Validation of the DENV2 assay with blood samples from Tanzania resulted in 23 samples detected by RT-LAMP , demonstrating that the assay is 100% specific and 95 . 8% sensitive ( positive predictive value of 100% and a negative predictive value of 85 . 7% ) ., All serum samples from Senegal , Sudan and Mauritania were detected and 3 untyped as DENV1 ., The sensitivity of RT-LAMP for DENV4 samples from Cambodia did not quite match qRT-PCR ., We have shown a novel approach to design LAMP primers that makes use of fast growing sequence databases ., The DENV1 and DENV2 assays were validated with viral RNA extracted clinical samples , showing very good performance parameters .

The co-existence of several dengue virus ( DENV ) serotypes within the same location and/or individuals as well as a single mosquito being able to carry multiple DENV serotypes highlight the necessity of specific diagnostic tools capable of detect and serotype DENV strains circulating worldwide ., In addition , these methodologies must be highly sensitive in order to detect the genome at low levels ( i . e . , before the onset of clinical symptoms ) and not cross-detect other flaviviruses ., Isothermal amplification methods ( such as reverse transcription loop-mediated isothermal amplification , RT-LAMP ) are affordable for laboratories with limited resources and do not need expensive equipment ., Because of the increasing number of publicly available full DENV genome sequences , traditional primer design tools are not able to handle such huge amount of information ., Therefore , to be able to cover all the diversity documented , we developed 4 complicated oligonucleotide mixes for the individual detection of DENV1-4 serotypes by RT-LAMP ., This approach combined Principal Component Analysis , phylogenetic analysis and LAVA algorithm ., Our assays are specific and do not cross-react with other arboviruses and DNA pathogens included in this study , they are sensitive and have been validated with samples from Tanzania , Senegal , Sudan , Mauritania and Cambodia , showing very good performance parameters .

dengue virus, medicine and health sciences, pathology and laboratory medicine, pathogens, rna extraction, microbiology, geographical locations, viruses, multivariate analysis, mathematics, rna viruses, statistics (mathematics), tanzania, molecular biology techniques, extraction techniques, africa, research and analysis methods, sequence analysis, sequence alignment, bioinformatics, artificial gene amplification and extension, medical microbiology, mathematical and statistical techniques, principal component analysis, microbial pathogens, molecular biology, people and places, flaviviruses, polymerase chain reaction, database and informatics methods, viral pathogens, biology and life sciences, physical sciences, statistical methods, organisms

journal.pntd.0005050

Factors Associated with the Time of Admission among Notified Dengue Fever Cases in Region VIII Philippines from 2008 to 2014

Dengue fever is endemic in the Philippines ., The Department of Health ( DOH ) reported 59 , 943 cases between January and September 2014 , with an incidence of 60 cases per 100 , 000 1 ., Sixty five percent of the cases in the country were admitted , while 35% were treated as outpatient 2 ., Mortality in Dengue cases is often associated with treatment delay 3 ., The delay arises mainly from late hospital admission ., Multiple factors influence the time of admission–for example , distance from hospitals or poor recognition of warning signs and symptoms 4 ., The ownership of hospitals ( ie . whether they are private or public ) also has a significant influence on health-seeking behaviour 5 , 6 ., In private hospitals , early consultations are deterred by the immediate costs 5 ., Data suggests that late admissions in public hospitals could be related to the lack of confidence in their services 6 ., The presence of an epidemic can also cause a change in the time of admission ., In a 2007/2008 epidemic in Rio de Janiero , Dengue cases frequently received primary medical care on the third day from the onset of illness 7 ., This study assessed the effect of the 2010 epidemic in Region VIII 8 , Philippines on the time of admission ., Differences in the admission time have also been associated with disease severity ., Non-severe cases were usually admitted on the third and fourth day since the onset of the disease , while severe cases were admitted later 7 ., In a 2006 study by Tomaschek in Puerto Rico , Dengue cases were seen by a clinician at least once , without appropriate assessment of the severe warning signs , before being admitted 9 ., Late admission of severe cases among adults has been reported by their unusual clinical presentation , such as severe Dengue with no fever 7 ., Early and improved management can reduce morbidity and mortality among Dengue patients ( 3 , 4 ) ., In general , case fatality rate ( CFR ) of 1 may be considered a consequence of insufficient management , late diagnosis and hospital admission 10 ., Therefore , understanding the factors related to the time of hospitalization can improve Dengue management ., In this study , we determined the factors associated with hospital admission time among Dengue cases ., The study period was from 2008 to 2014 in Region VIII , Philippines ., The results of this study may provide recommendations for organizational policies and treatment protocols to improve the admission time of Dengue patients ., Dengue cases are reported weekly to the Philippines Integrated Disease Surveillance and Response ( PIDSR ) through passive surveillance and active sentinel surveillance 11 ., We conducted an exhaustive retrospective sampling and analysis of the notified cases of Region VIII ., The study period lasted from January 1 , 2008 to December 31 , 2014 ., The case definition of Dengue was based on the 1997 World Health Organization ( WHO ) classification 11 ., For this study , Dengue fever ( DF ) was operationally defined as a mild case ., The Dengue hemorrhagic fever ( DHF ) and Dengue shock syndrome ( DSS ) were grouped together , and defined as a severe case ., We measured the outcome variable as the time of admission , measured in the number of days between the onset of illness and the time of admission ., The variable was categorized into early ( 0–2 days ) , regular ( 3–5 days ) and late ( 6 or more days ) admission ., These categories were arbitrarily derived from the clinical phases and period of Dengue: febrile , critical and recovery 12 ., We restricted the explanatory variables to the set of data available in the PIDSR , namely disease severity , age and sex of the patient , hospital sector , hospital level and period of admission ., The hospital sector was categorized into public and private ownership ., The hospital level was categorized into tertiary and non-tertiary hospitals 13 ., The tertiary hospitals were generally located in the urban area of the region ., The period of admission was divided into the epidemic ( 2010 ) and non-epidemic period ( 2008 , 2009 , 2011 , 2012 , 2013 , 2014 ) ., The year 2010 was defined as the epidemic period because of the remarkable increase in the number of cases compared with other years ., Only admitted cases were included in the study ., Values with a missing date of admission and onset of illness were excluded ., Cases with an admission time of over 90 days were excluded , since Dengue can only be confirmed serologically by IgM until 90 days 14 ., We established the association of the explanatory variables to the outcome variable through chi-square test in the univariate analysis and ordinal logistic regression in the multivariable analysis ., We also compared the case fatality rates ( CFR ) across varying times of admission by different factors using linear regression ., The p-value was set at <0 . 05 ., The model used was the proportional odds assumption for ordinal logistic regression 15 ., It assumed that the coefficients between outcome categories ( time of admission ) were similar ., We assumed the coefficients of early admission versus regular and late admission were similar , in regular admission versus late admission ., This is called the proportional odds assumption or the parallel regression assumption ., All statistical analyses were performed with RStudio Version 0 . 98 . 1103 and packages MASS , Hmisc , reshape2 , foreign , ggplot2 , rms , gridExtra ( The R Foundation for Statistical Computing , Vienna , Austria; http://www . r-project . org ) ., The study was authorized and exempted from the bioethics approval though the DOH Region VIII director ., The probability of physical , psychological , social , or economic harm occurring as a result of being included in the research study was minimal ., The names of the patients were replaced with unique keys , ensuring the confidentiality of the information gathered ., There were 21 , 480 reported cases from 2008 to 2014 in Region VIII ., The epidemic period ( 2010 ) accounted for 11 , 974 cases ( 56 . 68% ) ., Only 1 . 40% of these cases were confirmed with laboratory tests ., Most cases ( 85 . 74% ) were admitted in hospitals ., A total of 16 , 357 admitted cases ( 76 . 15% ) were included in the study after excluding the repeated cases ( 1 . 38% ) , non-Region VIII cases ( 0 . 27% ) , outpatients ( 14 . 00% ) , missing values of the outcome variable ( 7 . 29% ) , cases with missing values of the outcome variable entries ( 1 . 38% ) , and no values for the explanatory variable ( 0 . 52% ) , ( Fig 1 ) ., Males and females were equally distributed among admitted cases ., Most of the admitted cases were children ( 70 . 09% ) , and those suffering from a mild disease ( 64 . 00% ) ., Most of the cases were reported from the public sector ( 69 . 82% ) and non-tertiary hospitals ( 62 . 76% ) ., The number of cases was equally proportional during the epidemic and non-epidemic period ( Table 1 ) ., The univariate analysis ( Table 2 ) indicated significant association of the time of admission to disease severity , patient’s age , hospital sector , hospital level and period of admission ( p<0 . 05 ) ., There was no association between sex and time of admission ., Severe cases were admitted later than mild ones , and adult cases were admitted later than child cases ., Although elderly cases ( aged 65 and above ) were also admitted later than child cases , there were only 64 elderly cases in this study ., Public hospital cases were admitted later than private hospital cases ., Similarly , tertiary hospital cases were admitted later than non-tertiary ones ., Cases during the non-epidemic period were admitted later than cases during the 2010 epidemic ., We stratified and described the time of admission by disease severity for sex and age of the patient , hospital level , hospital sector and period of admission ( S1 Table ) ., Both mild cases and severe cases , indicated significant association of the time of admission with age , hospital sector , hospital level and period of admission ( p<0 . 05 ) but not with sex ., Mild cases comprised 69% of the cases in public hospitals , as opposed to 52% in private hospitals ., Mild cases made up 78% of the cases in non-tertiary hospitals , as compared with 40% in tertiary hospitals ., Both mild and severe cases from public and tertiary hospitals were admitted later than those in private and non-tertiary hospitals ., Both severe and mild cases have a significant association between the time of admission and age ( p<0 . 05 ) ., Severe cases made up 37 . 86% of cases among children , as opposed to 31 . 62% in adults ., There were only 22 severe cases among the elderly ., Adults were admitted later than children ., The time of admission for both severe and mild cases was also significantly associated to the epidemic or non-epidemic period ., We found significant association between disease severity , age of patient , hospital sector , hospital level and period of admission in our multivariate analysis ( Table 3 ) ., Late admission was more likely when the following factors occurred:, a ) severe case of the disease ( p<0 . 05 ) in comparison with the mild type;, b ) adult patients rather child patients ( p<0 . 05 ) ;, c ) public hospital admission as opposed to private hospital ( p<0 . 05 ) ;, d ) tertiary level hospital admission rather than non-tertiary ( p<0 . 05 ) ; and, e ) non-epidemic period as opposed to an epidemic period ( p<0 . 05 ) ., The factor with the highest influence was public sector hospital , with a higher probability of late admission in comparison with the private sector ., There was no association between sex and the time of admission ( p = 0 . 62 ) ., There was no significant difference in admission time between the elderly and children ( p = 1 . 37 ) ., The results obtained in S3 Table were used to assess the proportional odds assumption of the model ., The differences in the predicted coefficients in each level were only within the range of 0 . 01 to 0 . 4 ., This suggests that the coefficients at different levels of the time of admission were similar in all the risk factors assessed , and that the proportional odds assumption was held in the model ( 15 ) ., There were 106 deaths among the notified dengue cases with an overall CFR of 0 . 65 ., There was a general increase in the CFR across time of admission ., The CFR was equal or greater than 1 during late admission amongst children , severe disease , tertiary hospital , public hospital and females ( Fig 2 ) ., However , the CFR at different times of admission was only significantly associated with age , severity , hospital level and hospital sector ( p<0 . 05 ) ., It was not associated with sex ( p = 0 . 32 ) and epidemic period ( p = 0 . 15 ) ( S2 Table ) ., Severe cases were more likely to be admitted late in comparison to mild ones ., The CFR of severe cases exceeded 2 during late admission ., Patients with mild disease generally seek medical attention during the period 2–4 days after the onset of illness 14 , 21 ., Under the 1997 classification , severe cases of Dengue are categorised as cases where hemorrhage or shock occurs 3 , 12 ., These symptoms are generally present later than two days from the onset of illness , or at a late stage 14 , 16 , 18 which falls on the period of late admission ., Hence , late admission may have been favoured among severe cases with high CFR based on the 1997 classification ., In 2009 , WHO reclassified Dengue fever based on early warning signs to detect early on cases likely to deteriorate 22 ., This new classification has increased sensitivity to severe cases of Dengue 23 ., However , its use in the Region VIII surveillance was only initiated in 2013 , therefore it was not used in this study ., Dengue infection severity can vary among individuals ., It has been suggested that there are higher levels of viral load in severe cases correlating to the seriousnes of the disease 24 ., The mechanism remains to be determined through viral serotyping , investigating susceptibility and tracing the sequence of the infection 25 ., The symptoms of Dengue seemed to cause little alarm for adult patients , who were generally more likely to experience late admission than children ( Table 2 ) ., On the other hand admission time amongst the elderly were similar to those found amongst children ., A study in Southeast Asia found that cases of Dengue amongst adults showed mild signs and symptoms 25 ., In contrast , a majority of the severe cases occur in children aged 2–15 years ., Adults apparently acquire immunity from primary infection and avoid DHF 25 ., S1 Table of this study is consistent with literature revealing a higher percentage of severe cases in children rather than adults 16 , 25 ., Furthermore , the CFR among children during late admission exceeded 1% ., This emphasizes the important practice of early admission among children who are likely to have severe Dengue ., Dengue patients treated in public hospitals were more likely to be admitted late than those in private hospitals , particularly in mild cases ., There was no difference in admission time between private and public sectors in both mild and severe cases ( S1 Table ) ., The difference in admission time beween the hospital sectors suggests that there may be a disparity in the clinical practice between public and private hospitals ., The CFR in public hospital during late admission approached 1% in comparison to the private hospital ., The financial capacity and remote location of patients influence their health-seeking behaviour ., Public hospitals are frequently visited by patients with more limited economic resources who may endure symptoms of disease until they are critically ill 26 ., Three quarters of the patient load in our study were admitted by public hospitals ., In these hospitals , the doctor’s fee and accommodation are government-subsidized 27 ., However , public hospitals are often so full that there are no vacant beds to accommodate additional admissions ., They also have meager medical supplies and insufficient personnel , which can lead to long waiting hours 27 ., As a result of poor health facilities and the patients’ limited financial capacity , the late admission is more likely in public hospitals than private ones ., In contrast , private hospitals experience a more reduced occupancy rate 27 ., According to anecdotal descriptions from Region VII , private hospitals tend to have higher and faster admission rates , even in mild cases , in order to increase the number of clients for commercial purposes ., In a 2013 report by the National Statistics Coordinating Board , 3 public hospitals and 7 private hospitals in Region VII were classified as tertiary hospitals , while 46 public hospitals and 21 private hospitals were classified as non-tertiary hospitals 28 ., Most of the cases in this study were reported from non-tertiary hospitals ( 63% ) ., In general , tertiary hospitals admit Dengue cases later than non-tertiary ones ( Table 3 ) ., When we stratified the time of admission by disease severity , it appeared that late admission among tertiary hospitals was evident in both mild and severe cases ( S1 Table ) ., There may be a disparity in the time of admission for mild cases ., Most non-tertiary hospitals are funded by the private sector or local government ., Non-tertiary hospitals are generally perceived as providing low quality services because of their meager resources 27 ., Patients who bypass their services and seek treatment in better equipped tertiary hospitals have to travel to the urban area , where national public hospitals or large private hospitals are situated 27 ., These tertiary hospitals have a higher likelihood of late admission than non-tertiary hospitals ., The CFR in tertiary hospitals during late admission exceeded 1 ., Tertiary hospitals are overburdened , because Dengue cases are more common in densely populated urban areas where vectors proliferate 10 ., The 2010 epidemic accounted for the highest number of reported Dengue cases ., The likelihood of late admission was higher during the non-epidemic period than the epidemic period ., This is contrary to evidence suggesting that a high volume of patients can overburden and delay health services 29 , 30 ., During the 2010 epidemic , the number of admissions increased to five times the average per year as compared to the non-epidemic period ., The non-epidemic period covered 6 years which averaged 1 , 399 cases per year ., The 2010 epidemic had 7 , 963 cases in a single year ( Table 1 ) ., Anecdotal reports from the hospital staff revealed the use of the lobby and room extensions to accommodate the large number of patients ., During the 2010 epidemic , public campaigns on early consultation were conducted ., Individuals with fever lasting two or more days were advised to immediately seek consultation 8 ., The community became more prone to suspecting any febrile condition to be Dengue fever ., Anecdotal observations also suggest that increased awareness encouraged to promptly diagnose and admit Dengue cases ., This may explain the lower likelihood of late admission during the epidemic period ., However , the earlier admission may have occurred at the expense of the quality of services ., The 1997 WHO Dengue guidelines classification was used in this study ., The 1997 and 2009 Dengue guidelines use different criteria for categorizing clinical Dengue case severity 3 , 9 ., The 1997 guidelines favour late admission for severe cases , since haemorrhagic and shock presentations naturally occur during the late phase of the disease 3 ., The 2009 guidelines screen through warning signs that could occur during the early to late phase of the disease ., However , the 2009 WHO classification was only adopted in Region VIII surveillance data in 2013 ., As a result , our study only included a limited number of cases categorized under this new classification ., The majority of Dengue cases were suspected and not laboratory confirmed , and thus based on clinical presentation ., Due to the lack of serological laboratory confirmation , it is possible that cases that were not caused by Dengue were nonetheless included in the study ., Other febrile illnesses may have been classified as Dengue in the study , especially during the 2010 epidemic ., It is also worth bearing in mind that other diseases may have been classified as Dengue for insurance coverage reasons , seeing as a higher reimbursement is granted for cases of Dengue fever ., An overestimation of the number of cases may have occurred , but there is no evidence that this affected the distribution of the cases in terms of admission time ., Factors such as chronic diseases , concomitant diseases , immunologic status , Dengue virus serotype , secondary infection , body mass index , inter-hospital transfer , hospital accessibility , or socioeconomic level may also have influenced the time of admission ., However , these factors were not included in the analysis as they are not part of the surveillance information collected ., In Region VIII , the Philippines , late admission of Dengue cases was more likely among adults , public hospitals , tertiary hospitals and during non-epidemic period ., Among the factors , the highest likelihood of late admission was in public hospitals than in private hospitals ., Late admission was associated with hospital sector and level ., These may be influenced by patients’ financial capacity , a high patient load and lack of resources in healthcare facilities , the geographic location of hospitals , and noncompliance with Dengue guidelines ., The study also suggests a higher likelihood of early admission among children in comparison to adults ., This behavior should be encouraged , as severe cases are more common among children than adults ., The severe cases of Dengue were more likely to be admitted late than mild cases ., However , severe diseases have late presentation which may have favoured their late admission ., A CFR of 1 . 00 or greater was observed during the late admission of children , severe diseases , tertiary hospitals and public hospitals ., These suggest that admission time and management should be improved in these factors to minimize death ., Earlier admission was more likely during the 2010 Dengue epidemic in Region VIII than during the non-epidemic period , suggesting behavioural change attributable to increased awareness of the disease ., Improvements in behaviour related to admission time can be brought about through increased knowledge , practice and compliance to the Dengue guidelines by the healthcare workers and population ., We recommend the improvement and consistent implementation of hospital admission practices ., Firstly , in order to facilitate early admission and prevent fatality , health care workers should be able to identify severe Dengue fever cases ., The identification of severe cases may improve with the use of the 2009 WHO Dengue guidelines classification , which examines warning signs and indicates when the patient is admissible ., Trainings should be conducted among healthcare professionals ., The use of the 2009 WHO guidelines as the basis for admission , and the confirmation of the diagnosis through serological tests should be discussed and agreed on among doctors , surveillance staff and health insurance personnel ., The guidelines are expected to standardize the diagnosis and hospital admission time across different hospital sectors and levels ., Through public campaigns , the practice of early admission among children suffering from severe cases of the Dengue disease should be promoted among healthcare professionals ., During epidemics , the provision of examining stations exclusively for children can ensure the on-going practice of early hospitalization ., Secondly , the reasons for late admission in public and tertiary hospitals must be studied further in order to improve promptness in Dengue case management ., Hence , further research will be needed to assess how admission time is affected by the volume of cases , accessibility and out-of-pocket expense of these hospitals The quality of health care services must also be further explored to understand how the health system adapted during the epidemic , and why admission time was earlier during the epidemic rather than non-epidemic period .

Introduction, Methods, Results, Discussion

In cases of Dengue fever , late hospital admission can lead to treatment delay and even death ., In order to improve early disease notification and management , it is essential to investigate the factors affecting the time of admission of Dengue cases ., This study determined the factors associated with the time of admission among notified Dengue cases ., The study covered the period between 2008 and 2014 in Region VIII , Philippines ., The factors assessed were age , sex , hospital sector , hospital level , disease severity based on the 1997 WHO Dengue classification , and period of admission ( distinguishing between the 2010 Dengue epidemic and non-epidemic time ) ., We analysed secondary data from the surveillance of notified Dengue cases ., We calculated the association through chi-square test , ordinal logistic regression and linear regression at p value < 0 . 05 ., The study included 16 , 357 admitted Dengue cases ., The reported cases included a majority of children ( 70 . 09% ) , mild cases of the disease ( 64 . 00% ) , patients from the public sector ( 69 . 82% ) , and non-tertiary hospitals ( 62 . 76% ) ., Only 1 . 40% of cases had a laboratory confirmation ., The epidemic period in 2010 comprised 48 . 68% of all the admitted cases during this period ., Late admission was more likely among adults than children ( p<0 . 05 ) ., The severe type of the disease was more likely to be admitted late than the mild type ( p<0 . 05 ) ., Late admission was also more likely in public hospitals than in private hospitals ( p<0 . 05 ) ; and within tertiary level hospitals than non-tertiary hospitals ( p<0 . 05 ) ., Late admission was more likely during the non-epidemic period than the 2010 epidemic period ( p<0 . 05 ) ., A case fatality rate of 1 or greater was significantly associated with children , severe diseases , tertiary hospitals and public hospitals when admitted late ( p<0 . 05 ) ., Data suggests that early admission among child cases was common in Region VIII ., This behavior is encouraging , and should be continued ., However , further study is needed on the late admission among tertiary , public hospitals and non-epidemic period with reference to the quality of care , patient volume , out of pocket expense , and accessibility We recommend the consistent use of the 2009 WHO Dengue guidelines in order to standardize the admission criteria and time across hospitals .

A variety of factors affect the time of admission of Dengue fever cases ., These must be investigated , as delayed treatment of this disease can result in death ., The authors of this study determined the factors associated with the time of admission among notified Dengue cases of Region VIII , Philippines , from 2008 to 2014 ., The factors assessed were age and sex of the patient , hospital sector , hospital level , disease severity and the presence of Dengue epidemic ., A secondary surveillance data of Dengue was used ., The associations were determined using chi-square test and regression ., Late admission was more likely amongst adults , severe cases of the disease , public hospitals , tertiary level hospitals , and during the non-epidemic period ., In comparison , early admission was more likely in cases concerning children , mild cases of the disease , private hospitals , non-tertiary hospitals and during an epidemic period ., Case fatality was significantly associated to children , severe diseases , public hospitals and tertiary hospitals when admitted late ., The routine early admission of children should be promoted , as severe cases of Dengue fever are more likely among children ., Consistent admission criteria for Dengue should be implemented across all hospital sectors and levels .

medicine and health sciences, elderly, tropical diseases, geographical locations, health care, age groups, chi square tests, adults, mathematics, statistics (mathematics), neglected tropical diseases, research and analysis methods, infectious diseases, dengue fever, epidemiology, mathematical and statistical techniques, hospitals, people and places, statistical hypothesis testing, philippines, health care facilities, asia, disease surveillance, population groupings, viral diseases, geriatrics, physical sciences, statistical methods

journal.pgen.1003609

Re-Ranking Sequencing Variants in the Post-GWAS Era for Accurate Causal Variant Identification

The challenges of precise identification of disease-causing variants underlying GWAS signals have recently received much attention 1–3 ., For post-GWAS statistical analysis that aims to accurately identify potentially causal variants , a major hurdle is the development of methods to distinguish disease-causing variants from their highly-correlated proxies ., While GWAS-era statistical methods focused on identifying associated regions via tag SNPs at the coarse scale of GWAS arrays , next generation sequencing ( NGS ) technology offers the capability to not only detect associated regions , but to distinguish the causal SNPs within these associated regions ., Here we make a distinction between ranking SNPs across the genome to identify an associated region , and ranking to pinpoint the potential causal variant within an associated region ., Identifying an associated region requires that trait-associated SNPs be ranked above null SNPs , while identifying the causal variant requires that , among associated SNPs , associations due to causality are ranked above indirect associations due to other factors , e . g . linkage disequilibrium ( LD ) ., GWAS and imputation studies typically report the top-ranked SNP for each associated locus , and follow-up studies typically attempt replication for these top-ranked SNPs ( for further discussion of ranking see Text S1 ) ., Zaitlen et al 4 proposed a measure of performance for sequencing and fine mapping analysis , their localization success rate metric is the probability that the causal SNP has the top-ranked test statistic within an associated region ., When multiple SNPs are in high LD , the localization success rate drops dramatically 5 ., Udler et al ( 2010 ) investigated the difficulty in overcoming the stochastic effect of high LD among causal and non-causal SNPs 5 ., The sample size required to distinguish the causal SNP can be 1 to 4 times the size required to detect the association at genome-wide significance ., Zaitlen et al 4 showed that this problem could be overcome through joint analysis of samples from carefully selected populations with differing LD structure ., Although candidate causal SNPs will require further bioinformatic or functional study to ultimately delineate potential causal mechanisms , optimized study design and analysis can point to the best possible candidate causal SNP ( s ) and help develop testable hypotheses about biological mechanisms ., Studies of complex traits now underway are leveraging the cost efficiency of integrating GWAS , low- and high-coverage sequencing , and imputation to achieve sample sizes in the tens of thousands 6 , 7 ., For example , the Genetics of Type 2 Diabetes ( GoT2D ) study is combining low and high-coverage sequencing with 2 . 5M-SNP GWAS genotyping and imputation to achieve a total sample size of over 28 , 000 8 ., Sequencing the GWAS sample exploits the GWAS findings to ensure that an association signal is present at the genome-wide level and eliminates the cost of recruiting new individuals ., Analysis of sequenced and imputed SNPs ( post-GWAS data ) can thus be informed by previous GWAS results , allowing a prioritized use of post-GWAS data in fine-mapping regions surrounding significant GWAS tag SNPs 9–11 ., Selection of associated regions for further studies can also be based on combined GWAS and post-GWAS criteria 12 , 13 ., For example , the WTCCC 13 required a marginally significant ( p-value<10−4 ) GWAS SNP to support the evidence at a genome-wide significant imputed SNP ., However , these strategies lead to two important issues that have received little attention in the context of causal SNP identification: ( 1 ) the effect of the re-use of successful GWAS data and ( 2 ) the effect of genotyping error rates that differ between sequenced or imputed SNPs ., The re-use of GWAS data that had contributed to the identification of an associated region for post-GWAS analysis can adversely affect accurate causal SNP identification ., For example , the simulation study of Wiltshire et al 14 showed that when a significant GWAS tag SNP is followed up by sequencing in the same sample , the tag SNP is in fact ranked higher than the true causal SNP 30% to 63% of the time , depending on the genetic model and effect size ., When a GWAS tag SNP is selected based on small p-value , the magnitude of the association at the tag tends to be over-estimated; this form of selection bias is also known as the winners curse 15–19 ., To a variable extent , depending on the LD pattern , this selection bias is carried over from the GWAS tag to post-GWAS sequenced or imputed SNPs 20 ., While this earlier work empirically demonstrated the effect of selection for a significant GWAS tag SNP on the causal SNP , no work to date explores whether it also affects the rank of the causal SNP among all neighboring SNPs within an associated region , and if so how to correct for the bias ., High error rates and differences in error rates , due to differences in coverage , read length and depth , minor allele frequency ( MAF ) , GC content , local sequence structure , and other sequence-specific factors , are common to NGS SNPs and are well-recognized obstacles to analysis 21–29 ., Error rates for low-read-depth sequencing studies are estimated to be 1%–3% 22 , 30 , 31 , and as little as 1% error can produce a large loss in power 27 ., The strategy of low-coverage sequencing in a portion of GWAS samples has been used to discover sequencing variants and build a reference panel to drive imputation in the remaining samples , but the genotyping accuracy can be worse than if all individuals were sequenced 25 ., The choice of lower-coverage design is also motivated by reports that low-coverage sequencing in a large sample , alone or combined with GWAS and imputation data , can achieve superior power to detect associations compared to high-coverage sequencing in a small sample with similar cost 25 , 29 , 32 , 33 ., However , whether the localization success rate of the causal variants responsible for these associations is similarly high has not yet been examined ., High error rates that differ among SNPs also occur in high-coverage sequencing; for example , within targeted high-coverage regions , highly repetitive elements can be difficult to capture resulting in low accuracy for some SNPs 34 ., Differential genotyping accuracy between studies has been shown to reduce power of meta-analysis in the imputation setting 35 , and differential accuracy between cases and controls has been shown to cause confounding and elevated type I error 36 , 37 ., Accounting for differential genotyping accuracy in the association test can recover some of the lost power and reduce type I error 35 , 36 ., However , whether it affects our ability to distinguish causal SNPs from correlated SNPs , and how best to account for the effect of differential genotyping accuracy jointly for all SNPs ( GWAS tagged , imputed or sequenced ) is an open question ., In this report , we first demonstrate that: We develop an analytic description of how these factors influence the probability of localization success and evaluate this probability for a range of plausible parameter values ., We then show how to properly adjust for the adverse effects of these factors with a re-ranking procedure ., We evaluate the performance of the method with extensive simulation studies under a wide range of realistic scenarios , and we demonstrate the practical use of re-ranking with an application to the NCBI BPC3 aggressive prostate cancer GWAS with imputation 38 ., Suppose that M sequenced ( or imputed ) SNPs , Si , i =\u200a1 , … , M , in the region surrounding a significant GWAS tag SNP G are ranked by the magnitude of their association statistics in order to identify the causal SNP C . Table 1 provides the notation for the various parameters and statistics used throughout the report ., Briefly , TSi is the Wald test statistic at a sequenced SNP Si; is the sample Pearson correlation coefficient between the GWAS/imputed/sequenced genotypes ( most likely or fractional allele dosage ) for SNPs G and Si ( r2 is the well-known pair-wise correlation measure of LD between two SNPs ) ; is the estimated correlation between the true genotype and the called genotype for a sequenced SNP Si ( we use correlation as a measure of genotyping accuracy because of its simple interpretation in terms of power and genotyping quality; this quantity is provided by both MACH 24 and BEAGLE 39 software ) ; and are proportions of samples with non-missing genotypes ( termed call rates ) at SNPs G and Si , respectively , and is the joint call rate , the proportion of samples with non-missing genotypes at both SNPs , and is the call rate at the causal SNP ., Let be an estimate of the selection bias in genetic effect estimation at the tag SNP G ( described further below ) , that is the excess in the expected value of the test statistic at the tag SNP G induced by selection based on its small p-value ( or high rank ) ., We call this phenomenon the selection effect ( ΔG is zero if the region was not selected via a tag SNP that achieved the given significance or ranking criterion in the same sample ) ., Our proposed re-ranking statistic for a sequenced SNP Si is ( 1 ) Equation ( 1 ) depends on the selection effect , the tagging effect , the genotyping accuracy effect and scaling factors that depend on the call rates ., Justification for Equation ( 1 ) now follows in the remainder of this section ., ( Full details are provided in Text S2 . ), Without loss of generality , let >0 be the genetic effect ( e . g . the log odds ratio or the regression coefficient in the model relating the phenotype and genotype ) at the causal SNP C which could be: one of the sequenced or imputed SNPs Si , i =\u200a1 , … , M; the GWAS tag SNP G although this is unlikely; or neither if the genomic coverage was incomplete ., Let the tag SNP G be coded such that the coded allele is positively correlated with the causal allele ., Let be the genetic effect estimate and be the estimated standard deviation ( SD ) of the estimate from n observations ., We assume that the distribution of the Wald test statistic at the causal SNP , is approximately normal , , where ., The following also applies to test statistics that are asymptotically equivalent to the Wald test statistic ., Let be the difference between the observed test statistic and its expected value , ( 2 ) Here is the correlation between the genotypes of the causal C and the tag SNP G . ( We assume that the tag is coded so that it is positively correlated with the risk allele of the causal SNP . ) The value of is unobserved and needed only in the theoretical formation of the problem not in the practical implementation , which we discuss later ., The selection effect is most pronounced when there is low power at the tag SNP ., ( For discussion of this point see Text S3 ) ., The conditional distribution of the test statistic TSi at the sequenced SNP Si , conditional on the value of the observed test statistic at the tag SNP G , is ( 3 ) Derivation of this distribution is detailed in Text S2 ., The first term , , is the unconditional expected association signal at the sequencing SNP; the second term , , is the distortion due to the tag SNP selection propagated through correlation ., Therefore , ΔG , the selection effect at the GWAS tag SNP G carries through to each sequenced SNP Si in proportion to the correlation between G and Si ., The combination of attenuation due to LD and upward selection bias at the tag , ΔG , distorts the association evidence so that SNPs in high LD with the tag are more likely to be top-ranked ., We call this phenomenon the tagging effect , and use an estimate to remove bias from the conditional expected value of in ( 3 ) ., Third , differential call rates among SNPs ( , and ) and estimated genotyping accuracy ( is the estimated and is the actual correlation between the called genotype and true genotype ) of sequenced or imputed SNP Si appear in both the numerator and denominator of Equation ( 1 ) ., In the numerator , the tagging bias , , is scaled by a factor of because correlation between the test statistics depends on the individual and joint call rates at the two SNPs ( see Text S2 for derivation ) ., The bias-corrected statistic in the numerator is scaled by because ( 4 ) where is the correlation between the genotype of the causal SNP and the called or estimated genotype of the sequenced SNP ( in contrast to , for the true genotype of the sequenced SNP ) ., Assuming the probability of genotyping error is independent of the actual genotype , then ., It is clear that , without correction , smaller ρSi ( higher genotyping error ) and smaller ( higher missing data rate ) tend to lower the probability that SNP Si would be top-ranked ., We call this phenomenon the genotyping accuracy effect ., To conceptually demonstrate the joint effects of selection , tagging and genotyping accuracy on the localization success rate ( the probability that the causal SNP is topped ranked within an associated region ) , we first consider the simplified case of 2 SNPs , one causal ( from sequencing or imputation ) and one tag ( from GWAS ) with correlation between the two SNPs ranging from r\u200a=\u200a0 . 2 to 1 ( from almost no LD to perfect LD ) ., The inclusion of low LD value is motivated by the fact that correlation between the causal SNP and the best tag is often lower than expected ., The coverage of GWAS platforms tends to be overestimated for both sequenced and imputed SNPs ( see Text S4 for further discussion of this point ) ., We assume that the MAFs of both SNPs are 0 . 12 , the causal SNP has an additive odds ratio ( OR ) of 1 . 25 , and selection at the tag SNP , if present , is based on its association test p-value<0 . 05 in a sample of 1000 cases and 1000 controls ., Localization success rates ( before applying the proposed re-ranking procedure ) for all figures were computed based on Equations ( 2 ) – ( 3 ) and the equation in Text S3 and by numerically integrating over the following bivariate normal density function , ( 5 ) Analytical evaluations of Equation ( 5 ) were used to generate Figures 1–3 , which give insight into the relative influence of the tagging , selection , genotyping accuracy and sample size effects outlined in the Introduction and explicitly defined in Materials and Methods ., We find similar patterns of influence for a rare SNP ( MAF\u200a=\u200a0 . 02 , OR\u200a=\u200a1 . 5; Figures S2 , S4 and S6 ) and a higher frequency SNP ( MAF\u200a=\u200a0 . 25 , OR\u200a=\u200a1 . 25; Figures S3 , S5 and S7 ) , and when the number of non-causal SNPs increases ( Figures S8 , S9 , S10 ) ., The above analytical results demonstrate the need to correct for the joint effects of selection , tagging and genotyping accuracy on the localization success rate ., The practical implementation of the proposed re-ranking statistic in Equation ( 1 ) is as follows ., The estimated selection bias at the tag SNP G can be obtained using BR-squared that provides Bias-Reduced estimates via Bootstrap Resampling at the genome-wide level 40 , 41 ., ( The original program , designed to provide estimates for the genetic effect β , has been modified slightly to provide estimates for the test statistic T; see software documentation on authors website for details . ), The bootstrap estimator can be applied whether the region of interest was selected by rank or by p-value threshold ., Unlike the threshold-based likelihood and Bayesian methods 42–46 , the genome-wide bootstrap method incorporates information across the entire GWAS in order to account for the effects of LD and rank on the bias at each SNP ., The values of the individual and joint call rates are available from the dataset , and genotype correlation can be estimated from the sample ., Correlation between the actual and estimated genotypes at a sequenced SNP can be obtained from the mean posterior genotype ( e . g . MACH ratio of variances estimate , 24 ) or from the full genotype posterior probabilities ( e . g . BEAGLE allelic r2 estimate 39 ) ., An R script that implements Equation ( 1 ) is available ., The R script calls the BR2 software ( http://www . utstat . toronto . edu/sun/Software/BR2/ ) , which provides the essential quantity of if the original GWAS dataset was used for fine-mapping ., We conducted extensive simulation studies to empirically evaluate the performance of the re-ranking method under five general scenarios ( Table 2 ) : The parameter values in Table 2 were chosen to best reflect realistic scenarios ., For example , in order to address realistic tagging , we examined the Affymetrix 5 . 0 chip and identified the SNP that best captured each significant WTCCC T1D GWAS SNP ., The correlation between the two SNPs ranges from r\u200a=\u200a0 . 79 to, 1 . For the range of genotyping accuracy , we note that in practice , the average sequencing ρ can vary substantially from study to study ., For example , for low-coverage studies , it can vary from 0 . 63 to 0 . 99 depending on the coverage , MAF and sample size 25 ., When low-coverage sequencing ( 4× ) and imputation are combined , the average ρ can range from 0 . 89 to 0 . 99 depending on the reference panel size 24 ., Sequencing ρ also depends on MAF; the same error rate in a lower MAF SNP results in a smaller ρ ., Even when the average ρ is high , SNP-level ρ can vary widely within a single study ., Browning and Browning 39 found that imputation with a phased reference panel of 60 Hapmap CEU samples yielded a median ρ of 0 . 95 , however individual ρ was less than 0 . 77 for 20% of the SNPs ., We show that coverage rates can also vary widely between SNPs ( Figure S1 ) by examining the 1000 Genomes low-coverage whole-genome pilot data from chromosome 1 in the CHB and JPT samples ( Figure S1; October 2010 release; 1000 Genomes Project , 2010 ) ., We mimicked this variability in our simulations by randomly assigning each SNP in each dataset an error rate that ranged from zero to twice the overall average error rate ., No random error however was introduced into the genotypes of the tag SNP ( ρG\u200a=\u200a1 ) , because GWAS genotyping has been estimated to be over 99 . 8% accurate 13 , 47 ., In order to ensure realistic correlation structure among post-GWAS sequencing/imputation SNPs , we examined all SNPs in the regions surrounding the WTCCC T1D significant SNPs using the HapMap3 dataset ., The average correlation between adjacent SNPs in these regions was approximately 0 . 975 ., One of the main findings of the simulation study is that GWAS-based region selection or moderate genotyping error can substantially reduce the probability of correctly identifying the causal SNP ( Tables 3–4 and Tables S1 , S2 ) , consistent with that of the analytical study ., For example , results detailed in Table S1 demonstrate that the combined tagging and genotyping accuracy effect can reduce the localization success rate by over 30% ., The simulation study also shows that the proposed re-ranking procedure can recover much of this lost power to identify the causal SNP , increasing the localization success rates by 1 . 5- to 3-fold in many cases ( Table 3 ) ., When genotyping accuracy is high , the power lost due to tagging is small and so re-ranking tends to have little effect ., For studies using GWAS-based selection ( scenario 1 ) , the adverse effects of tagging and genotyping accuracy on localization success rate are strongest when the causal SNP is well tagged ( larger r ) and less accurately sequenced/imputed ( smaller ρ ) ( Tables 3 , 4 and S1 ) ., High-density GWAS followed up with low-coverage sequencing would fall into this category ., Well-tagged causal SNPs tend to suffer from lower localization success rates because the perfectly genotyped tag often captures the association better than the imperfectly sequenced or imputed causal SNP ., Re-ranking corrects this problem , so that the localization success rate does not depend on how well the causal SNP is tagged , except when the tag SNP is in fact the causal SNP ., In this case , the tagging and genotyping accuracy effects actually increase the localization success rate ., After re-ranking , the localization success rate is similar to levels seen when the tag is not causal ., We consider this a minor tradeoff , because the causal SNP is unlikely to be found among the GWAS SNPs for a number of reasons: GWAS SNPs are typically selected independent of the phenotype of interest and post-GWAS SNPs tend to greatly outnumber GWAS SNPs ., When the discovery sample is also used for fine-mapping , but significance is not required at the GWAS-tag SNP ( scenario 2 ) , the genotyping accuracy effect alone could still considerably reduce power to identify the causal variant ( Table 3 ) ., When an independent sample is used for fine-mapping ( scenario 3 , Table 3 ) , localization success rates are very similar to those seen in scenario, 2 . In both cases , the re-ranking method improves the probability of correctly identifying the causal SNP ., The improvement is most pronounced ( 2- to 4-fold improvement ) when genotyping accuracy is low ., When there is more than one causal variant ( scenario 4 , Table 3 ) , we find that re-ranking effectively increases localization success rates for both causal SNPs ., Imperfect call rates affect localization success rate in a similar manner to imperfect genotyping accuracy ( scenario 5 , Table 4 ) ., Equation ( 4 ) implies that a call rate of 0 . 80 should affect the distribution of the causal SNP test statistic in the same manner as a sequencing accuracy ρ of 0 . 89 , and this is borne out in our simulations ., The re-ranking procedure corrects for both missing data and genotyping error to the same degree ., In some cases , investigators are more interested in delimiting a set of best candidate causal SNPs instead of a single top SNP ., In the supplementary material , we include additional simulation results for this scenario ., We define an alternative localization success rate metric as the probability that the causal SNP is in the top 10% of SNPs by rank ( Table S2 ) ., Briefly , we examine the probability that the causal SNP is among the top 5 SNPs when there are 50 total SNPs ( ranked by test statistic or re-ranking statistic ) ., Without re-ranking , the probability that the causal SNP is in the top 10% of SNPs over the region is moderate ., Re-ranking provides an improvement up to 1 . 8-fold ., Machiela et al 28 used the August 2010 release of the 1000 Genomes Project European-ancestry ( EUR ) panel to impute 11 . 6 million variants in 2 , 782 aggressive prostate cancer cases and 4 , 458 controls ., These subjects were genotyped as part of the NCI Breast and Prostate Cancer ( BPC3 ) Cohort Consortium aggressive prostate cancer GWAS 48 , 49; genotyping platforms varied across the seven BPC3 studies , although all used versions of the Illumina HumanHap arrays and most used the Illumina HumanHap 610 Quad array ., The correlation between imputed genotype dosage and genotypes thus varied across studies ., Imputation and association analyses using imputed genotype dosages were conducted separately for each study , and the association results were combined via fixed-effect meta-analysis ., For each imputed SNP , studies with imputation r2<0 . 8 were excluded from the meta-analysis test statistic , leaving a total of 5 . 8 million GWAS and imputed SNPs ., Fine-mapping in the meta-analysis context ranks SNPs by the meta-analysis test statistic ., Re-ranking requires that we compute the correlation between the meta-analysis test statistic on the Z-score scale ( i . e . normally distributed test statistic ) with and without accounting for genotyping error ., Assume Zj is the normally distributed test statistic for study j , and wj is the weight for study j , the meta-analysis test statistic used for the standard naïve ranking isIf is an estimate of pair-wise correlation between the actual and imputed genotypes in study j ( e . g . the square root of allelic-r2 39 , or ratio of variances r2 24 ) , it follows that the estimated correlation between the meta-analysis test statistic computed with perfectly genotyped SNPs ( Zact ) and the meta-analysis test statistic computed with the observed imperfectly genotyped SNPs ( Zobs ) isThe re-ranking statistic in the meta-analysis case iswhere is the meta-analysis test statistic Z scaled for variance of, 1 . Machiela et al 38 reported five statistically independent associated regions within the 8q24 . 21 locus and one for each of 11q13 . 3 and 17q24 . 3 ., We selected all SNPs in LD ( r2>0 . 2 ) with the index SNP from each region for analyses ( Figures 4 and 5 , and Figures S11 , S12 , S13 ) ., In the application , we first ranked SNPs using the naïve test statistics 38; and excluded any SNP with MAF <0 . 01; but unlike Machiela et al 38 we did not exclude any studies ., Machiela et al selected significant regions by examining all imputed and genotyped SNPs at once and so we corrected for the imputation accuracy effect only ( i . e . ) ., Re-ranking identifies new top SNPs for 2 of the 3 associated loci: 8q24 . 21 and 17q24 . 3 ( Figures 4 and 5 respectively ) ., In addition to the most significant region at 8q24 . 21 ( Figure 4 ) , re-ranking also identifies a new top SNP for the third most significant region ( Figure S11 ) ., For both regions re-ranking also identifies SNPs that may have otherwise been missed due to imperfect imputation ., After re-ranking , 2 SNPs in the most significant region at the 8q24 . 21 locus ( Figure 4 ) and 8 SNPs at the 17q24 . 3 locus ( Figure 5 ) move from the lower ranks into the top 10 percent ., On the other hand , SNPs in the top 10% are moved down by only a few ranks ., In this way , re-ranking keeps highly significant SNPs identified by the naïve ranking and adds a few SNPs that would have otherwise been missed ., When the top test statistics are of similar size , re-ranking may identify a new top SNP ., When most SNPs are well-genotyped , re-ranking makes only subtle changes ( Figure S11 , S12 and S13 ) ., There is one poorly imputed SNP at 17q24 . 3 ( rs1014000 , r2\u200a=\u200a0 . 20 ) that moves from the naïve rank of 245 to the new rank of 16 after adjustment ., This SNPs apparent association is largely driven by data from a single study: the naïve rank in the EPIC study is 10 ., When we remove this study from the meta-analysis , the naïve rank is 306 and the adjusted rank is 119 ., No other SNP in the top 10% is this drastically affected when the EPIC study is removed from the analysis ., In the meta-analysis context , we recommend examining top SNPs for heterogeneity among studies when re-ranking produces dramatically different results ., Overall , we observed that the tagging and genotyping accuracy effects are non-trivial sources of bias that could obscure association evidence at the causal SNP ., The proposed re-ranking procedure is simple to implement and can substantially increase the probability of identifying the causal SNP ., For low-coverage sequencing , we recommend the re-ranking method to improve causal SNP identification ., For imputation and high-coverage sequencing , we recommend that unfiltered SNPs in associated regions be examined to see if correlation varies across SNPs and if so , we recommend adjustment with the re-ranking method ., Large changes in rank should be carefully examined for underlying issues such as heterogeneity among meta-analysis studies or differential accuracy between cases and controls , and procedures to correct for these issues should be incorporated ., Re-ranking is most beneficial when genotyping accuracy is moderate to low , that is , the average correlation between the actual and estimated genotypes of post-GWAS ( sequenced or imputed ) SNPs is less than 0 . 97 ., A large number of post-GWAS SNPs in a study may appear to be significant , but when not all were directly genotyped with high accuracy , re-ranking can help select the most probable causal SNPs for follow-up ., High density genotyping followed by low-coverage sequencing in the same sample can produce misleading results , as demonstrated by our simulations , so we do not recommend this design for identifying causal variants ., Our re-ranking method tends to down-rank the tag SNP ., If the tag SNP is suspected to be causal ( e . g . based on prior study ) , we recommend examining the rank of the tag SNP using both the naïve and re-ranked methods when selecting SNPs for further study ., Several imputation and sequencing software packages provide accurate estimates of ρ or quantities from which ρ can be computed 24 , 39 ., Re-ranking depends on accurate estimates of ρ ., Recalibration of sequencing quality scores can greatly improve accuracy and so we recommend this step prior to re-ranking 27 ., Re-ranking is especially important when study-specific factors exacerbate the effects of GWAS-based selection and genotyping error ., Such factors include: high genetic diversity which makes sequencing reads difficult to align 27; low LD among SNPs or lack of population-specific reference panels which makes some populations particularly difficult to impute ( e . g . some African populations 50 ) ; and imputation error which can be as high as 10% for these populations ., Low MAF SNPs tend to suffer from both low power ( which exacerbates the tagging effect ) and high genotyping error ., Re-ranking can be applied to rare and low MAF SNPs with allele counts large enough for test statistics to reach asymptotic normality ., Very low ( 1×−2× ) and extremely low ( 0 . 1×−0 . 5× ) read depth sequencing has received recent attention as a way to maximize cost efficiency and make use of off-target sequencing data 29 , 32 ., Error rates for such regions would be both very high and highly variable among SNPs and so re-ranking to account for errors in the estimated genotypes would be crucial ., When genotyping accuracy is extremely poor , the re-ranking method may not be able to sufficiently improve the localization success rate to ensure useful results ., We recommend that investigators consider the accuracy thresholds recommended by the genotype calling or imputation algorithm they are using before re-ranking is applied ., We emphasize that re-ranking improves the localization success rate when applied to SNPs under the alternative , i . e . SNPs that are themselves causal or in LD with a causal SNP ., Including null SNPs in the re-ranking procedure increases the number of SNPs the causal must out-compete , and so we recommend that only SNPs suspected to be under the alternative be included ., In our application we included all SNPs that had squared pairwise correlation ( r2 ) with the index SNP ( most significant SNP in the region ) greater than 0 . 2 ., Existing methods that incorporate genotype uncertainty into tests for association to reduce power lost due to genotyping error or missing data e . g . 51–54 do not completely recover lost power , and so the genotyping accuracy effect will remain ., The simplest way to deal with genotype uncertainty in a test is to use the expected additive genotype ( i . e . the posterior mean or dosage ) in the standard linear or logistic regression ., In this case , the re-ranking method can be applied using the allele dosages in place of called genotypes as described above ., Guan and Stephens 55 compared several frequentist and Bayesian methods that incorporate genotype uncertainty into tests for association ., The re-ranking procedure could be extended to any case where the correlation between test statistics or Bayes factors can be worked out ., We expect that re-ranking will play an important role as sequencing costs fall and GWAS platform coverage increases ., Ultra-high density GWAS platforms are more likely to include tag SNPs in very high correlation with the causal SNP , which increases power to detect indirect association at the tag SNP ., However , without re-ranking , strong tagging also decreases power to correctly identify the causal SNP in

Introduction, Materials and Methods, Results, Discussion

Next generation sequencing has dramatically increased our ability to localize disease-causing variants by providing base-pair level information at costs increasingly feasible for the large sample sizes required to detect complex-trait associations ., Yet , identification of causal variants within an established region of association remains a challenge ., Counter-intuitively , certain factors that increase power to detect an associated region can decrease power to localize the causal variant ., First , combining GWAS with imputation or low coverage sequencing to achieve the large sample sizes required for high power can have the unintended effect of producing differential genotyping error among SNPs ., This tends to bias the relative evidence for association toward better genotyped SNPs ., Second , re-use of GWAS data for fine-mapping exploits previous findings to ensure genome-wide significance in GWAS-associated regions ., However , using GWAS findings to inform fine-mapping analysis can bias evidence away from the causal SNP toward the tag SNP and SNPs in high LD with the tag ., Together these factors can reduce power to localize the causal SNP by more than half ., Other strategies commonly employed to increase power to detect association , namely increasing sample size and using higher density genotyping arrays , can , in certain common scenarios , actually exacerbate these effects and further decrease power to localize causal variants ., We develop a re-ranking procedure that accounts for these adverse effects and substantially improves the accuracy of causal SNP identification , often doubling the probability that the causal SNP is top-ranked ., Application to the NCI BPC3 aggressive prostate cancer GWAS with imputation meta-analysis identified a new top SNP at 2 of 3 associated loci and several additional possible causal SNPs at these loci that may have otherwise been overlooked ., This method is simple to implement using R scripts provided on the authors website .

As next-generation sequencing ( NGS ) costs continue to fall and genome-wide association study ( GWAS ) platform coverage improves , the human genetics community is positioned to identify potentially causal variants ., However , current NGS or imputation-based studies of either the whole genome or regions previously identified by GWAS have not yet been very successful in identifying causal variants ., A major hurdle is the development of methods to distinguish disease-causing variants from their highly-correlated proxies within an associated region ., We show that various common factors , such as differential sequencing or imputation accuracy rates and linkage disequilibrium patterns , with or without GWAS-informed region selection , can substantially decrease the probability of identifying the correct causal SNP , often by more than half ., We then describe a novel and easy-to-implement re-ranking procedure that can double the probability that the causal SNP is top-ranked in many settings ., Application to the NCI Breast and Prostate Cancer ( BPC3 ) Cohort Consortium aggressive prostate cancer data identified new top SNPs within two associated loci previously established via GWAS , as well as several additional possible causal SNPs that had been previously overlooked .

genome analysis tools, mathematics, genetic association studies, statistics, genetics, biology, genomics, biostatistics, computational biology, human genetics

journal.pntd.0001005

A Low-Tech Analytical Method for Diethylcarbamazine Citrate in Medicated Salt

The World Health Organization has called for an effort to eliminate Lymphatic Filariasis ( LF ) around the world ., 1 A nematode worm ( Wuchereria bancrofti ) is the cause of 90% of lymphatic filariasis cases globally ., Mosquito bites transmit larval nematodes ( microfilariae ) present in the blood stream of infected persons , and although the adult nematodes are resistant to medical treatment , human transmission in endemic regions can be stopped by administering drugs , such as diethylcarbamazine ( DEC ) , that kill the microfilariae ., DEC has had a long history of safe use in mass drug administration ( MDA ) LF eradication programs , 2–4 and so far , W . bancrofti do not appear to have developed resistance to DEC ., 5–6 A course of treatment of 6 mg/kg per day of DEC citrate for 12 days ( daily dose around 300 mg ) can significantly reduce the microfilariae count in an infected person ., However , in regions where the disease is endemic , yearly drug administration to infected individuals must be continued over the adult worm lifetime of 4–6 years to eradicate the disease ., As an alternative to pill-based MDA , DEC can be administered to local populations in the form of medicated cooking salt , with DEC citrate present at 0 . 2–0 . 4% w/w , which corresponds to a daily dose of 20–40 mg DEC citrate ., Local production and distribution of medicated salt fortified with DEC has proved to be a particularly effective method 7–8 for eradicating LF from endemic regions 9–10 ., A partner of the Notre Dame Haiti program , Group SPES in Port-au-Prince , Haiti , produces a double-supplemented salt called “Bon Sel” ., 11 Coarse salt is pre-washed and sprayed with a solution of DEC citrate and potassium iodate ., Iodine levels are routinely monitored on site by a titrimetric method ., However , as of 2010 , the factory had no analytical process for monitoring DEC levels ., Critical analytical issues include, 1 ) determining whether the amount of DEC citrate in each lot of Bon Sel is within safe and therapeutically useful limits ,, 2 ) monitoring variability within and between production runs , and, 3 ) determining the effect of a common local practice ( washing salt before use ) on the availability of DEC ., The “gold standard” assay for DEC citrate uses high-performance liquid chromatography ( HPLC ) ., 12 Sending samples out for analysis would impose unwanted costs and prevent real time analysis of production runs , yet it was impossible to implement this process at the factory in Haiti , which has no access to an HPLC or to the supplies and expertise necessary to maintain one ., Color tests and spectrophotometry have been used for monitoring DEC-medicated salt production , 13–15 although usually for qualitative monitoring ., 16 The facility in Haiti wanted quantitative information but did not have a spectrometer ., The goal of our group was to develop a back titration assay for DEC citrate in medicated salt requiring only a balance , volumetric glassware , and burets , equipment that most iodized salt production programs have on hand for monitoring iodine levels , and compare this method against the benchmark HPLC method ., Samples of untreated NaCl and pharmaceutical grade DEC citrate ( EPICO ) were obtained from the Bon Sel plant in Haiti; pure DEC citrate for HPLC standardization was obtained from Sigma-Aldrich ., The untreated NaCl was a coarse grade produced by evaporation of seawater and had visible contaminants ( dirt , sand , plant matter ) ., 0 . 0040 M HCl was prepared by sequential volumetric dilution of concentrated HCl , and stored in a plastic bottle ., Dilute NaOH solutions are unstable due to reaction with atmospheric CO2 ., A 0 . 200 M NaOH stock solution should be prepared ( it is stable for at least 4 weeks ) and diluted each day to give the working 0 . 0100 M NaOH solution ., Phenolphthalein indicator solution was prepared by dissolving 0 . 5 g of phenolphthalein ( Aldrich ) in 500 mL of a 50% ethanol:water solution ., Standards: DEC citrate standards ( 0 . 05% , 0 . 125% , 0 . 25% , and 0 . 50% w/w of salt ) are prepared in the same matrix as the medicated salt samples ., The final solutions are 10% w/v in salt , thus , to prepare the 0 . 50% standard , 10 g NaCl and 0 . 0500 g DEC citrate are mixed with DI water to give a final volume of 100 ml ., Samples: 5 . 00 g of medicated salt is dissolved in deionized or distilled water to a final volume of 50 . 00 ml with vigorous shaking ( or 10 g/100 ml final volume ) ., A small amount of insoluble residue is usually present in these samples ., Standard DEC citrate used in this study ( from Sigma-Aldrich ) was identical by NMR ( spectra acquired in D2O and d6-DMSO at 400 MHz ) to a sample of the DEC citrate ( manufactured by EPICO ) that is used at the Bon Sel factory in Haiti ., The 1∶1 DEC∶citrate stoichiometry was confirmed by integration of the 1H-NMR peaks from the diastereotopic methylene groups on the citrate and the triplet from the ethyl groups on the DEC ( predicted for a one-to-one stoichiometry of DEC∶citrate: 4∶6 , found 4 . 2∶6 . 0 . ) From the DEC∶citrate stoichiometry , each equivalent of DEC citrate ( see structure in Figure, 1 ) contains three acidic protons ( two carboxylic acids and one protonated tertiary amine ) ., These three acidic protons are visible as a very broad peak at 10 . 5 ppm when the spectrum is acquired in dry DMSO-d6 ., Direct titration of DEC citrate with base did not prove analytically useful ., Due to the range of pKa values in the polyprotic citrate , the end point of the titration was not clear enough ., However , back titration gave a clear endpoint ., In the back titration , a sample of DEC citrate is added to a known excess of the strong base sodium hydroxide , which reacts completely with the acidic protons ., The remaining hydroxide is titrated with standard HCl , giving a clear endpoint with the common indicator phenolphthalein ., Bon Sel also contains small amounts of potassium iodate to supply 40 ppm iodine as a nutritional supplement ., Calibration with DEC citrate standards compensates for any matrix effects from the salt or interference from the iodate ., It should be noted that this analytical method is not as specific or generally useful as the HPLC analysis , because any acidic or basic compound will interfere with the back-titration ., Thus , this test cannot be applied to complex matrices ( e . g . , determination of DEC concentration in cooked food or in body fluids ) ., Titration of standard samples gave a linear calibration curve ( Figure 2 ) ; the linear least-squares parameters were determined in Excel using the LINEST function and used to fit unknown samples ., The linear range extends from 0 . 050% to 0 . 88% ( w/w DEC citrate in salt ) , which covers the normal therapeutic range of DEC in salt ( 0 . 1–0 . 6% , recommended 0 . 2–0 . 4% ) ., 17 The average relative standard deviation ( RSD ) for the concentration of known DEC samples at Notre Dame was 16±9% by the titration method , based on triplicate analysis of samples ranging from 0 . 10% to 0 . 90% DEC citrate ., Samples analyzed in Haiti gave an average RSD of 33±7% ., The limit of detection ( LOD\u200a=\u200a3*s/m ) and limit of quantification ( LOQ\u200a=\u200a10*s/m ) were calculated; 18 m is given by the least square fit to the slope of the calibration curve , and s is the standard deviation of 7 determinations of DEC concentration for the 0 . 050% standard sample ., The LOD is 0 . 029% and the LOQ is 0 . 096% for the titration method ., To compare the titration method and the HPLC method , multiple standards and unknowns were analyzed with both methods ., Figure 3 shows the results plotted against each other; the observed slope of the line is 1 . 014 ( for perfect agreement it would be 1 . 00 ) ., The accuracy of the titration method was indistinguishable from that of the HPLC method ., Applying the paired t-test 19 for the 10 samples listed in Table 1 , the mean difference between the titration and HPLC results was −0 . 0018 , the std deviation was 0 . 016 , and tcalc is 0 . 35 ., This indicates that the difference between the titration and HPLC results was not statistically significant for samples at concentrations of 0 . 1%–0 . 8% , although the precision of the HPLC method was superior ( RSD <5% for HPLC ) and its LOQ was much lower ., Analysis of Bon Sel samples from seven production runs in mid-2009 showed that all seven production lots ranged from 0 . 09–0 . 13% DEC citrate , with an average of 0 . 10%±0 . 01% ., ( Table, 2 ) This shows that spray coating is an effective technique for achieving uniform DEC loading on salt at the kg-to-kg and lot-to-lot level ., The loading achieved , while in the therapeutic range ( 0 . 1–0 . 6% w/w ) , was lower than the desired loading of 0 . 2–0 . 4% w/w ., The loading is a function of the solubility of the DEC citrate in the spraying solution , the drying rate of the salt , and the salt feed rate , and could not be improved with the equipment on hand ., However , the group in Haiti tried an experimental run where a finished batch of salt was dried and fed back into the sprayer; this double-sprayed salt analyzed at 50±7 ppm iodine and 0 . 28±0 . 7% w/w DEC citrate ( Table 2 , entries X1 ( single sprayed ) and X2 ( double sprayed ) ) ., To monitor heterogeneity within the bags of Bon Sel , three 10 g grab samples from each of several 1 kg bags of Bon Sel ( taken from different lots ) were tested; the levels of DEC citrate varied from 0 . 08 to 0 . 15% for samples taken within the same bag of Bon Sel ., This heterogeneity was not due to errors in the titration analysis , as the results were confirmed by HPLC analysis , which has a much higher precision ., Because the DEC is sprayed onto the salt , which contains both coarse ( low surface area ) and fine ( high surface area ) crystals , DEC loading is expected to be a function of salt crystal size ., Two lots of Bon Sel from the mid-2009 production runs were screened to separate particles >4 mm in size from particles <4 mm in size; in each case , the large crystals had significantly lower loading of DEC than the small crystals ., For example , in one lot , the large crystals gave a DEC loading of 0 . 034±0 . 001% while the small crystals came in at 0 . 085±0 . 002% ( these low loadings were measured using HPLC to obtain more precise results ) ., The variation in loading with crystal size appears to be large enough to account for most of the heterogeneity in the within-lot analyses , and suggests that more uniform spray coating and higher loadings would be achieved by crushing the salt before spraying it ., The salt available in Haitian markets is often of low purity , and many people rinse the salt before using it in cooking ., Although Bon Sel is pre-washed and the packaging advises consumers not to wash the salt , habits can be hard to break , and some people probably still wash the Bon Sel ., Tests on the effect of hand rinsing ( ∼5 seconds swirling in a bowl of water , or a similar time under a stream of water ) showed retention of 40–50% of the DEC citrate and 60–70% of the iodate after the medicated salt was washed ., This result suggests that a fortification level of 0 . 3–0 . 4% DEC citrate , at the high end of the recommended scale , would be likely to deliver therapeutically useful doses to consumers of the medicated salt regardless of whether or not they rinse it ., A simple titration-based assay allows determination of diethylcarbamazine ( DEC ) citrate concentrations in medicated salt produced in Haiti for an anti-lymphatic filariasis program ., The assay can be carried out with widely available equipment and materials and thus offers a useful tool for quality control and field analysis of DEC ., The development of this method , which allows quantification of the medication , DEC citrate , has already proven useful for quality control in the Haiti plant where salt fortification takes place ., Historically , identification and communication of flaws in the salt fortification levels have taken several months as samples were sent back to the US for analysis ., Using the back titration analysis of DEC , chemists in Haiti can now identify variation in DEC loading as batches of Bon Sel are produced ., This analysis will allow the Bon Sel plant to act more rapidly and independently in their effort to supply the area with properly medicated salt ., An increased efficiency in Bon Sel production should bolster the endeavor to reduce and eventually eliminate lymphatic filariasis in Haiti .

Introduction, Materials and Methods, Results/Discussion

The World Health Organization has called for an effort to eliminate Lymphatic Filariasis ( LF ) around the world ., In regions where the disease is endemic , local production and distribution of medicated salt dosed with diethylcarbamazine ( DEC ) has been an effective method for eradicating LF ., A partner of the Notre Dame Haiti program , Group SPES in Port-au-Prince , Haiti , produces a medicated salt called Bon Sel ., Coarse salt is pre-washed and sprayed with a solution of DEC citrate and potassium iodate ., Iodine levels are routinely monitored on site by a titrimetric method ., However , the factory had no method for monitoring DEC ., Critical analytical issues include, 1 ) determining whether the amount of DEC in each lot of Bon Sel is within safe and therapeutically useful limits ,, 2 ) monitoring variability within and between production runs , and, 3 ) determining the effect of a common local practice ( washing salt before use ) on the availability of DEC ., This paper describes a novel titrimetric method for analysis of DEC citrate in medicated salt ., The analysis needs no electrical power and requires only a balance , volumetric glassware , and burets that most salt production programs have on hand for monitoring iodine levels ., The staff of the factory used this analysis method on site to detect underloading of DEC on the salt by their sprayer and to test a process change that fixed the problem .

As researchers develop more sophisticated technologies , parts of the world are left behind ., The front lines of fighting many diseases lie in regions where expensive technology is not feasible ., As part of the effort to eradicate lymphatic filariasis in Haiti , our groups goal was to design an assay that would allow a chemist , with basic equipment , to quantify the levels of diethylcarbamazine citrate on medicated salt ., With access to university research facilities , we were able to devise and test a back-titration procedure that can measure the medication levels with sufficient accuracy and precision ., Our method capitalized on the fact that the medication is acidic ., This characteristic allows us to combine an unknown , medicated salt sample with a known quantity of base and then back-titrate with acid to determine diethylcarbamazine citrate concentration based on the neutralization point ., Developing this protocol has put the power of quality control into the hands of the Haitian factory producing the medicated salt ., With the ability to better monitor dosing levels , we have increased the effectiveness of this program in Haiti ., Using modern research facilities to produce effective , low-tech methods could be a useful approach for tackling many worldwide medical and environmental issues .

medicine, infectious diseases, analytical chemistry, filariasis, global health, neglected tropical diseases, chemistry, chemical analysis, parasitic diseases

journal.pcbi.1005638

Inherent limitations of probabilistic models for protein-DNA binding specificity

The study of protein-DNA interactions has a long history and includes binding to both single- and double-stranded DNA and both non-specific and sequence-specific interactions 1 ., Our interests are primarily in the sequence-specific interactions of transcription factors ( TFs ) that bind to DNA to regulate gene expression ., Detailed modeling of the in vivo interactions of TFs with genomic DNA that control gene expression requires accounting for many complicating factors , including competition and cooperativity with other TFs , competition with nucleosomes for occupancy of specific DNA regions and how sequences flanking the TF binding sites can affect occupancy 2–11 ., Regardless of the modeling approach , one component is always the specificity of the TF , how its binding affinity varies depending on the DNA sequence of the binding site ., Representations of specificity typically employ matrix-based models where the positions within the binding site are assumed to contribute independently to the TF’s binding affinity 12–14 ., In various methods the elements of the matrix may represent probabilities ( or log-probabilities ) of each base occurring at each position , or energetic contributions from each base at each position , or more generally just abstract scores that are related to the functional contributions of each base at each position 13 ., The data used to estimate the matrix parameters may come from many types of either in vivo or in vitro experiments , employing various types of algorithms 13 , 15–38 ., To determine the intrinsic specificity of a TF , how its binding affinity varies between different sequences , in vitro methods are preferred because the data are unconfounded by the complications that exist in vivo ., In this paper , we compare two methods of matrix representation of TF specificity , a probabilistic model in which the matrix elements are probabilities of each base at each position , and a biophysical model in which the matrix elements are the binding energy contributions of each base at each position ., Either type of matrix could be obtained from the same types of data , and we show that there are inherent limitations of the probabilistic model that do not apply to the biophysical model , suggesting that energy matrices should be preferred in general ., Probabilistic models ( PMs ) for DNA binding proteins were initially introduced by Harr et al . for E . coli promoters that even treated variable length binding sites 39 ., Soon after , Staden converted to the use of log-probability to put the model into a weight matrix ( additive ) model , also including parameters for variable spacing 40 ., Schneider et al . drew connections between the probabilistic models and information theory and introduced the log-odds model that accounts for the background distribution of bases 41 and later introduced the popular logo graphical representation of specificity 42 ., The probabilistic model was also the basis of the earliest motif discovery algorithms 33 , 34 , 43 ., Since then there have been many different algorithms for motif modeling and discovery using probabilistic models ( reviewed in 12 , 13 , 15 , 24 , 44 ) ., Even earlier von Hippel introduced an energy-based model of protein-DNA interactions 14 ., At the time , there were almost no data on actual binding sites so the paper used first principles to describe the informational specificity required for functional regulatory sites ., The paper made simplifying assumptions such as the independence between positions and that every mismatch from the preferred sequence had the same energy difference ., The first assumption , of independent contributions , has proven to be a reasonably good approximation for most transcription factors , whereas differences in contributions of alternative bases at each position are now well known and form the basis of most specificity modeling approaches ., Berg and von Hippel derived an energy model that was identical to the probabilistic one under some simplifying assumptions and connections between the energy approach and the information theory models of specificity became clear 45–47 ., Hwa and colleagues put the energy modeling approach into a more general biophysical model that accounts for the effects of protein concentration on binding probabilities 48 , 49 ., Djordjevic et al . pointed out the importance of the biophysical approach in modeling specificity 50 ., They further provided an algorithm that is guaranteed , for any collection of known binding sites , to predict the minimum number of additional sites in a genome , thereby minimizing the number of false positive predictions , although the method is not guaranteed to provide a more accurate model of the true specificity 50 , 51 ., Regression methods have been used to find optimal energy parameters and Foat et al . provided the first regression algorithm for motif discovery of optimal energy models 16 , 52 ., Since then several related methods have been developed to determine biophysical ( energy ) models of protein specificity from various types of high-throughput experimental data 18–20 , 25 , 27 , 28 , 31 , 37 , 38 , 53–55 ., Despite the development of several high-throughput experimental methods for measuring the specificity of protein-DNA interactions 22 , 56 and the algorithms described above for modeling them with the biophysical approach , probabilistic models remain the most popular ., The purpose of this report is to point out that when good energy models are available there is no advantage to using the probabilistic models ., In fact , due to inherent limitations the probabilistic models can be misleading and are highly sensitive to the samples used for inference of the parameters ., Energy models can be readily obtained and can easily accommodate non-independent contributions between positions 13 , 52 , 57 ., We conclude that energy modeling should become the approach generally used for modeling specificity and predicting protein-DNA interactions ., This model is based on a probability matrix PM ( b , j ) for each base b ∈ ( A , C , G , T ) at each position j = 1 , 2 , ⋯ , m for an m-long binding site ., Any particular DNA sequence Si can be encoded as a similar matrix , Si ( b , j ) , of 1s and 0s , where a 1 represents the base that occurs at position j and all other elements are 0 13 ., From the model , the probability of the sequence Si being among the bound sites is:, P˜ ( Si|B ) =∏j=1m∏b=ATPM ( b , j ) Si ( b , j ), ( 1 ), Often this is converted to a log-odds weight matrix WM ( b , j ) = logPM ( b , j ) /P ( b ) , where P ( b ) is the background , or prior , probability of base b 13 , 41 ., For simplicity , we assume the prior probability is a constant , 0 . 25 for each base , and therefore the two approaches give equivalent results ., Importantly , this is the probability of observing the sequence Si given a binding site , whereas what is desired is usually the probability that a sequence Si is bound , P ( B|Si ) ., Instead , searching sequences with probabilistic models generally just provides a list of sites within some probability range of the preferred site , including a predicted relative probability for each sequence ., This model is based on the thermodynamics of the interaction between two molecules , the protein T and a binding site Si ( additional details provided in S1 Supporting Information ) ., The association constant , which we refer to as the affinity , can be determined by measuring the concentrations of free reactants ( protein and DNA ) and of the complex:, KA ( T , Si ) =T⋅SiTSi≡Ki, ( 2 ), It is common to assume that the positions contribute independently to the binding affinity , just as the probabilistic model assumes the positions contribute independently to the site probability ., This is represented as a matrix of affinity contributions K ( b , j ) such that, Ki=∏j=1m∏b=ATK ( b , j ) Si ( b , j ), ( 3 ), From that one can determine the probability of a sequence Si being bound based on the protein concentration ( or really the chemical potential of the protein which is related to its free concentration ) and the association constant Ki:, P ( B|Si ) =T⋅SiT⋅Si+Si=KiTKiT+1=11+eEi−μ, ( 4 ), where Ei = −ln Ki is the free energy of binding to sequence Si and μ = lnTF is the chemical potential ., The probability of sequence Si in the bound sequences is obtained by Bayes’ rule and is dependent on the chemical potential , which differs from the probabilistic approximation ( noted above with P˜ ) :, P ( Si|B ) =P ( B|Si ) P ( Si ) P ( B ) ∝KiTKiT+1=11+eEi−μ, ( 5 ), if P ( Si ) is the same for all Si ., More importantly the true probability of sequence Si in the bound sequences has a non-linear relationship with its binding affinity ., This becomes pronounced at high protein concentrations where the energy can be additive across the positions of the binding site and yet the probabilities of the bases at each position are not independent ., From Eqs ( 1 ) and ( 5 ) , it is clear that probabilistic models of protein binding specificity provide approximations to true binding probabilities , P˜ ( Si|B ) ≈P ( Si|B ) ., We used simulations ( see the details in the Methods section ) to measure the accuracy of the approximation under various values of the chemical potential and for different methods of estimating the PM from the observed binding sites ., Of particular interest is how well the rank order of binding site probabilities is preserved ., At different protein concentrations , the PMs derived from binding probabilities are usually different ., As shown in Eq ( 4 ) , the binding probability of a sequence Si depends on both its binding affinity Ki ( or energy Ei ) and the protein concentration T ( or chemical potential μ ) ., If KiT ≪ 1 , there is a linear relationship between affinity and probability , but that occurs only when P ( B│Si ) ≪ 0 . 5 , which is unlikely to be the case in vivo for true regulatory sites ., At high protein concentrations , where KiT > 1 and the preferred binding site is highly occupied , the non-linear relationship between binding probability and affinity has several consequences ., One is that the PM itself depends on the protein concentration , whereas the binding energy does not ., Fig 1a and 1b show one example from the simulation of an energy matrix and its associated energy logo 13 , 16 ., Note that in the matrix the lowest energy base at each position is assigned energy 0 ( using the convention of Berg and von Hippel 45 ) , and in the logo the average energy for each position is set to 0 , with the lower energy ( higher affinity ) bases on top ., Since only the differences in energy matter for relative binding affinities , both representations lead to the same results ., Fig 1c and 1d show the information logo 42 and the PM for that protein obtained at very low protein concentration , μ = −3 ., At low μ the PM corresponds very closely to the independent contributions of each base to the binding affinity ( Eq ( 5 ) converges to Eq ( 1 ) ) ., But at high protein concentration , such as μ = 3 , the logo and PM are different ( Fig 1e and 1f ) ., The second logo shows that the information is “compressed” at μ = 3 , with the mean column information content ( MCIC ) decreasing from 0 . 9 bits to 0 . 7 bits ., But the change in probability is not evenly distributed ., Comparing the two PMs ( Fig 1d and 1f ) , at high protein concentration the base probabilities tend to move toward the mean , 0 . 25; the high probability bases decrease in probability and the low probability bases increase ., But each position is normalized independently so that the magnitude of the change varies from position to position ., For each position , the rank order of probability for the bases remains unchanged , but because the positions are normalized independently , the probabilities of different binding sites may change rank order ., That is , even though the binding energy is completely additive across the positions , the probabilities of bases do not factor accurately across the positions ., The rank correlation ( the square of the Spearman’s rank correlation coefficient ) between the predicted and true all-sequence distributions depends on the protein concentration and how the PM is computed ., Table 1 shows the mean values and standard deviations of r2 for 100 simulations of 8-long binding sites with μ = −3 , 0 and 3 ( which correspond to the preferred sequence being bound at 0 . 05 , 0 . 5 and 0 . 95 probability , respectively ) ., The rank correlation is shown for the complete distribution of binding sites based on PMs generated from the full distribution of binding data and from just the top 1% of sites , either weighted or unweighted ., At μ = −3 there is a nearly perfect fit to the true ranking when the PM is derived from the entire distribution ., However , when it is based on the top 1% of sites , the ranking is slightly less accurate ( 0 . 994 ) when the sites are weighted by their true probability ., In both of those cases the PM provides a very good approximation to the true ranking of binding sites ., If the top 1% are used unweighted to make the PM , the fit to the true ranking is 0 . 984 and that is true regardless of the value of μ because the top 1% of sites is the same and their probabilities are ignored ., When μ = 0 , the results are very similar ., For μ = 3 the rank correlation drops to 0 . 988 when the weighted top 1% of sites are used to obtain the PM ., Fig 2 plots the logarithms of the predicted and true relative binding probabilities for the case of the protein of Fig 1 and μ = 3 ., In each case the overall fit is quite good but the width of the plots indicates some degree of mis-ranking of the binding sites ., While the overall rankings are quite good , it is the highest affinity sites that are of primary interest ., In fact , all DNA-binding proteins exhibit a non-specific binding affinity 49 such that there is a minimum binding affinity below which the sequence no longer matters ., In addition , functional regulatory sites must have sufficient occupancy to fulfill their roles , so only sites within some restricted range of the optimum are likely to be functional ., Table 2 shows the rank correlations for the same PMs used in Table 1 , but now focusing on the 1% highest affinity binding sites ., Note that the values in Table 2 are all lower than for Table 1 , indicating that the accuracy is lower when the PM is used to predict the highest affinity sites ., Fig 3 shows a subset of the plots in Fig 2 , including only the top 1% of sites ., For μ = 3 the rank correlation drops to 0 . 970 even when the entire distribution is used to generate the PM and the scatter of the points shows that there are clear differences between the true and predicted rankings ., In fact , the true top 1% is not precisely equivalent to the predicted top 1% of sites ., When the top 1% of sites are used to generate the PM , weighted by their probability , the rank correlations drop substantially for all values of μ , but especially for μ = 3 where it is only 0 . 876 ., If the unweighted top 1% are used for the PM , the rank correlation drops to 0 . 840 for all values of μ ., The plots in Fig 3 all show substantial mis-ranking of sites ., The results in Tables 1 and 2 show that the quality of PMs , their ability to correctly rank binding sites , vary widely depending on both the protein concentration at which the binding data was obtained and the set of binding sites used to derive the PM ., The effect of the protein concentration is most evident at high values of μ where the non-linearity of Eq ( 4 ) is largest and non-independence of the position probabilities is most pronounced ., The effect of site sampling is due to the sensitivity of the PM to the exact set of example sites used ., Another consequence of the non-linear relationship between binding affinity and probability is that pairs ( and higher order combinations ) of positions have non-independent effects on binding probability , even though the contributions to binding affinity are completely independent ., We show this with one example in Fig 4 based on the protein with energy matrix shown in Fig 1 at μ = 3 ., When the preferred binding site , TGGTAACG with binding probability of 0 . 95 , is mutated to TGGTAAAG , the binding probability decreases to 0 . 79 , about a 17% decrease in binding probability ., If the same C to A mutation occurs in another sequence , TGGCAACG to TGGCAAAG , the binding probability decreases from 0 . 62 to 0 . 23 , a 63% decrease ., This apparent non-independence , where the effect of the mutation varies depending its context , is an artifact of the PM because the change in binding affinity ( 1 . 69 kT , Fig 1b ) is completely independent of context ., In the preceding simulations , we compared the probabilistic model to the true biophysical model for the specificity of TFs ., However , in real experimental data the observed probabilities of binding to specific sites will include noise that will affect both the probabilistic and biophysical models ., To measure the effects of noise on the accuracy of rank predictions , we included errors in the observed probabilities from which both the probabilistic and biophysical matrices were obtained ( details in S1 Supporting Information ) ., We added random noise , with mean of 0 and standard deviation of 0 . 5 kT , to the energy of each sequence prior to generating its probability ., That amount of noise is larger than what can be obtained with methods such as Spec-seq , where we typically get standard deviationsof about 0 . 2 kT , at least for the high affinity sequences 54 , 58 ., The results for the probabilistic models are slightly worse than those without noise , described above ., For the biophysical models the fits are much better , decreasing to 0 . 97 when only the top 1% of sites are used to estimate the parameters ( see S1 Supporting Information ) ., This is because the noise is added to each sequence independently , whereas the models include the parameters for each base at each position , which are averaged over all of the sequences containing those bases ., One advantage of low-dimensional models , such as all types of positional matrices , is that experimental noise is averaged out ., In fact , even if the true interaction is not precisely independent between the positions , by averaging over all of the contexts one can obtain models that are good representations of the true specificity ., Probabilistic models of protein-DNA interactions are commonly used because they are easy to obtain and they provide an intuitive representation of specificity ., However , they do not provide the information usually desired , the probability that a specific sequence is bound , P ( B│Si ) , but rather an approximation to the probability of observing a specific sequence given a binding site , P˜ ( Si|B ) ., From that one can obtain a predicted rank order of all possible binding sites and , if one assumes a specific probability , or occupancy , for the preferred sequence , the predicted probabilities for all other sequences ., To obtain binding probabilities from the biophysical model one needs to know the chemical potential , but just as with the probabilistic model if one assumes the probability , or occupancy , of the preferred sequence , then the probabilities of all other sequences can be obtained from the model ., Since both models really return the same information , a predicted ranked list of binding sites and relative binding probabilities , they should be judged on the accuracy of those predictions and the ease of obtaining the model parameters ., The accuracy of PMs is limited by availability of binding site affinity data ., When a PM is based on the entire probability distribution of binding sites it is a good approximation overall , even at high μ ., However , it does have discrepancies that include mis-ordering of the ranks of binding sites as well as the appearance of non-independence between positions that are in fact independent ., These effects are due to the intrinsic lack of proportionality between binding probability and binding affinity that is most problematic at high protein concentrations ., More severe defects occur due to incomplete information about the binding probability distribution ., Obtaining the full distribution of binding probabilities requires in vitro experiments , such as protein binding microarrays , HT-SELEX ( or SELEX-seq ) or other high-throughput methods 20 , 22 , 27 , 35 , 53 , 59–64 , but many algorithms utilize only the highest affinity binding sites ., PMs can be derived from in vivo data of binding site locations , and have the advantage of being easily derived from such data using many motif discovery algorithms 13 , 24 , 32–34 , 43 , 65 ., But in those cases , the PM is derived from only a fraction of the binding sites ., Functional regulatory sites will be among the high affinity sequences and in ChIP-seq experiments the peaks will also tend to contain the highest affinity sites ., And if the sample size is small , those sites are not even weighted by their binding probabilities ., In addition , confounding factors occurring in vivo , such as competition and cooperativity with other proteins , lead to incomplete information about the probability distribution and that causes further inaccuracies in the PMs ., Good binding models are still important after the advent of high-throughput methods and their parameters can be readily determined by using appropriate algorithms ., Binding affinities to small numbers of sequences can be obtained with arbitrarily high accuracy using a variety of experimental techniques 66 ., If the additivity ( positional independence ) assumption is valid , the relative affinities , compared to the preferred sequence , of only the 3m single nucleotide variants are needed for the full energy model ., Of course , additivity is unlikely to be completely accurate , but there are still only 3m + 9 ( m − 1 ) single variants plus double variants at adjacent positions , where the non-additivity is likely to be most prevalent ., But multiple high-throughput methods are now available that provide quantitative binding data from which accurate energy models can be obtained by using appropriate algorithms 16 , 18–20 , 25 , 27 , 28 , 31 , 37 , 38 , 53 , 54 , 62 ., From sufficiently abundant and accurate quantitative binding data one can even skip the modeling and just use the list of relative binding energies to all possible sites ( or at least the highest affinity sites that are likely to function as regulatory sites ) , avoiding approximations entirely ( to the degree allowed by the measurement accuracy ) ., However , models are still useful because they provide a compact representation of specificity , usefully visualized with logos 13 , 16 , 42 , and can provide insight into the mechanisms of binding specificity , such as the contribution of DNA structure to binding specificity 67 , 68 ., It is also important to have good specificity models obtained from in vitro binding experiments to compare to data obtained in vivo ., This allows one to identify cases where interacting TFs alter the specificity of individual TFs , which one can only infer by having good models for each TF alone 69–71 ., We conclude by pointing out that when accurate energy models are available for DNA binding specificity there is no advantage to using probabilistic models , and in fact they can be misleading and provide inaccurate predictions ., There are now good high-throughput methods for measuring relative binding affinities to very large collections of sites and good algorithms for determining accurate energy models ., We propose that such models become the standard approach for representing specificity and predicting binding sites in vivo ., We developed a program , BEnDS ( Binding Energy Distribution Simulations ) , to generate random energy matrices of a user-specified length , m ., One base is randomly chosen as the preferred base at each position and assigned an energy of 0 ., Energies for the other bases are drawn randomly from a normal distribution with a user-specified mean and standard deviation ( with default values of N ( μ = 2 . 5 , σ = 1 . 0 ) ) ., We assume perfect additivity of binding energies so that the energy for any sequence is the sum of the energies for its bases at each position ( equivalent to Eq ( 3 ) ) ., This model , implemented in different programs , achieved the best overall performance in a test of various programs on modeling the specificity of DNA-binding proteins based on protein binding microarray ( PBM ) data 25 ., Given an energy matrix , probabilities of binding to all possible sites are obtained using Eq ( 4 ) for various ( user-specified ) values of μ ., For every set of site probabilities , PMs were determined ., This was done both for the entire distribution and from a subset of high affinity sites , such as the top 1% ( as might be expected to be functional sites ) ., When only the top 1% of sites are used , PMs from the sites could be obtained either weighted by their probabilities , or just from the list of sites unweighted , as one might expect from a collection of known regulatory sites or from ChIP-seq type of experiment with a limited sample of observed binding sites ., Simulations that include noise in the energies , and therefore the probabilities , of each sequence are described in S1 Supporting Information ., In those cases the energy matrix is obtained using non-linear regression on the site probabilities , similar to BEESEM 37 but without the need to infer the binding site position ., The probabilistic model does not attempt to report the probability that a site is bound , P ( B│Si ) , it only reports the predicted relative probability , and therefore the rank order , of different sites being bound ., We compare the true rank order of the sites from their binding energies to the predicted rank order based on the PM at different protein concentrations ., We report the square of the Spearman’s rank correlation coefficient , r2 .

Introduction, Results, Discussion, Methods

The specificities of transcription factors are most commonly represented with probabilistic models ., These models provide a probability for each base occurring at each position within the binding site and the positions are assumed to contribute independently ., The model is simple and intuitive and is the basis for many motif discovery algorithms ., However , the model also has inherent limitations that prevent it from accurately representing true binding probabilities , especially for the highest affinity sites under conditions of high protein concentration ., The limitations are not due to the assumption of independence between positions but rather are caused by the non-linear relationship between binding affinity and binding probability and the fact that independent normalization at each position skews the site probabilities ., Generally probabilistic models are reasonably good approximations , but new high-throughput methods allow for biophysical models with increased accuracy that should be used whenever possible .

Transcription factors ( TFs ) , a class of DNA-binding proteins , play a central role in the regulation of gene expression ., TFs control the rate of transcription by binding to the genome in a sequence-specific manner ., Thus , one important aspect in the study of gene regulation mechanism is to model the binding specificities of TFs , namely the features of the DNA sequences that a TF prefers to bind ., Multiple models have been proposed to characterize the binding specificities of TFs , among which the class of probabilistic models is the most popular ., In this study , we point out several major limitations of the well-established probabilistic model by comparing it with the biophysical model ., Through simulations we demonstrate that the probabilistic model is only an approximation of the biophysical model ., The latter has most of the advantages of the former , and is a more accurate representation of binding specificities ., We propose a shift from the probabilistic model to the biophysical model in future studies of protein-DNA interactions .

protein interactions, applied mathematics, dna-binding proteins, simulation and modeling, algorithms, mathematics, sequence motif analysis, research and analysis methods, sequence analysis, bioinformatics, proteins, biophysics, approximation methods, physics, biochemistry, biochemical simulations, database and informatics methods, biology and life sciences, physical sciences, computational biology

journal.pcbi.1006370

Dorsal anterior cingulate-brainstem ensemble as a reinforcement meta-learner

In the next two subsections of the Introduction we describe qualitatively the computational principles of the RML ( The RML: General description ) and the main novelties introduced by the model ( The RML: Innovations ) ., In the subsequent Results section , we describe the experimental paradigms we used to test the RML and the results , together with domain-specific discussion paragraphs ., Next , in the domain-general Discussion section we broadly frame and connect the results , comparing our model with other models from recent literature ( Relationships to Other Models ) ., We also propose future experimental paradigm to test RML predictions ( Experimental Predictions ) , including possible applications to translational research , and we describe some limitations of our work ( Limitations ) ., Finally , in the Methods section we provide the full mathematical description of the RML ., At the basis of our model is the idea that a macrocircuit involving dACC-VTA-LC represents a core computational unit for optimizing both behaviour and internal states that modulate behaviour itself ( meta-learning ) ., Fig 1 represents an overview of the RML architecture ., The RML dynamics is based on two inter-related loops connecting four computational modules: dACCBoost , dACCAct , VTA , and LC ., An external loop represents the interaction between the dACC modules and the environment , while an internal loop covers the interaction between the dACC modules and the brainstem nuclei ( VTA and LC; orange and red bidirectional arrows in Fig 1 ) ., This double loop structure is aimed at optimizing performance ( i . e . , maximizing reward ) while minimizing two different types of costs: the costs of motor actions ( external loop; e . g . the metabolic cost of climbing a stair ) , and the boosting costs of neuromodulators release ( internal loop; e . g . the cost of neurotransmitters depletion ) ., Connectivity and functional studies corroborate the hypothesis underlying this architecture , because they show that there is an anatomical overlap between the midfrontal sub-region related to the meta-learning processes discussed above , and the midfrontal sub-region maximally connected with both LC and VTA nuclei 13 , 18–21 , both located within the dACC area ., In the RML , the dACC plays the role of a performance monitoring system , which compares expectations about environmental states and executed actions with environmental outcomes ( cf . 22 ) ., Discrepancies between expectations and outcomes generate prediction error ( PE ) signals ( Figure G in S2 File ) , which are used to update the expectations themselves 3 ., This monitoring process is at the basis of two dACC modules , and in substantial agreement with the experimental literature ( see 3 for a review ) ., One dACC module ( dACCAct in Fig 1 ) receives environmental states and selects actions directed toward the external environment ( part of the external loop ) ., Although value-based action selection involves also other subcortical and cortical structures ( e . g . the dorsolateral prefrontal cortex , DLPFC ) , here we frame both value estimation and action selection within the dACC for both modeling parsimony and because also the MFC , with its motor components , plays an important role in action selection ( see 3 for a review ) ., A second dACC module ( dACCBoost in Fig 1 ) receives environmental states and consequently modulates ( that is , boosts ) the release of catecholamines from the brainstem nuclei LC and VTA ( part of the internal loop ) ., Catecholamines , in turn , control the internal dynamics of the dACC in real time ( i . e . while the RML is interacting with the environment ) , by modulating the magnitude of reward signals ( by VTA module ) and the amount of effort ( by LC module ) that the RML exerts to execute a task ., Although the dACCBoost module is the main responsible for catecholaminergic modulation , the dACCAct module , too , is in recurrent interaction with the brainstem nuclei , providing the VTA with a reward prediction signal ., The latter is used by the VTA to compute non-primary rewards , which are sent back to the dACCAct ( like in a TD-learning algorithm; 23 ) allowing the system to learn complex tasks without the immediate availability of primary rewards ( higher-order conditioning ) ., Importantly , both dACC modules have dynamic learning rates ( λ ) , ensuring that knowledge is updated only when there are relevant environmental changes ( volatility ) ., Learning rate adaptation emerges from the interaction between the LC and both dACC modules ., Each dACC module feeds the LC with reward prediction and PE signals , while the LC analyzes these “raw data” from the cortex ( approximating a Bayesian learner ) , estimating volatility and adjusting the modules’ learning rate as a consequence ., Finally , the RML can be connected to other neural models ( e . g . a visuo-spatial working memory model , see Simulation 2c ) ., This allows the effort-related signal from the LC to modulate processing in other brain areas for performance optimization ( Fig 1 , orange arrows; see Methods for details ) ., In this section we briefly introduce the main theoretical novelties of the RML ., For a more detailed analysis we address the reader to the Discussion section , where we also relate the RML in detail to previous models , describe explicit experimental predictions that derive from the model , and speculate on the potential application of the RML to translational research ., The RML is an autonomous agent able to near-optimally adapt to a diverse range of environments and tasks , with no need of task-specific parameters setting: across all the reported simulations the RML autonomously controlled its internal dynamics as a function of the environmental challenges , with no offline parameters optimization or human intervention ( i . e . one parameter set was used for all the simulations ) ., From here , four major novelties can be identified ., Adaptive control of learning rate is a fundamental aspect of cognition ., Humans can solve the tradeoff between stability and plasticity in a ( near ) Bayesian fashion 4 , 29 , distinguishing between variability due to noise versus variability due to actual changes of the environment; thus they can increase the learning rate only when volatility is detected 30 , 31 ., At the neural level , a currently unexplained dissociation exists between dACC and LC activity , recorded during decision-making tasks where uncertainty due to noise and uncertainty due to volatility were systematically manipulated ., The LC activity ( and thus NE release ) has been shown to track specifically volatility 30 , 32 , 33 , while the results about the dACC role in volatility estimation are less consistent ., Indeed , while in the seminal study by Behrens et al . 4 , the dACC was hypothesized to track volatility , more recent study suggested that dACC activity in volatile environments are driven rather by PE coding , rather than specifically by volatility estimation 21 ., These empirical findings seem to attribute different roles to LC and dACC in uncertainty coding , without providing a computational rationale for their functional specialization ., In this simulation , we will investigate to what extent the model accounts for human adaptive control of learning rate at both behavioural and neural levels , and whether it can explain the dACC/LC dissociation ., A long list of experimental results indicates that DA and NE neuromodulators are not only crucial for learning environmental regularities , but also for exerting cognitive control 37–41 ., Although these mechanisms have been widely studied , little is known about how the brainstem catecholamine output is controlled to maximize performance 31 , 42 , 43 , and how the dACC is involved in such a process ., In this section , we describe how the dACCBoost module learns to regulate LC and VTA activity to control effort exertion , at both cognitive and physical level 19 , 44 , 45 ., In Simulation 2a , we test the cortical-subcortical dynamics in experimental paradigms involving decision-making in physically effortful tasks , where cost/benefit trade off must be optimized 46–48 ., In Simulation 2b , we show how the LC can provide a NE signal to external neural modules to optimize cognitive effort 19 , 20 allocation and thus behavioural performance in a visuo-spatial working memory ( WM ) task ., In both simulations , we also test the RML dynamics and behaviour after cortical and subcortical lesions ., Deciding how much effort to invest to obtain a reward is crucial for human and non-human animals ., Animals can choose high effort-high reward options when reward is sufficiently high 46 , 47 ., The impairment of the mesolimbic DA system strongly disrupts such decision-making 46 , 47 ., Besides the VTA , experimental data indicate also the dACC as having a pivotal role in decision-making in this domain 19 , 20 , 48–50 ( see also51 for a review ) ., In this simulation , we show how cortical-subcortical interactions between the dACC , VTA and LC can drive optimal decision-making when effortful choices leading to large rewards compete with low effort choices leading to smaller rewards ., We thus test whether the RML can account for both behavioral and physiological experimental data from humans and nonhuman animals ., Moreover , we test whether simulated ACC lesion or DA depletion can replicate the disruption of optimal decision-making , and , finally , how effective behaviour can be restored ., Simulation results will be compared with behavioural data from rodents ( 47 , see also Simulation 2a in S1 File ) , and with physiological data from nonhuman primates 35 and humans 44 ., Rodent data from Walton et al . 47 were chosen for comparison to study how the cost-benefit trade-off could be affected by ACC damage and by DA lesion and how behavioural performance could be partially recovered with environmental intervention ( Simulation 2b ) ., We express the caveat that DA depletion studies in the literature we cited ( 46 , 47 , to compare with RML performance ) either deplete DA systemically , or are focused more on the mesolimbic-accumbens path than on DA afferents to the medial prefrontal cortex ., Our assumption that mesolimbic DA lesion affects dACC functioning is neurophysiologically sound , because functional and anatomical connectivity indicates strong nucleus accumbens ( NAc ) —dACC connectivity 12 , 13 , 52 , 53 , probably contributing to convey reward-related information to the dACC ., For this reason , lesioning the NAc may also disrupt the information flow from VTA to the dACC ., Moreover , our simulations lead to the experimental prediction that DA lesion to dACC generates effects similar to mesolimbic DA lesions ., In DA lesioned subjects , the preference for HR option can be restored by removing the difference in effort between the two options 47 , that is , by removing the critical trade-off between costs and benefits ., In Simulation 2b , we show how the RML can recover a preference toward HR options , as demonstrated empirically in experimental paradigms used in rats ., We focused specifically on recovery after DA lesion ., Our choice was aimed at investigating the consequences of DA lesion at cortical-subcortical level and how these can be modulated by the environment , to open a view on future translational scenarios about DA-related neuropsychiatric disorders ., We elaborate on the latter topic in the Experimental Predictions section ., NE neuromodulation also plays a crucial role in WM , improving signal-to-noise ratio by gain modulation mediated by α2-A adrenoceptors 37 , 56 ., A low level of NE transmission leads to WM impairment 57 , 58 ., At the same time , as described above , NE is a major biological marker of effort exertion 35 , 59 ., Besides NE release by the LC , experimental findings showed that also dACC activity increases as a function of effort in WM tasks 19 , 20 , 60 ., Here we show that the same machinery that allows optimal physical effort exertion ( Simulation 2a ) may be responsible for optimal catecholamine management to control the activity of other brain areas , thus rooting physical and cognitive effort exertion in a common decision-making mechanism ., This is possible because the design of the RML allows easy interfacing with external modules ( Fig 1 and Methods ) ., Animal behavior in the real world is seldom motivated by conditioned stimuli directly leading to primary rewards ., Instead , behavior is guided by higher-order conditioning , bridging the gap between reward and behavior ., However , a unifying account explaining behavioral results and underlying neurophysiological dynamics of higher-order conditioning is currently lacking ., First , at the behavioral level , literature suggests a sharp distinction between higher-order conditioning in classical versus instrumental paradigms ., Indeed , although it is possible to train animals to execute complex chains of actions to obtain a reward ( instrumental higher-order conditioning , 62 ) , it is impossible to install a third- or higher-order level of classical conditioning ( i . e . when no action is required to get a reward 63 ) ., Although the discrepancy has been well known for decades , its reason has not been resolved ., Second , a number of models have considered how TD signals can support conditioning and learning more generally 64 , 65 ., However , no model addressing DA temporal dynamics also simulated higher-order conditioning at behavioural level ., Here we use the RML to provide a unified theory to account for learning in classical and instrumental conditioning ., We show how the RML can closely simulate the DA shifting in classical conditioning ( Simulation S2 and Fig F in S2 File ) ., We also describe how the VTA-dACC interaction allows the model to emancipate itself from primary rewards ( higher-order conditioning ) ., Finally , we investigate how the synergy between the VTA-dACCBoost and LC-dACCBoost ( the catecholamines boosting dynamics ) is necessary for obtaining higher-order instrumental conditioning and how this process could be considered one of the foundations of intrinsic motivation ., This provides a mechanistic theory on why higher-order conditioning is possible only in instrumental and not in classical conditioning ., As VTA can vigorously respond to conditioned stimuli , it is natural to wonder whether a conditioned stimulus can work as a reward itself , allowing to build a chain of progressively higher-order conditioning ( i . e . not directly dependent on primary reward ) ., However , for unknown reasons , classical higher-order conditioning is probably impossible to obtain in animal paradigms 63 , 66 ., We thus investigate what happens in the model in such a paradigm ., Differently from classical conditioning paradigms , animal learning studies report that in instrumental conditioning it is possible to train complex action chains using conditioned stimuli ( environmental cues ) as reward proxies , delivering primary reward only at the end of the task 62 ., The flexibility of RML , and the explicit neurophysiological hypotheses on which it is based , allow several experimental predictions ., In this paper we aimed at presenting the general potential and the theoretical value of the RML , comparing , in a qualitative fashion , the results from our simulations with experimental data from many different domains ., A larger use of quantitative approaches to test the experimental predictions derivable from the RML ( e . g . model-based data analysis ) will be necessary in future work ., Here we list some potential experiments deriving from RML predictions ., The first three are sufficiently specific to potentially falsify the model ( at least in its neurophysiological interpretation ) , the others are currently formulated as working hypotheses ., First , the RML architecture suggests that PE signals are generated by the dACC and then converge toward the brainstem nuclei ., This hypothesis implies that dACC lesion disrupts DA dynamics in higher-order conditioning , with a consequent impairment in higher-order instrumental conditioning; further , dACC lesion should disrupt LC dynamics related to learning rate control , with a consequent impairment of behavioural flexibility optimization ., A second prediction concerns the mechanisms subtending higher-order conditioning and the difference between classical and instrumental paradigms ., In the RML , higher-order conditioning is possible only when the agent plays an active role in learning ( i . e . , instrumental conditioning ) ., We predict that hijacking the dACC decision of boosting catecholamines ( e . g . , via optogenetic intervention ) would make possible higher-order conditioning in classical conditioning paradigms ( ref . simulations 3a-b ) ., Third , the DA-lesioned RML shows stronger dACC activation during an easy task ( without effort ) in the presence of a high reward ( see Simulation 2a , Fig 4B ) ., This finding can be interpreted as a compensatory phenomenon allowing to avoid apathy ( i . e . refusal to engage in the task ) if a small effort can make available a big reward ., This is an explicit experimental prediction that could be tested both in animal paradigms and in mesolimbic DA impaired humans 76 , or in patients with Parkinson’s disease on and off medication 51 , therefore providing also possible translational implications ., Fourth , as shown above , the model provides a promising platform for investigating the pathogenesis of several psychiatric disorders ., In a previous computational work , we proposed how motivational and decision-making problems in attention-deficit/hyperactivity disorder ( ADHD ) could originate from disrupted DA signals to the dACC 77 ., In the current paper , we also simulated a deficit related to cognitive effort ( Simulation 2c ) in case of DA deficit ., Together , these findings suggest how DA deficit can cause both motivational and cognitive impairment in ADHD , with an explicit prediction on how DA deficit can impair also NE dynamics 78 in ADHD ., This prediction could be tested by measuring performance and LC activation during decision-making or working memory tasks , while specifically modulating DA transmission in both patients ( via pharmacological manipulation ) and RML ., Fifth , another clinical application concerns a recent theory on autism spectrum disorder ( ASD ) pathogenesis ., Recent studies 79 , 80 proposed that a substantial number of ASD symptoms could be explained by dysfunctional control of learning rate and overestimation of environment volatility ., This qualitative hypothesis could be easily implemented and explored quantitatively by altering meta-learning mechanisms in the RML leading to chronically high learning rate and LC activation ., The RML framework has three main limitations ., First , in the RML DA plays a role only in learning ., As with any other neuromodulator , experimental results suggest a less clear-cut picture , with DA being involved also in performance directly ( e . g . attention and WM via DLPFC modulation ) 39 , 81–83 ., The goal of our simplified characterization of DA function was to elucidate how the two neuromodulators can influence each other for learning ( DA ) and performance ( NE ) ., Moreover , other theories stress the importance of direct ( and hierarchically organized ) interaction between the medial prefrontal cortex and the DLPFC in cognitive control 84 and WM function 68 ., From this perspective , reduced DA signal to the dACC could directly disrupt the dACC-DLPFC interaction , impairing cognitive control and WM without the involvement of the NE modulation ., dACC-DLPFC interaction is a neglected aspect in our model that should be investigated in future works ( see next section ) ., The second limitation is the separation of the LC functions of learning rate modulation ( λ ) and cognitive control exertion ., The cost of this separation between these two functions is outweighed by stable approximate optimal control of learning rate and catecholamines boosting policy ., It must be stressed that the ACCBoost module receives the LC signal λ related to learning rate in any case , making the boosting policy adaptive to environmental changes ., Third , the RML reacts to environmental changes by learning rate modulation , while human and nonhuman primates can use specific events that occurred ( episodic control 85 ) , to trigger policy change for adapting to novel situations ., There is also converging evidence that primate dACC ( and most likely its homologous area in rats ) is critical to perform this type of higher-order inference ( see 7 for a short review ) , and that LC bursts could work as circuit breakers to reset ongoing neural representations and trigger behavioural adaptation driven by episodic control 86 ., The lack of contribution by episodic knowledge in behavioural optimization is clearly a limitation of our model , especially if we consider that episodic control can also optimize motivational signals to modulate cognitive effort 84 ., We believe that these two adaptive processes ( i . e . learning rate control and episodic control ) are complementary and run in parallel and that their integration ( a possibly arbitration on influencing behaviour ) should receive future theoretical investigation ., The RML shows how meta-learning involving three interconnected neuro-cognitive domains can account for the flexibility of the mammalian brain ., However , our model is not meant to cover all aspects of meta-learning ., Many other decision-making dimensions may be optimized by meta-learned too ., One obvious candidate is the stochasticity ( temperature ) of the decision process 87 , which arbitrates the exploration/exploitation trade-off ., We recently proposed that this parameter is similarly meta-learned trading off effort costs versus rewards 6 ., It must be noted that experimental findings indicated a link between LC activation and the arbitration on exploration/exploitation trade-off 88 , 89 , suggesting that the same mechanism used for learning rate optimization could be extended also to this domain ., Other aspects from the classical RL modeling framework include discounting rate or eligibility traces 90; future work should investigate the computational and biological underpinnings of their optimization ., Moreover , considering the strong empirical evidence attributing to the dACC a prominent role in foraging ( e . g . 91 ) , future work should focus on how the RML can also face this class of problems , where it is studied not only how mammals optimize choices within a task , but also how they decide when it is convenient to switch to another task , to maximize reward in the long run ., Given the exceptionally extended dACC connectivity 12 , other brain areas are likely relevant for the implementation of decision making in more complex settings ., For example , we only considered model-free dynamics in RL and decision-making ., However , both humans and nonhuman animals can rely also on complex environment models to improve learning and decision making ( e . g . spatial maps for navigation or declarative rules about environment features ) ., In this respect , future work should particularly focus on dACC-DLPFC-hippocampus interactions 92 , 93 , in order to investigate how environment models can modulate reward expectations , how the nervous system can represent and learn decision tree navigation 94 and how reward expectations can modulate goal-directed DLPFC representations 84 ., Another anatomo-functional aspect that could be investigated concerns the anatomical segregation of the twofold dACC function we described here ( dACCAct and dACCBoost ) ., Although we remain agnostic about this question , it would be interesting to investigate whether the neural units performing these two types of decision-making operations are overlapping , intermixed , or even segregated in different dACC sectors ., Finally , the RML can work in continuous time and in the presence of noise ., These features are crucial to make a model survive outside the simplified environment of trial-level simulations , and allow simulating behaviour in the real world , like , for example , in robotic platforms ., RML embodiment into robotic platforms could be useful for both neuroscience and robotics ., Indeed , testing our model outside the simplified environment of computer simulations could reveal model weaknesses that are otherwise hidden ., Moreover , closing the loop between decision-making , body and environment 95 is important to have a complete theory on the biological and computational basis of decision-making ., At the same time , the RML could suggest new perspectives on natural-like flexibility in machine learning , helping , for example , in optimizing plasticity as a function of environmental changes ., RML architecture was implemented in two versions: a discrete model ( simulating inter-trial dynamics ) and a dynamical model ( a dynamical system simulating also intra-trial dynamics ) ., Both implementations share the same architecture displayed in Fig 1 , and follow the same computational principles ., All the results reported above were obtained with the dynamical model ., Here we introduce the mathematical form of the discrete model , which provides a clearer and more compact RML description ., All the simulations ( with exception of Simulation 2c , which requires intra-trial dynamics ) were replicated with the discrete model ( Figures S9-S12 in S2 File ) , demonstrating that the computational principles founding the RML are independent from specific implementations ., We used a single set of parameters across all simulations both for the discrete model ( Table 1 ) and for the dynamical model ( Table A in S1 File ) ., Parameters were hand-tuned to ensure acceptable performance in a simple 2-armed bandit task and second-order conditioning task ., The mathematical description of the dynamical model can be found in the S1 File ., We designed the model such that communication with the external environment is based on 9 channels ( Fig 9A ) ., Six channels represent environmental states ( s ) and RML actions ( a ) ( 3 states and 3 actions ) ., The first two actions are aimed at changing the environmental state ( e . g . turning right or left ) , while the 3rd action means “Stay” , i . e . refusing to engage in the task ., There are two other input channels , one dedicated to reward from environment ( RW ) and the other to signal costs of motor actions ( C ) ., Finally , there is one output channel conveying norepinephrine ( NE ) signals to other brain areas ., The RML is scalable by design , i . e . there is no theoretical limit to the number of state/action channels , and neither the number of parameters nor their values changes as a function of task type/complexity .

Introduction, Results, Discussion, Methods

Optimal decision-making is based on integrating information from several dimensions of decisional space ( e . g . , reward expectation , cost estimation , effort exertion ) ., Despite considerable empirical and theoretical efforts , the computational and neural bases of such multidimensional integration have remained largely elusive ., Here we propose that the current theoretical stalemate may be broken by considering the computational properties of a cortical-subcortical circuit involving the dorsal anterior cingulate cortex ( dACC ) and the brainstem neuromodulatory nuclei: ventral tegmental area ( VTA ) and locus coeruleus ( LC ) ., From this perspective , the dACC optimizes decisions about stimuli and actions , and using the same computational machinery , it also modulates cortical functions ( meta-learning ) , via neuromodulatory control ( VTA and LC ) ., We implemented this theory in a novel neuro-computational model–the Reinforcement Meta Learner ( RML ) ., We outline how the RML captures critical empirical findings from an unprecedented range of theoretical domains , and parsimoniously integrates various previous proposals on dACC functioning .

A major challenge for all organisms is selecting optimal behaviour to obtain resources while minimizing energetic and other expenses ., Evolution provided mammals with exceptional decision-making capabilities to face this challenge ., Even though neuroscientists have identified a heterogeneous and distributed set of brain structures to be involved , a comprehensive theory about the biological and computational basis of such decision-making is yet to be formulated ., We propose that the interaction between the medial prefrontal cortex ( a part of the frontal lobes ) and the subcortical nuclei releasing catecholaminergic neuromodulators will be key to such a theory ., We argue that this interaction allows both the selection of optimal behaviour and , more importantly , the optimal modulation of the very brain circuits that drive such behavioral selection ( i . e . , meta-learning ) ., We implemented this theory in a novel neuro-computational model , the Reinforcement Meta-Learner ( RML ) ., By means of computer simulations we showed that the RML provides a biological and computational account for a set of neuroscientific data with unprecedented scope , thereby suggesting a critical mechanism of decision-making in the mammalian brain .

learning, medicine and health sciences, neurochemistry, chemical compounds, decision making, brain, social sciences, neuroscience, organic compounds, learning and memory, hormones, simulation and modeling, cognitive psychology, cognition, memory, amines, neurotransmitters, catecholamines, research and analysis methods, behavior, chemistry, brainstem, biochemistry, behavioral conditioning, psychology, organic chemistry, anatomy, biogenic amines, biology and life sciences, physical sciences, cognitive science

journal.pgen.0030206

The Role of Carcinine in Signaling at the Drosophila Photoreceptor Synapse

An exceedingly complex regulation is involved in the synthesis , release , activity , and recycling or degradation of neurotransmitter in the nervous system of animal species ., This regulation may involve neurotransmitter degradation within the synapse 1 , reuptake of neurotransmitter by presynaptic neurons 2 , recycling of neurotransmitter by neighboring cells 3 , and/or activation of receptors that trigger positive/negative feedback loops resulting in an increase/decrease of neurotransmitter release in presynaptic neurons 4 ., Many of the mechanisms and machinery components associated with this regulation have been well conserved across species , ranging from Caenorhabditis elegans to humans 5 ., D . melanogaster is often utilized when studying neurotransmitter dynamics because of its malleable genetics , strong phenotypes in the presence of neurotransmission defects , and its sensitivity to numerous neuropharmacological compounds that have been shown to exert similar effects in humans ( for review , see 6 ) ., The major neurotransmitter released by photoreceptor cells in Drosophila is histamine 7 , which is biosynthesized from histidine by the enzyme histidine decarboxylase ( Hdc ) found within the photoreceptor cell 8 ., Upon excitation by light , the photoreceptor cell depolarizes and releases histamine , which then binds to a postsynaptic histamine-gated chloride channel , resulting in the hyperpolarization of the postsynaptic neuron 9 , 10 ., Each synaptic cartridge is surrounded by three glial cells that invaginate into the photoreceptor terminals by means of fingerlike projections known as capitate structures 11 ., While the exact site of histamine reuptake is currently unknown , the histamine remaining in the synaptic cleft is thought to be rapidly taken up by glial cells , possibly at the site of these capitate structures 12 ., In glial cells this histamine is converted by the enzyme N-β-alanyl dopamine synthase , encoded by the ebony gene in Drosophila , into β-alanyl-histamine , also known as carcinine 3 , 13 ., This carcinine is then released by the glial cell as an “inactive” conjugate of histamine , again possibly at the site of capitate structures , where it is then taken up by the presynaptic neuron ., Once in the photoreceptor cell , the carcinine is converted by the enzyme N-β-alanyl dopamine hydrolase , encoded by the tan gene , back into the original neurotransmitter histamine 13 , 14 ., It is proposed that the combination of histamine biosynthesis through histidine decarboxylase and the recycling of histamine by ebony and tan enzymes defines the total pool of histamine available at the photoreceptor cell synapse ., The transporters responsible for histamine uptake by glial cells , and for carcinine internalization by photoreceptor cells , are currently unknown ., The ine gene is believed to encode a putative neurotransmitter transporter , and two ine cDNAs have been sequenced and identified 15 , 16 ., The shorter cDNA , ine-RB , encodes the protein Ine-P2 , while the longer cDNA , ine-RA , encodes the protein Ine-P1 , which contains an additional ∼300 amino acids at its N terminus ., The function of the additional N-terminal region of ine-P1 is currently unknown ., Despite efforts to identify the neurotransmitter transported by inebriated in transfected Xenopus laevis oocytes , the substrate of the inebriated protein has remained elusive 17 ., Mutations in the ine gene result in an increase in the rate of onset of long-term facilitation at the larval neuromuscular junction 18 , as well as an increase in the neuronal excitability associated with mutations in the Shaker gene , which encodes the α-subunit of a potassium channel 19 ., Both of these neuronal excitability phenotypes are believed to be caused by the defective reuptake of an unknown neurotransmitter , and thus the overstimulation of postsynaptic neurons ., A third , and less understood , phenotype associated with ine mutations is manifested as an aberrant electroretinogram ( ERG ) 15 , 20 ., ERGs measure the mass retinal response of the eye to a stimulus of light , and the ERG of ine mutants is characterized by several defects , including most noticeably a series of strong oscillations in the presence of light 15 , 20 ., Recently , in an excellent and comprehensive review of histaminergic neuronal signaling in arthropods , it was proposed that the ine gene in Drosophila might encode the carcinine neurotransmitter transporter 21 ., Here , we provide genetic and pharmacological evidence linking the mutant ine-associated phenotype with the buildup of carcinine in the photoreceptor synaptic cleft and with the activity of a putative H3 receptor in the Drosophila eye ., The repo-GAL4 , “long”-GMR-GAL4 , w1118 , tan1 , tan2 , e1 , e11 , ortPbac fly lines were all obtained from Bloomington Stock Center ., HdcP218 and ort5 fly mutants were obtained from W . Pak ( Purdue University ) while the UAS-ine-RB and ine2 transgenic fly lines were obtained from M . Stern ( Rice University ) ., All stocks were maintained in constant darkness at room temperature ., Flies carrying two or three mutations/transgenes were generated by standard genetic methodologies ., All wild-type flies were of the w1118 background ., Flies were anesthetized by exposure to carbon dioxide and immobilized within a rotating disc using a drop of molten myristic acid ( Akros ) ., To record voltage changes within the eye , an electrode filled with signa gel ( Parker Labs ) was placed on the surface of the eye , while a second gel-filled electrode was gently inserted into the thorax ., For light treatments a halogen lamp controlled by a Model T132 Uniblitz shutter was used ., All light treatments , unless otherwise stated , were performed using a 580 nm filter and were 4 s in duration ., To reduce the effects that exogenous sources of histamine might have on ine2HdcP218 flies during ERG analysis , these animals were starved for 24 h before testing ., To induce depolarization spikes in ort5 flies , two 4-s pulses of 480 nm light , followed by two 4-s pulses of 580 nm of light , were delivered to the flies , and a trace was taken during the second 580 nm pulse for analysis ., Voltage changes were amplified using a DAM50 amplifier ( World Precision Instruments ) recorded using Powerlab 4/30 ( AD Instruments , Colorado Springs , CO ) and analyzed using Chart 5 software ( AD Instruments ) ., Oscillation frequency was determined by counting and averaging the number of repolarization spikes observed within 0 . 2 s of light exposure in either ine2 or carcinine-treated fly ERG recordings ., Thioperamide , immepip , and histamine were obtained from Sigma and carcinine was obtained from Peninsula Laboratories ., All compounds were reconstituted in sterilized water for long-term storage ., Flies were treated overnight in vials containing Kimwipes soaked with 200 μl of 1% sucrose solution with or without drug compound ., Histamine was delivered to flies at a concentration of 10% 22 ., Thioperamide and immepip were used at 0 . 5% and carcinine at either 5% or 10% ., Flies were starved for 24 h before drug treatment ., RNA was purified from either embryos or adult fly heads of the w1118 background by employing an RNeasy Mini Kit ( Qiagen Sciences ) ., cDNA was generated from purified RNA by utilizing MMLV-Reverse Transcriptase ( Fisher Scientific ) ., Forward primers specific to either the ine-RA or ine-RB transcripts and a reverse primer common to both transcripts were obtained from Integrated DNA Technologies ., The ine-RA forward primer was ATCGATGGCCACTTCCGGATTACA , the ine-RB forward primer was ATCAGTTGCCACTCCCAGTTTCCA , and the reverse primer used to generate PCR product from both transcripts was TATCCTATGCAGGCCAGGACGAAT ., Products were generated and amplified by means of PCR using Taq polymerase , buffers obtained from Invitrogen , and a TECHNE , TC-312 thermocycler ( Bartoworld Scientific ) using the following parameters for 35 cycles ( 94 °C for 30 s , 55 °C for 30 s , and 72 °C for 90 s ) ., PCR products were separated in a 1% agarose gel and then stained using ethidium bromide ., An ERG recording from a wild-type fly ( Figure 1A ) contains a receptor component , or the depolarization response upon exposure to light , and on and off transient spikes that indicate the response downstream of the photoreceptor cell ( arrows , Figure 1A ) ., The ERG of ine mutants contains an intact receptor component but with the addition of an initial depolarization spike ( unfilled arrowhead , Figure 1B ) and prominent oscillations superimposed on the depolarization response ( Figure 1B ) ., These oscillations have a wide range of frequencies from 40–90 Hz ., These mutants also possess reduced on and off transients ( arrows , Figure 1B ) , indicating impaired photoreceptor synaptic transmission ., Finally , these ine mutant ERGs often display a hyperpolarization following a light response ( arrowhead , Figure 1B ) ., We observed all of these previously described ERG phenotypes when using either ine2 or ine3 allele flies ., The ine3 allele is the result of a deletion of the majority of the ine open reading frame common to both ine-RA and ine-RB 16 , while the mutation associated with ine2 was identified as being a nonsense mutation in codon 125 of the ine gene and is believed to only affect the ine-RA-encoded isoform 23 ., Because the ine3 allele is associated with reduced viability , and because we could discern no observable difference between the ERG traces of ine2 and ine3 flies , we made use solely of the ine2 fly line for all of our experiments and genetic crosses ., Intracellular voltage recording experiments suggest that the oscillations observed in ine mutants originate in the photoreceptor cell , and that they are not the result of synaptic feedback 24 ., However , if ine does encode a neurotransmitter transporter , its expression and localization are not necessarily restricted to photoreceptor cells , as neurotransmitter transporters often function from neighboring glial cells ., Indeed , a previous study demonstrated that expression of ine in either neurons or glial cells was sufficient to rescue several mutant ine-associated defects at the neuromuscular junction 23 ., In order to confirm that the inebriated protein is needed at the photoreceptor cell synapse , we tested whether the ine2 mutant phenotype could be rescued by expressing the ineRB transcript in photoreceptor and glial cells ., The UAS-ine-RB transgenic fly line 23 , 25 contains the ine-RB cDNA under the control of the upstream activator sequence of the yeast Gal4 transcription factor ., These UAS-ine-RB flies will only express Ine-P2 when crossed with a second line of flies expressing Gal4 ., The Gal4 lines utilized were “long”-GMR-GAL4 , which expresses Gal4 protein specifically in photoreceptor cells 26 , and repo-GAL4 , which expresses Gal4 in glial cells 27 ., A strong rescue of the ine2 ERG phenotype was observed when ine-RB was expressed in either photoreceptor or glial cells ( Figure 1C and 1D ) compared to non-rescued ine2 or wild-type control ( w1118 ) ERGs ( Figure 1A and 1B ) ., As expected , the UAS-ine-RB transgene failed to rescue the ine2 phenotype if neither GAL4 transgene was present ( unpublished data ) ., The oscillations observed in ERG traces from these transgenic rescued animals were either absent or greatly reduced , and the hyperpolarization response was also significantly diminished ., Finally , the rescued ERG traces contained larger on and off transients than the ine2 non-rescued controls ( arrows , Figure 1C and 1D ) ., Expression of ine-RB in glial cells appears to give a stronger and more consistent rescue of the ine2 ERG phenotype than when expressed in photoreceptor cells ., This may be due to stronger expression of the Gal4 transcription factor in repo-GAL4 flies than in GMR-GAL4 animals , or it may be due to the need for full-length ine-RA , rather than ine-RB , expression in photoreceptor cells ., It is also surprising that ine-RB expression has the ability to rescue the ine2 ERG phenotype , as the ine-RB transcript was previously thought to remain intact in ine2 mutant flies 25 ., These results suggest that ine-RB is normally expressed at only low levels compared to ine-RA , and that overexpression of ine-RB is sufficient in compensating for the loss of ine-RA associated with ine2 mutants ., Reverse transcriptase PCR experiments confirm these suspicions; ine-RB is expressed at low levels in adult wild-type heads compared to robust expression of ine-RB in the developing embryo ( Figure 1E ) ., The ine-RA transcript was found at high levels in both wild-type embryos and adult heads ( Figure 1E ) ., The ability to rescue the ine2 ERG response by expressing inebriated protein in photoreceptor and glial cells suggests that inebriated functions primarily at the site of the photoreceptor cell in the eye ., Histamine is believed to be the predominant neurotransmitter that signals between photoreceptor cells and second order laminar neurons in the optic lobe 28 , and it is possible that inebriated serves as a histamine transporter ., However , previous studies showed that when inebriated protein from the tobacco hornworm Manduca sexta , which has significant homology to the inebriated protein from Drosophila , was expressed in Xenopus laevis eggs , it was unable to transport histamine across the cell membrane 17 ., However , these authors do propose that a second unknown protein may be required to assist inebriated in proper neurotransmitter transport function , or that inebriated may possess different substrates in Manduca compared to Drosophila ., Thus , histamine could still be the substrate of inebriated in Drosophila ., Histamine is generated by the activity of histidine decarboxylase , encoded by the Hdc gene in Drosophila ( Figure 2A ) ., Mutations in the Hdc gene , such as in the case of the HdcP218 allele , result in flies possessing disrupted photoreceptor synaptic transmission , as demonstrated by the lack of on and off transients in their ERGs ( 8 , and Figure 2C ) ., Approximately 80% of flies that were homozygous for both the HdcP218 and ine2 alleles displayed ERGs with no oscillations ( Figure 2D ) when compared to ine2 controls ( Figure 2B ) ., There was a small percentage ( ∼20% ) of HdcP218 ine2 flies that displayed weak or delayed oscillations ., However , the ERGs from these flies that displayed this weak rescue also possessed on and off transients , suggesting that the HdcP218 allele was either not fully penetrant in these double mutants , or that their food provided an outside source of histamine ., This is not surprising , as Drosophila photoreceptors are known to regain some function from exogenous histamine taken up at minute levels in their food 22 ., The ERGs from HdcP218 ine2 also often lack the hyperpolarization response characteristic of ine2 flies ( Figure 2D ) ., These findings suggest that histamine production or signaling plays a strong role in the oscillation and hyperpolarization phenotype observed in ine2 traces ., The postsynaptic receptor for histamine in Drosophila is a histamine-gated chloride channel ( Figure 2A ) , and a subunit of this channel is encoded by the ora transientless ( ort ) gene 10 ., The ort5 and ortPbac alleles both result in reduced activity of this histamine receptor in Drosophila , as shown by the lack of on and off transients in their ERG traces ( 10 and Figure 2E ) ., If ine encodes a histamine neurotransmitter transporter , then reduced function of this protein may result in an excess of histamine in the synaptic cleft , and this excess of neurotransmitter may be acting upon this postsynaptic histamine receptor to somehow generate the observed oscillations ., If this were the case , then ine2;ort double mutants should have reduced oscillations ., However , neither ine2;ort5 ( Figure 2F ) nor the ine2;ortPbac ( unpublished data ) double mutants exhibited rescue of the oscillation or hyperpolarization components of the ine2 ERG recordings , indicating that the oscillations do not arise from histamine signaling in downstream neurons ., Moreover , since ort mutations block signaling in laminar neurons , these data are consistent with the oscillations being generated in the photoreceptor cells ., Surprisingly , the ort5 allele , which is the result of a frameshift mutation and therefore likely serves as a null allele for this gene 10 , often displays strong depolarization spikes of its own in the receptor component of its ERG trace ( Figure 2G ) ., Therefore , mutations in Hdc , which block the formation of histamine , rescue ine2 whereas mutations in ort , which still allow for the synthesis of histamine , fail to rescue ., The ablation of the ine mutant ERG phenotype upon the introduction of Hdc , but not ort , mutations , suggests that histamine is involved in generating the ine2 ERG phenotype , but that histamines downstream signaling in the optic lobe is not ., The recycling pathway of histamine in the eye has been well elucidated 13 , 14 , 29 , 30 ., It has been shown that following release into the synaptic cleft , histamine is rapidly taken up by neighboring glial cells and is converted by the β-alanyl-dopamine synthase , encoded by the gene ebony , into β-alanyl-histamine , also known as carcinine ( Figure 3A ) ., This carcinine is then transported into the presynaptic photoreceptor cell and is converted back into histamine by β-alanyl-dopamine-hydrolase , encoded by the gene tan , for use as a recycled source of neurotransmitter ( Figure 3A ) ., Both tan and ebony mutations in Drosophila are associated with significant reductions in size of the on and off transients in ERG traces ( Figure 3C and 3E ) , due to the loss of this recycled pool of histamine in the eye ., Introduction of either tan2 ( unpublished data ) or tan1 mutations into an ine2 background failed to have any effect in reducing the size of the oscillations or hyperpolarization response when compared to ine2 mutants alone ( Compare Figure 3B with Figure 3D ) ., However , ine2;ebony1 ( unpublished data ) or ine2;ebony11 double mutants displayed complete rescue of the oscillation phenotype in all flies tested ( Compare Figure 3B with Figure 3F ) ., These data , combined with the fact that histamine synthesis is necessary for the presentation of a mutant ERG phenotype in ine2 flies , provide genetic evidence that carcinine is involved in generating ine2-associated oscillations ., If ine encodes a carcinine neurotransmitter transporter , as the genetic evidence above suggests , than a potential cause of the aberrant ERG phenotypes seen in ine mutants could be the buildup of carcinine within the photoreceptor synaptic cleft ., In order to test whether or not carcinine is able to induce an ine2-like ERG phenotype in wild-type animals , w1118 flies were treated with 5% carcinine overnight and then subjected to ERG analysis ., Approximately 35% of the w1118 flies treated with carcinine displayed occasional weak oscillations or brief depolarization/repolarization spikes in the photoreceptor response of their ERG traces ( Figure 4B and 4C , compare to Figure 4A ) ., While these spikes exhibit no consistent frequency , unlike the oscillations seen in ine2 recordings , the carcinine-induced ERG disturbances were never observed in untreated starved flies ., If carcinine was delivered to ebony11 flies , which lack the ability to synthesize carcinine from histamine , they surprisingly displayed an abnormal ERG trace ., All ebony11 flies treated with carcinine manifested phenotypes reminiscent of those seen in ine2 ERG traces , including sharp depolarization spikes in response to a light response , weak oscillations , and a hyperpolarization peak upon the termination of light ( Figure 4E and 4F , compare to Figure 4D ) ., The oscillations observed in carcinine-treated ebony11 mutants , while only appearing briefly during the initiation of light exposure , were seen at a similar frequency as those found in ine2 ERG recordings ( 63 spikes/s ) ., As discussed below , a possible mechanism underlying these carcinine-induced ERG disturbances may involve the sensitization of a putative histamine/carcinine receptor ., If inebriated does serve as a carcinine transporter , and if carcinine is indeed building up within the synaptic cleft in ine2 mutants , then this uncleared carcinine appears to somehow be acting on some synaptic receptor to initiate this aberrant oscillation phenotype ., The ine2;ort5 experiments suggest that this receptor is not the post-synaptic histamine-gated chloride channel ., In mammals and various other vertebrate systems , presynaptic histaminergic neurons often contain their own histamine receptors , known as H3 receptors ., The H3 receptor is a G-protein–coupled receptor that was first identified in 1983 by Arrang et al . 4 and is now known to act as a presynaptic autoreceptor that inhibits histamine release from histaminergic neurons in the brain ( for review , see 31 ) ., Thus , H3 receptors serve to negatively regulate histamine release and synthesis in the presence of high histamine levels in the synaptic cleft ., While no H3 receptor has been identified yet in Drosophila , there are several candidate genes that may encode such a putative receptor ., There are numerous well-characterized pharmaceutical compounds that act as agonists , antagonists , or inverse agonists of the H3 receptor in vivo in mammals , and recently carcinine was identified as being an inverse agonist of this receptor in mice 32 ., It was shown that , rather than reduce histamine release , as occurs in the case of histamine binding to an H3 receptor , carcinine had the opposite effect and induced both histamine synthesis and release from presynaptic histaminergic neurons in vivo ., A possible scenario to explain the oscillations seen in ine2 ERGs is that histamine and uncleared carcinine are competing for binding to putative H3 receptors , resulting in opposing signaling cascade responses in the photoreceptor cell ., If this is the case , disrupting this balance of histamine and carcinine binding to the putative H3 receptor in ine2 fly eyes should result in a rescue of ERG oscillations ., Indeed , treatment of ine2 flies with 10% carcinine resulted in a rescue of oscillations in 35% of flies ( unpublished data ) , and treatment of ine2 flies with 0 . 5% thioperamide , another well-characterized and potent inverse agonist of the H3 receptor in mammals , resulted in the consistent and complete ablation of oscillations in ERG traces in all flies tested ( Compare Figure 5A with Figure 5C ) ., In addition , treatment of ort5 flies with 0 . 5% thioperamide resulted in a loss of ort5-associated depolarization spikes ( unpublished data ) ., Surprisingly , treatment of wild-type control flies with 0 . 5% thioperamide resulted in the loss of on and off transients in their ERGs ( compare Figure 5B with Figure 5D ) ., It should also be possible to disrupt the hypothetical balance of histamine and carcinine binding to a photoreceptor cell-specific H3 receptor in ine2 by introducing an H3 receptor agonist , such as histamine itself ., Indeed , treatment of ine2 flies with 10% histamine ( unpublished data ) or 0 . 5% immepip ( Figure 5E ) , another potent H3 receptor agonist , resulted in a strong rescue of oscillations in >50% of flies tested ., Occasionally , weak oscillations and depolarization spikes were still observed in immepip- or histamine-treated ine2 flies ( Figure 5E ) ., Neither histamine nor immepip treatment had a strong or consistent effect on the on and off transients seen in wild-type control ERGs ., Since immepip and thioperamide are known to be specific agonists and inverse agonists of the mammalian H3 receptor , these pharmacological experiments suggest that an H3 receptor may exist in Drosophila and that abnormal stimulation of this H3 receptor is occurring in the eyes of ine2 Drosophila mutants ., Previous studies involving intracellular voltage recordings of ine mutants led the authors to conclude that the oscillations observed in ine mutant ERGs were the result of a defect occurring within the photoreceptor cell 24 ., We were able to support these conclusions by expressing ine specifically in photoreceptor cells and demonstrating a rescue of the ine2-associated oscillations ., Neurotransmitter transporters are often able to function from either the presynaptic neuron or from neighboring glial cells , as shown at the neuromuscular junction in ine mutants 23 ., We found that glial cell–specific expression of the ine gene in ine2 flies resulted in a complete rescue of the ine mutant ERG phenotype ., It was somewhat unexpected that ine expression in glial cells rescued the ine2 phenotypes , as glial cells have been shown to lack tan protein and thus would be unable to convert carcinine back to a recycled pool of histamine 30 ., However , it is possible that glial cells do express trace amounts of the enzyme tan to hydrolyze carcinine and generate a renewable source of histamine for photoreceptor cells , and it is also possible that the inebriated protein is expressed in a non-autonomous manner and can be transported from glial cells to photoreceptors in the fly eye ., The finding that an ERG recording can exhibit oscillations is somewhat surprising ., An ERG does not record the electrical response of a single photoreceptor , but rather is a collective measure of the retinal photoresponse ., Thus , if the mutant ine-associated ERG defects are indeed localized to the photoreceptor synapse , as our data and that of previous labs suggest , then one would expect that different photoreceptors would be excited/inhibited at different timepoints , ultimately resulting in the oscillations simply canceling themselves out ., The fact that oscillations are indeed observed , and appear to be due to a defect occurring at the photoreceptor synapse , implies the existence of an uncharacterized and complex synchronization of photoreceptor cell de-/repolarization ., The lack of rescue of ine2-associated oscillations in flies carrying additional mutations in the postsynaptic histamine receptor gene ort , the finding that mutant ine oscillations were detected within single photoreceptor cells 24 , and our observations that the mutant ine phenotype can be rescued when ine is expressed in photoreceptors , all combine to strongly suggest that the oscillation phenotype is likely a result of a defect occurring within the photoreceptor itself ., In addition , by crossing ine2 animals with HdcP218 flies , we demonstrated that the ine2-associated oscillations are dependent upon histamine synthesis ., All of these results indicate that histamine is somehow contributing to the aberrant ERG witnessed in ine2 flies , and that histamine appears to be acting on the presynaptic photoreceptor cell to induce this oscillation phenotype ., Further epistatic analyses also revealed that ebony , but not tan , activity is required for the generation of oscillations in ine2 ERGs ( Figure 6A ) ., These genetic experiments are consistent with ine encoding either a carcinine importer found in the photoreceptor cell or a carcinine exporter found in glial cells ., The homology of inebriated with other known Na+/Cl− neurotransmitter transporters ( which import neurotransmitter into cells ) 16 suggests that inebriated protein is transporting carcinine into the photoreceptor , and not out of glial cells ., While ebony is known to act on multiple substrates , such as dopamine to generate β-alanyl-dopamine 13 , the requirement of histamine synthesis for the maintenance of ine2-associated oscillations suggests that it is β-alanyl-histamine , or carcinine , that is somehow responsible for the oscillations observed in ine2 ERGs ., It should be noted , however , that ebony mutations were not sufficient in rescuing the hyperpolarization response observed in mutant ine ERG traces ( Figure 3F ) ., The origins of this hyperpolarization response are still unclear and further research will be required to elucidate its exact meaning ., In tan mutants , one would predict that there would be a buildup of carcinine ., However , this buildup does not give rise to an ERG recording similar to that of ine2 ., This is due most likely to the presence of functional inebriated protein in tan mutant flies , which should effectively clear the carcinine from the synaptic cleft for degradation within the photoreceptor cell ., By treating wild-type and ebony11 flies with carcinine and subsequently inducing components of the ine2-ERG phenotype , we provide further evidence suggesting that the sharp depolarization spike , the oscillations , and the hyperpolarization response all seen in ine2-ERGs are due to a buildup of carcinine within the photoreceptor synaptic cleft ., While the oscillations observed in carcinine-treated wild-type flies do not mimic exactly the oscillations seen in ine2 ERG recordings , it is presumably difficult to replicate the carcinine and histamine balance occurring in the eyes of ine2 animals ., Indeed , treatment of wild-type flies with higher ( 10% ) or lower ( 1% ) concentrations of carcinine were less effective in inducing the oscillations than the described 5% carcinine dose ( unpublished data ) ., It is possible that carcinine is being degraded or modified by the fly before the compound is able to exert its effects at the photoreceptor cell ., In order to eliminate the activity of one enzyme known to be involved in carcinine metabolism , tan1 flies were treated with 5% carcinine overnight ., Surprisingly , none of the tan1 flies treated with carcinine showed an aberrant ERG phenotype ( unpublished data ) ., It was surprising that carcinine treatment had a strong effect in flies of the ebony11 , but not the tan1 , background ., While the results of these tan1 and ebony11 carcinine-treatment experiments are unexpected , one possible explanation may involve the regulation of carcinine clearance/degradation ., The tan1 flies presumably suffer from a perpetual excess of carcinine even before exogenous carcinine treatment , and these flies , in order to reduce their sensitivity to this compound , may consequently decrease the levels of a putative carcinine receptor , increase their rate of carcinine degradation , or increase the levels of inebriated protein for carcinine clearance ., However , ebony11 flies are relatively “naïve” to the effects of carcinine , as their ability to synthesize this compound has been greatly diminished , and as a result these animals may have an increased level of the supposed carcinine receptor , a decrease in inebriated receptor levels or a decrease in carcinine degradation , ultimately making them more sensitive to the effects of carcinine treatment ., It remains to be seen whether or not all of the mutant ine-associated phenotypes , including increased neuronal excitability 19 , 23 and increased sensitivity to osmotic stress 25 , are due to the inability of these flies to transport carcinine ., It is possible that the inebriated protein transports other compounds that perhaps share the common feature of β-alanine conjugation ., This might help explain why none of the more common neurotransmitters were taken up by ine-transfected Xenopus oocytes 17 ., In order to assist in confirming that inebriated is indeed a carcinine neurotransmitter transporter , in vitro experiments , such as neurotransmitter uptake assays , will need to be performed ., In addition , the ability of inebriated protein to take up other β-alanyl-neurotransmitters/osmolytes also should be examined ., The oscillations present within the photoreceptor response of ine2 ERGs appear as sharp depolarization/repolarization spikes , and this oscillation phenotype is dependent upon both histamine synthesis and ebony activity ( Figure 6A ) , and is sensitive to drugs that target mammalian H3 receptors ., It is perplexing that the synthesis of a single metabolite , carcinine , could be responsible for both the depolarization and repolarization spikes observed within ine mutant ERGs ., We speculate that these oscillations are the r

Introduction, Materials and Methods, Results, Discussion, Supporting Information

The Drosophila melanogaster photoreceptor cell has long served as a model system for researchers focusing on how animal sensory neurons receive information from their surroundings and translate this information into chemical and electrical messages ., Electroretinograph ( ERG ) analysis of Drosophila mutants has helped to elucidate some of the genes involved in the visual transduction pathway downstream of the photoreceptor cell , and it is now clear that photoreceptor cell signaling is dependent upon the proper release and recycling of the neurotransmitter histamine ., While the neurotransmitter transporters responsible for clearing histamine , and its metabolite carcinine , from the synaptic cleft have remained unknown , a strong candidate for a transporter of either substrate is the uncharacterized inebriated protein ., The inebriated gene ( ine ) encodes a putative neurotransmitter transporter that has been localized to photoreceptor cells in Drosophila and mutations in ine result in an abnormal ERG phenotype in Drosophila ., Loss-of-function mutations in ebony , a gene required for the synthesis of carcinine in Drosophila , suppress components of the mutant ine ERG phenotype , while loss-of-function mutations in tan , a gene necessary for the hydrolysis of carcinine in Drosophila , have no effect on the ERG phenotype in ine mutants ., We also show that by feeding wild-type flies carcinine , we can duplicate components of mutant ine ERGs ., Finally , we demonstrate that treatment with H3 receptor agonists or inverse agonists rescue several components of the mutant ine ERG phenotype ., Here , we provide pharmacological and genetic epistatic evidence that ine encodes a carcinine neurotransmitter transporter ., We also speculate that the oscillations observed in mutant ine ERG traces are the result of the aberrant activity of a putative H3 receptor .

During signaling in the nervous system , individual nerve cells transfer information to one another by a complex process called synaptic transmission ., This communication involves the release of a specific neurotransmitter into the synaptic cleft , which then triggers signaling in the downstream neuron by binding to and activating specific cell surface receptors ., In order to terminate the neuronal signal , the neurotransmitter must be rapidly removed from the synaptic cleft ., This is done by two mechanisms: the neurotransmitter can be degraded or modified , or the transmitter can be taken up by the presynaptic neuron and packaged into vesicles for reuse ., In the compound eye of the fruitfly D . melanogaster , the photoreceptor cell responds to light and releases histamine into the synaptic cleft ., This signal is terminated by the removal of histamine from the synapse and the enzymatic conversion of histamine to carcinine ., We have shown that it is not sufficient just to modify the histamine neurotransmitter , but it is also important to remove carcinine from the photoreceptor synapse ., The failure to adequately remove carcinine results in defects in the visual transduction process ., Moreover , the work suggests that carcinine itself modulates vision by regulating histamine release into the synapse .

drosophila, cell biology

journal.pcbi.1005255

The Power of Malaria Vaccine Trials Using Controlled Human Malaria Infection

In 2015 , malaria caused an estimated 438 , 000 deaths ( 236 , 000–635 , 000 ) 1 , most of which were associated with Plasmodium falciparum ( Pf ) infections ., Hypothetically , a malaria vaccine targeting the sporozoite parasite stage ( pre-hepatic vaccine ) , hepatic parasite stage ( hepatic vaccine ) , and/or blood stage parasites ( erythrocytic vaccine ) could prevent many such deaths ., In the past ten years , over 40 malaria vaccines have reached the clinical trial stage ., So far , only the RTS , S vaccine has shown promising results ( 30%–65% protection against clinical malaria ) 2–6 , with a recent large , multi-center phase three trial showing 45 . 7% protection against clinical malaria in infants and children aged 5–17 months over a period of 18 months after three vaccine doses 7 ., In response , the World Health Organization has recommended RTS , S for pilot implementation studies in Africa 8 ., However , before any malaria vaccine can be tested in the field , its efficacy and safety need to be evaluated in controlled settings , which is most often done by means of controlled human malaria infection ( CHMI ) 9 , 10 ., CHMI is currently considered to be a powerful tool in clinical vaccine development , as it allows researcher to control the otherwise highly variable infection rates ., It is traditionally conducted by exposing a limited number of volunteers to laboratory-reared Anopheles spp ., mosquitoes carrying Pf sporozoites 10 , 11 , and more recently , also through needle injection of a defined number of aseptic cryopreserved sporozoites 12 , 13 ., Parasitaemia can be monitored by means of blood smear microscopy but increasingly by sensitive quantitative real-time polymerase chain reaction ( qPCR ) 14–17 ., The traditional study endpoint is detection of blood stage parasites by microscopy , at which point a study subject is treated with a curative regimen of an anti-malarial drug 10 ., Under these conditions and with some additional precautionary measures , CHMI studies are considered to be safe , though the risk of severe adverse effects and the burden of venipuncture up to three times per day have to be weighed against the benefits of the information to be gained 18–20 ., Power calculations for CHMI-based vaccine trials are therefore critical to ensure an appropriate number of included participants to obtain meaningful estimates of protective efficacy ., Here we provide an optimized method for power calculations for CHMI-based hepatic and erythrocytic vaccine trials in malaria-naive individuals , using qPCR and bloodsmear data from mosquito-based CHMI experiments in 57 non-vaccinated individuals ., We developed a novel non-linear model for parasite kinetics during the first two weeks of an experimental infection , based on an earlier statistical model for cyclical patterns in CHMI 21 ., We implemented the model in a Bayesian hierarchical framework to capture important sources of variation both within and between individuals ( i . e . by means of random effects ) , and extended the model to explicitly capture the processes leading to measurements below the qPCR detection limit ( censoring ) and termination of the experiment due to positive blood microscopy ., With this model , we performed power calculations for hypothetical hepatic and erythrocytic malaria vaccine trials under varying assumptions about blood sampling schemes and vaccine impact on parasite growth ., In Fig 1 we provide an example of the model fit to the data for a subset of individuals , and illustrates the cyclical pattern in both the data and model predictions ., Note how censored observations ( black triangles ) coincide with model predictions near or under the qPCR detection limit ( dashed line ) ., Similar plots for all individuals can be found in S1 Fig . For a minority of subjects no clear cyclical pattern in parasitaemia levels was visible , resulting in relatively wide credible intervals for predicted parasitaemia levels ., Data from one individual were excluded ( ZONMW1 . 1270 ) because they were prohibitively divergent from the rest to effectively fit the model ( first observation above detection limit at day ten and no clear cyclical pattern while this individual–like all others–had no vaccine protection ) ., S2 Fig depicts the association between parasite concentration in the blood and probability of positive blood microscopy in comparison to the data , illustrating good agreement between the model and the data ., We estimated that the average number of first generation parasites among the 56 included study subjects was 635 Pf/ml ( 95%-BCI 406–945; Table 1 ) ., Inter-individual variation in the number of first generation parasites was estimated at a standard deviation ( SD ) of 1 . 40 on the natural logarithmic scale , which is equivalent to a relative standard deviation ( RSD ) of exp\u2061 ( 1 . 402 ) −1=247% ., Assuming that infected hepatocytes release on average 6 . 0 merozoites per mL blood ( based on 30 , 000 merozoites per hepatocyte and an average of 5 litres of blood per person , as assumed before 21 ) , we can deduce from the average concentration of first generation parasite that on average about 107 hepatocytes were infected and that each experimentally infected mosquito transmitted on average about 21 sporozoites ( that successfully infected hepatocytes ) ., The average time of appearance of first-generation parasites was at day 6 . 87 ( 95%-BCI 6 . 76–6 . 99 ) , with little variation between individuals ( SD 0 . 036 , RSD 3 . 6% ) ., The average multiplication rate of P . falciparum parasites within individuals was estimated at 11 . 8 per parasite cycle ( 95%-BCI 9 . 0–15 . 3 ) , with inter-individual variation estimated at a SD of 0 . 47 on the natural logarithmic scale ( RSD 50% ) ., In Fig 2 , we illustrate the correlation between time of appearance of first generation blood parasites , peak concentration in blood of first generation parasites , the parasite multiplication rate , and the relative odds of an individual having positive blood microscopy ., As there was no clear correlation pattern between these four individual-level parameters , we did not further explicitly model their joint distribution ( i . e . we assume they are independently distributed when simulating data for power calculations ) ., Next , we used the model to simulate ten thousand repeated data sets for each of various hypothetical vaccine trial designs ., In each simulated vaccine trial , we define vaccine efficacy as the relative reduction in total number of released merozoites induced by a hepatic vaccine , or the relative reduction in parasite growth rate induced by an erythrocytic vaccine ., To explicitly account for previously excluding one out of 57 original study subjects , we allowed each individual in the simulated vaccine trials to be dropped with a 1/57 probability ( which turned out to be of little consequence for power estimates ) ., Fig 3 illustrates how trial power increases with the number of individuals per trial arm , allowing one to deduce the required number of individuals for a desired power level ., Study power is most dependent on the number of individuals in each treatment arm; in contrast , inter-individual variation in vaccine efficacy and the number of blood samples taken per day matter relatively little for study power ., Due to high inter-individual variation in the number first-generation parasites , hepatic vaccine trials required significantly more study subjects than erythrocytic vaccine trials ., For instance , over 30 individuals per trial arm are needed to achieve 80% study power when hepatic vaccine efficacy is 70% or lower ., S2 File provides a graphical user interface to look up power of all simulated vaccine trial settings ., The effective probability of a Type 1 error when using T-tests ( assuming unequal variances and setting α = 0 . 05 ) was approximately 5% and even somewhat lower for small sample sizes ( S3 Fig ) , which confirmed the validity of T-tests for identifying a difference between trial arms ., S4 Fig also depicts the association between trial power and number of individuals per treatment arm ( like Fig 3 ) , but under the assumption that inter-individual variation in peak first-generation parasite density in terms of SD is halved ( e . g . by using needle-injected instead of mosquito-based CHMI ) ., This figure illustrates how such a reduction in inter-individual variation may reduce the number of individuals required for hepatic vaccine trials ., For instance , only 11 to 15 individuals per trial arm ( depending on number samples per day ) would be needed to achieve 80% study power when hepatic vaccine efficacy is 70% , instead of around 30 or more when using mosquito-based CHMI ( Fig 3 ) ., For comparison , we further repeated the power analysis using the log-linear sine model 22 , which we also improved to explicitly capture censored observations and inter-individual variation in slope , intercept , and timing of the log-linear sine curve ., Table S2 in S1 Text summarizes the assumed prior distributions and posterior parameter estimates for the sine model; S5 Fig depicts the model fit to the data ., The log-linear sine model resulted in a higher estimate of the parasite growth rate than our non-linear model: 4 . 80 per day , or 17 . 9 per parasite cycle ( vs . 11 . 8 per cycle ) , assuming a parasite cycle duration of 1 . 84 days ( as estimated by the non-linear model ) ., Further , the sine model resulted in a higher estimate of qPCR measurement error ( 1 . 22 vs . 0 . 98 on the natural logarithmic scale , or RSD of 155% vs . 129% ) , which is an indication of the sine model providing an inferior fit to the data compared to the non-linear model , as it attributed some of the temporal variation in the data to ( random ) measurement error ., As a result , the sine model produced more optimistic estimates of study power than our non-linear model , especially for hepatic vaccines ( S6 Fig ) ., We present robust power analyses for malaria vaccine trials based on controlled human malaria infection ( CHMI ) experiments , using a Bayesian non-linear model for blood parasite kinetics ., Our model adequately accounts for the cyclical nature of P . falciparum concentrations in the host blood , based on robust estimates of biological parameters underlying CHMI ., Our study supersedes earlier models and power calculations: we prevent overly optimistic estimates of trial power by directly modeling the biological processes behind the cyclical patterns in CHMI to best capture temporal patterns in the data , and by jointly and robustly estimating all model parameters in a Bayesian hierarchical model framework ., Furthermore , we avoid excessive weighing of outliers in data simulation , appropriately model the impact of detection limits , and account for five important sources of variation between individuals: time of first parasite appearance , first cycle amplitude , parasite multiplication rates , probability of positive blood microscopy ( and consequent termination of the experiment ) , and vaccine efficacy ., This makes our model a more suitable tool for power calculations of CHMI-based vaccine trials than previous approaches to date ., Our study confirms that for vaccine trial power , inter-individual variation in vaccine efficacy and the number of blood samples taken per day matter relatively little , and that erythrocytic vaccine trials require significantly fewer study subjects than hepatic vaccine trials 23 ., Importantly , our analyses suggest that loss of only a few study subjects ( e . g . due to experimental failure ) may lead to drastic loss of trial power , unless the trial power is already close to 100% , which is important to consider during trial design ., For instance , if the anticipated efficacy of a hepatic vaccine is 80% ( reduction in number of first-generation parasites ) , 15 people per treatment arm will provide just over 80% power to detect this level of efficacy , but drop out of even one study subject per treatment arm may diminish the trial power below 80% ., Further , our results suggest that a 50% reduction in the inter-individual variation of the number of first-generation parasites ( e . g . by using needle-injected instead of mosquito-based CHMI ) may result in a substantial reduction in the number of individuals required per trial arm , especially so for hepatic vaccine trials ( reductions up to 50% ) ., These estimates will be further refined based on forthcoming data from needle-injected CHMI experiments ., The average multiplication rate of P . falciparum parasites estimated by our non-linear model is very similar to earlier estimates 22 , 24 , 25 or somewhat higher 21 , 23 , 26 ., However , our results do not confirm an earlier report by Sheehy et al . of negative correlation between the parasite multiplication rate and the peak concentration of first generation parasites in mosquito-based CHMI 13 , and suggest that this previous finding should be revisited ., Possibly , Sheehy et al . overestimated the multiplication rate in study subjects with low initial parasite loads ( two parasites per mL ) as typically , parasite loads below the detection limit for such individuals are set to some arbitrary low value ( e . g . half the value of the detection limit26 ) while ignoring measurement error ( i . e . true levels may be above the detection limit ) ., If Sheehy et al . had left out observations with low initial parasite loads ( <10 Pf/mL ) from their analysis , very little correlation would have remained ., The parameter estimates from our study are intended for power calculations for phase 1/2 hepatic and erythrocytic vaccine trials using CHMI of malaria-naive individuals with NF54 or similar strains ( i . e . first-time infections only ) , and using qPCR to monitor parasitaemia levels ., To avoid variation due to differences between labs and strains , future vaccine trials using CHMI should either be performed in a single lab using a single strain ( as was the case for the data used in this study ) , or in case of multi-center trials , each lab should cover each trial arm to allow within-lab comparisons ., Of course , the validity of power estimates based on our model relies on understanding one’s potential vaccine efficacy in terms of its impact on the number of first generation blood parasites and the parasite replication rate at the time of challenge ( in contrast to impact on clinical outcomes such as time until positive blood microscopy or duration of sterile protection ) ., The potential vaccine efficacy in a power calculation is ideally based on a target product profile ( TPP ) defined as part of a vaccine development process ., We therefore recommend that TPPs are defined not just in terms of clinical outcomes but also in terms of impact of a vaccine on parasite growth ., The link between impact on parasite growth and clinical outcomes in healthy volunteers and inhabitants of endemic areas is a topic of ongoing research ., Based on our study in health volunteers , however , we can say that to achieve sterile protection in healthy volunteers , an erythrocytic vaccine would have to reduce the parasite multiplication rate by at least 92% ( 1 − 1/β2 , where β2 is the parasite multiplication rate per cycle ) ., Further , our estimates for biological parameters can be used as prior information ( in a Bayesian framework ) in studies that aim to estimate biological and/or vaccine efficacy parameters ., If there is reason to believe that the study population is somehow different from the population covered by the current study ( e . g . a different parasite replication rate is expected ) , it is important that a sensitivity analysis be performed by incrementally diluting the relevant prior information ( i . e . increasing prior variance on the population mean and variance of e . g . replication rates ) so that the prior information becomes weaker and the model is relatively more informed by the data at hand ., Here , we predict that mosquito-based CHMI vaccine trials may require several tens of volunteers per trial arm , especially if the expected vaccine efficacy is under 70% ( in terms of impact on patterns in parasite growth ) and/or if a hepatic vaccine is being tested ., In our experience and as reported in literature 27 , CHMI experiments are typically executed with 5–8 volunteers at a time , or in several batches of that size ., Working with larger groups is impractical and expensive because many volunteers will turn positive for malaria infection on the same day , requiring more attending physicians and lab personnel ., However , future vaccine candidates should preferably have an impact on parasite growth that is much higher than 70% , which for erythrocytic vaccines can be demonstrated with 80% power using only 5 to 15 volunteers per trial arm ., Achieving similar study power for hepatic vaccine trials with the same number of volunteers might be made possible by the use of needle-based CHMI ., It has been argued that the non-linear model that we build on here 21 is unnecessarily complex as simpler models may yield very similar estimates of relevant biological parameters , and requires the estimation of vaccine-irrelevant parameters and the assumption that some of those parameters are fixed 26 ., We argue that log-linear models ( a straight line through the log-transformed parasite concentrations 24 ) and sine models ( log-linear model in which the slope is multiplied with a sine function 22 , 28 ) are too simple and require assumptions that clearly contradict the data ( linearity of data; data before first parasite generation are ignored; parasite cycles follow a symmetric , sinusoid pattern ) , making them less fit for power calculations ., In power analyses , it is imperative to consider uncertainty in all relevant parameters ( like we do here ) 29 , including the processes leading to censoring of observations below the detection limit and termination of an experiment in case of positive blood microscopy ., In this study , we demonstrate that the log-linear sine model ( even after properly accounting for censoring and variation between individuals ) results in higher power estimates than our non-linear model , which we attribute to the fact that the sine model does not capture temporal patterns in CHMI data as well as our non-linear model , as indicated by the sine model’s higher estimate for qPCR measurement error ., The bootstrapping approach to power calculation used by Roestenberg et al . 23 is a major improvement over log-linear and sine models , but it is based on the assumption that the empirical distribution of data observed so far is representative of the distribution in the population , which may cause it to assign excessive weight to outliers in power calculations , especially when based on small datasets ., Rather , our model , with its hierarchical design , shrinks such outliers towards the population mean when estimating population-level parameter values , while retaining the ability to simulate such outliers by chance ., Because we more robustly quantify parameter uncertainty , our power estimates are more realistic and dictate higher required sample sizes compared to those by Roestenberg et al . 23 Further , in the current study , we relaxed the assumption of fixing parameters by jointly estimating all parameters in a Bayesian framework ., A possible limitation in the application of our model for mosquito-based CHMI data , is the assumption that first-generation malaria parasites appear in the blood in a single wave ( an assumption shared with the sine model 22 , 28 ) ., Obviously , this was not the case for subjects in whom no clear cyclical pattern in parasitaemia levels was visible ., In this respect , CHMI data based on needle-infected PfSPZ may provide less noisy data than mosquito-based CHMI 12 , 13 , and may even result in fewer asynchronous parasite cycles , although this remains to be evaluated ., In short , like all other statistical models 22–24 , 28 , our model relies on assumptions that are not always fulfilled; nevertheless , our model best approximates and quantifies uncertainty in the biological mechanisms behind CHMI and is therefore the most suitable for performing power calculations ., As already mentioned , only the RTS , S vaccine has shown promising results so far ( 30%–65% protection against clinical malaria ) 2–6 , with a recent large , multi-center phase three trial showing 45 . 7% protection against clinical malaria in infants and children aged 5–17 months over a period of 18 months after three vaccine doses 7 ., Unfortunately , RTS , S vaccine trials have only used blood microscopy to evaluate infection levels in trial participants so far , which does not allow evaluation of vaccine impact in terms of the effect on parasite growth patterns ( and thus a link with our power analyses ) ., This obstacle may be readily overcome by the use of qPCR to monitor infection levels in Phase 1/2 studies , which may also help to reduce study sample sizes ., In conclusion , to maximize the probability of identifying effective candidate malaria vaccines , and to keep the risk of severe adverse events and the number of invasive procedures to a minimum , it is important to perform power analyses ., With this study , we provide robust power estimates for malaria vaccine trials using mosquito-based CMHI , superseding previous models and power analyses ., Given the simulation-based nature of our approach , it is straightforward to implement more complicated assumptions for future vaccine trials ., Last , our model suggests that using needle-injected instead of mosquito-based CHMI may improve the power of hepatic vaccine trials ., We used data from 57 volunteers participating in 8 different sporozoite challenge trials at Radboud University Medical Center ( RUMC , Nijmegen , The Netherlands ) from 1999 to 2011 ( Table 2 ) 14 , 21 , 30–35 ., The data set included subjects from immunological studies ( n = 20 ) , infectivity controls from immunization trials ( n = 18 ) , and non-protected subjects from a malaria vaccine trial ( n = 19 ) ., Volunteers were challenged by bites of 4–7 ( n = 20 ) or 5 infected mosquitoes ( n = 37 ) for 10 minutes; the number of bites was unknown as exposure to mosquito bites took place in the dark ( under a cloth ) , and not all individuals developed a clearly visible skin reaction to every bite ., Mosquitoes were laboratory-reared and infected with the NF54 strain of Pf ., Presence of sporozoites in mosquitoes was confirmed by salivary gland dissection ., Trial subjects were followed 2–3 times daily from day 5 after challenge until 3 days after antimalarial curative treatment ., At every visit , blood samples were collected and assessed for presence of parasites by microscopy ( threshold 4 parasites/μL ) and quantified by qPCR ( detection limit 20 or 200 parasites/mL ) 14 ., Ethical approval was obtained for each trial separately from the RUMC institutional review board and/or for some trials from the Dutch Central Committee on Research Involving Human Subjects ( 0004–00900 , 0011–0262 , 2001/203 , 2002/170 , NCT00442377 , NCT00757887 , NCT00509158 , NCT01002833 , NCT01236612 ) ., We modeled parasite kinetics in a novel Bayesian non-linear statistical model , which is based on a previous , simpler model for cyclical patterns in CHMI 21 ., The model predicts parasite concentration in the blood as measured by qPCR as a function of days since infection , mimicking successive cycles of appearance and disappearance ( sequestration ) of blood parasite generations ., The model parameters capture the following biological processes: the total number of first generation blood parasites per mL blood ( β1 ) ; the blood parasite multiplication rate , i . e . the number of next generation parasites per current generation parasite ( β2 ) ; the average time from inoculation to appearance of first generation blood parasites ( μ1 ) ; the average duration of the blood parasite stage ( μ2 ) ; the average time from parasite sequestration ( i . e . when a parasite cannot be detected in the blood ) to appearance of next generation parasites ( μ3 ) ; and the standard deviation of time of appearance and disappearance of individual blood parasites from a given generation , which represents the rate at which parasite concentrations change ( σ1 , lower values mean higher rate ) ., We assumed that qPCR measurement error follows a lognormal distribution ., We extended the original model 21 with regard to the following points ., We allowed parasite kinetics to vary between individuals by including random effects for model parameters β1 , β2 , and μ1 ., Given that the nine CHMI experiments ( Table 2 ) were performed at the same lab , using the same strain , and were performed by the same person ( CCH ) on the same PCR machine , we assumed that inter-study variation is negligible relative to inter-individual variation ( i . e . no random intercept for study ) ., Observations below the detection limit of the qPCR ( i . e . censored observations ) were explicitly modeled , rather than assuming a value equal to half the detection limit 21 , 26 ( which would introduce artificial information ) or leaving out such observations altogether 28 ( which would ignore information in the data ) ., We further jointly modeled the probability of detecting parasites through blood microscopy as a function of the predicted parasite load ( i . e . before measurement error or censoring ) , using a standard hierarchical logistic regression model ( random intercept for individuals ) ., Model parameters were jointly estimated in a Bayesian framework , giving several advantages over previously applied classic ( frequentist ) approaches 21 , 25 , 26 ., The Bayesian approach allowed simultaneous estimation of all model parameters and the associated uncertainty ( including the model parameters for positive blood microscopy ) , without the need to fix a subset of parameters ., Furthermore , the Bayesian framework allowed for exact rather than approximate inferences based on normality assumptions ., See S1 Text for a detailed description of the statistical model and the parameter estimation procedure ., Model parameter estimates are summarized in terms of posterior mean and a 95%-Bayesian credible interval ( 95%-BCI ) , which we defined as the 2 . 5th and 97 . 5th percentiles of the posterior samples for each parameter ., We performed power calculations for erythrocytic and hepatic vaccine trials , using combinations of presumed mean vaccine efficacy ( 30% , 40% , 50% , 60% , 70% , 80% , 90% , or 95% reduction in number of first-generation parasites or parasite multiplication rate ) , inter-individual variation in vaccine effect ( beta distribution with standard deviation 0 . 05 or 0 . 10 ) , number of study participants in each group , and the blood sampling frequency: one ( 8am ) , two ( 8am , 4pm ) , or three ( 8am , 4pm , 10pm ) per day , or once every two days ( even days ) , either in the morning ( 8am ) or afternoon ( 4pm ) ., For each combination of assumptions , we simulated ten thousand repeated vaccine trials , using one posterior draw of model parameters to generate one set of trial data ., We explicitly simulated the probability of blood microscopy turning positive ( and consequent termination of the experiment ) as a function of predicted blood parasite concentrations to arrive at the most realistic individual time series possible ., For each repeated vaccine trial , all individual-level random effects were drawn independently from each other ( i . e . we did not re-use the random effects estimated from the data ) ., The qPCR detection limit was set to 20 parasites per mL blood ( i . e . the current practice at Radboud University Medical Center , Nijmegen , The Netherlands ) ., Because the Bayesian non-linear model used in this study requires a substantial amount of data ( i . e . at least two , preferably three samples per day ) , each simulated vaccine trial was analyzed using simple frequentist statistical tests as previously described 23 ., First , censored observations were set to half the value of the detection limit ., Next , we categorized data by parasite cycle ( days 6 . 5–8 . 5 , 8 . 5–10 . 5 , 10 . 5–12 . 5 ) and calculated the mean log-transformed blood parasite concentration per parasite cycle and individual ., For hepatic vaccine trials , we compared first-cycle log-blood concentrations between vaccine and control arms with two-sided t-tests ( assuming unequal variances due to potential censoring ) ., For asexual erythrocytic vaccine trials , we calculated the differences in average log-blood concentrations between consecutive cycles for each individual , and then averaged these differences over cycles within individuals ., The average differences were compared between groups , again with two-sided T-tests ( assuming unequal variances due to potential censoring ) ., The predicted power of a vaccine trial was expressed as the proportion of repeatedly simulated trials that resulted in a p-value equal to or lower than 0 . 05 ., The validity of using the T-test ( i . e . the effective probability of a Type 1 error ) was checked in a similar fashion , but by simulating vaccine trials with zero effect in both treatment arms ., To explore the potential impact of using needle-injected CHMI on vaccine trial power , we repeated the power calculations assuming that the inter-individual variation in the number of first-generation parasites is half that of mosquito-based CHMI ( i . e . σβ1 , needle=σβ1/2 , based on data digitally extracted from Fig 4A in Sheehy et al . 2013 13 ) ., We further repeated the power calculation based on an analysis of the data using the simpler log-linear sine model 22 , assuming that the parasite generation time is 1 . 84 days ( as estimated by the main model ) ., For the sake of comparison , we improved the log-linear sine model by adding random effects for starting parasitaemia levels , parasite growth rate , and timing of the parasite cycle in each individual , and explicitly modeled observations under the qPCR detection limit ( see S1 Text for model details ) ., For the purpose of the power analysis we assumed that two full parasite cycles would be observed in each individual ( the log-linear sine model does not provide a prediction for termination of an experiment ) ., Other than that , this power analysis was executed in exactly the same fashion as that based on the main analysis ., None of the funders were involved in the writing of the manuscript or the decision to submit it for publication ., The authors have not been paid to write this article by a pharmaceutical company or other agency ., The corresponding authors had full access to all the data in the study and had final responsibility for the decision to submit for publication .

Introduction, Results, Discussion, Materials and Methods

Controlled human malaria infection ( CHMI ) in healthy human volunteers is an important and powerful tool in clinical malaria vaccine development ., However , power calculations are essential to obtain meaningful estimates of protective efficacy , while minimizing the risk of adverse events ., To optimize power calculations for CHMI-based malaria vaccine trials , we developed a novel non-linear statistical model for parasite kinetics as measured by qPCR , using data from mosquito-based CHMI experiments in 57 individuals ., We robustly account for important sources of variation between and within individuals using a Bayesian framework ., Study power is most dependent on the number of individuals in each treatment arm; inter-individual variation in vaccine efficacy and the number of blood samples taken per day matter relatively little ., Due to high inter-individual variation in the number of first-generation parasites , hepatic vaccine trials required significantly more study subjects than erythrocytic vaccine trials ., We provide power calculations for hypothetical malaria vaccine trials of various designs and conclude that so far , power calculations have been overly optimistic ., We further illustrate how upcoming techniques like needle-injected CHMI may reduce required sample sizes .

Controlled human malaria infection ( CHMI ) in healthy human volunteers is an important and powerful tool in clinical malaria vaccine development ., However , to obtain meaningful estimates of protective efficacy , it is important to include an appropriate minimum number of participants , while minimizing the risks and burden for volunteers ., Existing power calculations have limited value due to important influential assumptions ., To optimize power calculations for malaria vaccine trials , we developed a non-linear , Bayesian statistical model for parasite kinetics as measured by quantitative real-time polymerase chain reaction , using existing data from mosquito-based CHMI experiments ., Using our model , we provide improved , robust power calculations for various hypothetical malaria vaccine trials , taking account of important sources of variation between and within individuals ., We conclude that so far , power calculations for malaria vaccine trials have been overly optimistic ., We further illustrate how upcoming techniques like needle-injected CHMI may reduce required sample sizes .

medicine and health sciences, body fluids, parasite replication, immunology, tropical diseases, parasitic diseases, parasitic protozoans, parasitology, vaccines, preventive medicine, protozoans, mathematics, statistics (mathematics), infectious disease control, vaccination and immunization, public and occupational health, infectious diseases, malarial parasites, hematology, statistical models, blood, anatomy, physiology, biology and life sciences, malaria, physical sciences, organisms

journal.pgen.1001191

Endogenous Viral Elements in Animal Genomes

Viral infection of germ line cells ( i . e . gametes , or cells of the early embryo ) can lead to viral genes or genomes becoming integrated into chromosomes and inherited as host alleles 1 , 2 ., These insertions , which we refer to here as endogenous viral elements ( EVEs ) , are usually eliminated from the host gene pool within a small number of generations ., However , they can also increase in frequency , and some eventually reach fixation 3–11 ., In animal genomes , the majority of EVEs are derived from reverse transcribing RNA ( rtRNA ) viruses ( i . e . retroviruses ) 5 , 12 , 13 ., Retroviruses are the only animal viruses that integrate into the genome of the host cell as an obligate step in their replication strategy , and are thus predisposed to enter the host germ line ( Figure 1 ) ., EVEs derived from viruses that use other genome replication strategies also occur , but are much less common 6 , 7 , 9 , 11 , 14 , 15 ., Genomic integration of non-retroviral viruses may be mediated by non-homologous recombination with chromosomal DNA 16–18 or by interactions with retroelements in the host cell 11 , 19–22 ( Figure 1 ) ., EVEs reveal complex evolutionary relationships between viruses and their hosts ., For example , endogenous retroviruses have shaped vertebrate genome evolution , not only by acting as genetic parasites 23 , 24 , but also by introducing useful genetic novelty ., Indeed , the role of exapted retroviral genes ( i . e . integrated retroviral genes that have adapted to serve a function in the host genome ) in mammalian reproduction 25 , 26 identifies EVEs as a key factor in the evolution of placental mammals from egg-laying ancestors ., Similarly , in parasitoid wasps , genes derived from ancestral nudiviruses have been exapted to facilitate a parasitic lifestyle 9 ., These remarkable examples demonstrate an important role for gene flow from viruses to hosts in animal evolution ., EVEs also constitute an invaluable resource for reconstructing the long-term history of virus and host evolution 27 , 28 ., Viruses exhibit the potential for extremely high rates of nucleotide substitution , host switching , and lineage extinction , and this sets limitations on what can be reliably inferred from observations of contemporary isolates 29 , 30 ., EVE sequences effectively represent the ‘molecular fossils’ of ancient viral genomes , preserving information about ancient virus and host interactions that would otherwise be difficult , if not impossible , to infer ., For example , EVEs are subject to host rates of evolution and can thus be dated relatively reliably with molecular clock-based approaches , in which genetic divergence correlates linearly with time 31 ., In contrast , structural constraints in exogenous viruses may lead to the decoupling of short and long-term rates of viral evolution , rendering molecular clock assumptions unusable over longer timescales 30 , 32–34 ., Furthermore , the identification of orthologous EVE insertions allows the incorporation of independent age estimates based on host species divergences ( see Figure 1 ) 35 ., Despite the large quantity of published genome sequence data , the diversity of non-retroviral viruses in animal genomes has not been systematically explored ., In this report , we use an in silico approach to screen the genomes of mammals , birds and insect vector species for endogenous sequences derived from non-retroviral mammalian viruses ., We identify sequences derived from a very broad range of viruses , revealing an extensive history of non-retroviral genome invasion ranging back to at least the late Mesozoic Era ( ∼93 million years ago ) ., We demonstrate that these sequences can be highly informative;, ( i ) revealing novel virus diversity;, ( ii ) providing a timescale for virus evolution;, ( iii ) indicating the likely host range of virus groups , and;, ( iv ) identifying rare instances of horizontal transmission ., Furthermore , using a novel approach , we show that at least some of the EVE sequences identified here are likely to have been exapted during their evolution ., The implications of these findings are discussed ., An algorithm for in silico screening of genomes for endogenous non-retroviral insertions was developed ., We selected all non-retroviral virus genera that infect mammals , and constructed a library of representative peptide sequences ( restricted to viruses with typical genome sizes of <100 Kilobases ( Kb ) ) ( Table S2 ) ., The tBLASTn program was used to screen low coverage and complete genome assemblies for sequences exhibiting similarity to viral peptides in this library ., We screened the genomes of likely reservoirs ( birds , n\u200a=\u200a2 ) and vector species ( mosquitoes , n\u200a=\u200a3; ticks , n\u200a=\u200a1 ) as well as all available mammal genomes ( n\u200a=\u200a44 ) ( Table S1 ) ., Sequences that matched viral peptides with e-values <0 . 001 were extracted ( along with flanking sequences ) and putative protein sequences were inferred through a combination of automated and manual alignment ., These sequences were assigned to taxonomic groups ( family , genus ) based on the most closely related exogenous viral sequences in searches of PFAM and Genbank databases ( Tables S3 , S4 , S5 , S6 , S7 ) ., For EVEs that were found to encode uninterrupted open reading frames ( ORFs ) , putative protein sequences were used with the tBLASTn program to search expressed sequence tag ( EST ) databases for the corresponding mRNA ., For all EVEs disclosing similarity to contemporary virus isolates , putative EVE protein sequences were aligned with representative viral protein sequences , and maximum likelihood phylogenies were constructed ., We identified numerous , highly significant matches ( i . e . e-values <1×10−9 ) to RNA viruses in the genomes of mammals and insect vectors ( Table 1 , Tables S3 , S4 , S5 ) ., EVEs related to a total of seven families were identified including double stranded RNA ( dsRNA ) viruses ( Reoviridae ) and positive sense RNA ( RNA+ve ) viruses ( Flaviviridae ) , as well as both segmented ( Orthomyxoviridae , Bunyaviridae ) and non-segmented ( Borna- , Filo- and Rhabdoviridae ) families of negative sense RNA ( RNA-ve ) viruses ., Consistent with an integration process involving viral mRNA ( rather than genomic RNA ) , all EVEs derived from RNA viruses had genetic structures that spanned a single viral transcript ( or fragments derived from single transcripts ) ., EVEs derived from different genes never occurred as contiguous sequences , and consequently we could not determine whether EVEs derived from distinct genes of a given virus family originated from the same or distinct virus lineages/infections ., In mammals , matches to RNA virus proteins that spanned complete genes were typically flanked by target site duplications ( TSDs ) and 3′ poly-A tails , consistent with LINE-mediated retrotransposition of viral mRNAs 36 ., In insects , similar features were not apparent for any EVE insertion , even when the boundaries of host and viral sequences were clearly identifiable ( Figure S1 ) ., Notably , putative 3′ poly-A tails could be identified in the expected position for some mammal genome sequences that matched only weakly to RNA virus peptides , suggesting the presence of EVEs at the limit of detection to our search strategy ., We identified highly significant matches to three families of viruses with DNA genomes in the genomes of mammals and birds ( Table 1 , Tables S6 and S7 ) ., These included matches to two single stranded DNA ( ssDNA ) virus families ( Parvoviridae and Circoviridae - the first ssDNA virus EVEs to be described in mammals - and one family of reverse transcribing DNA ( rtDNA ) viruses ( Hepadnaviridae ) - the first rtDNA EVEs to be described ., A single match to a double stranded DNA ( dsDNA ) virus family ( Adenoviridae ) was identified in the kangaroo rat genome , but this sequence was unambiguously viral across its entire length ( ∼17 Kb ) , encoding thirteen completely intact viral ORFs ( Figure S2 ) , and is thus likely to have derived from free virus and not an EVE ., A subset of parvovirus-related EVEs represented complete or nearly complete viral genomes ( Figure 5a ) ., For one insertion in the M . lucifugus genome , we identified putative 5′ and 3′ terminal non-coding regions encoding characteristic inverted terminal repeats ( Figure S3 ) ., In general , however , DNA virus EVEs occurred as genomic fragments , with no particular region of the viral genome being obviously favored , with the exception of the circoviruses , for which only the Rep gene was found ., We identified a number of EVE insertions that were orthologous between species , allowing minimum ages for families to be inferred from host divergence dates ( see Figure 2 ) ., Using previously estimated mammalian divergence dates 42 we obtained minimum ages for the Parvo , Circo and Bornaviridae of 30 , 68 and 93 million years respectively , demonstrating the ancient origins of these families ( Figure 6 ) ., During completion of this manuscript , orthologous filovirus EVEs were reported in the mouse and rat genomes 22 ., These sequences were identified by BLAST searching using EVEs as probes , and were not picked up in our screen , which relied on matches to exogenous viruses ., On the basis of the mammalian divergence dates used here 42 , these EVEs provide a minimum age of 30 million years for the Filoviridae ( Figure 6 ) ., The EVEs identified in this study extend the host range of several families ( Parvo- , Circo- , Hepadna , Borna- and Filoviridae ) with respect to their known range as exogenous viruses ( Figure 6 ) ., Dependovirus EVEs are particularly widespread and occur in diverse mammalian hosts , despite their apparent low probability of germ line integration in AAV-derived gene therapy vector in vivo models 43 ., Filoviruses have only been identified as exogenous infections in bats and primates 44 ., However , filoviruses EVEs were identified not only in North American bats ( M . lucifugus ) and Asian primates ( tarsier ) , but also in insectivores , rodents , and in both South American and Australian mammals ( Figure 6 ) ., In concordance with the recent identification of Ebola Reston in swine 45 , this unexpected result indicates that the distribution of filoviruses is likely much broader than has previously been recognized ., Highly discordant host ranges among closely related EVEs ( or EVEs and exogenous viruses ) can provide information about transmission events ., In this regard , we note that a dependovirus EVE in the bottlenose dolphin ( Tursiops truncatus ) genome grouped robustly with avian dependoviruses ( rather than mammalian isolates ) in NS1 trees ( Figure 6d ) , suggesting cross-class transmission of parvoviruses between birds and mammals may have occurred in the past ., EVEs that are neutral or only slightly deleterious in their hosts may fortuitously drift or hitchhike 46 to fixation , accumulating mutations at the host neutral rate ., Alternatively , EVE insertions may confer an advantageous phenotype on the host and spread through the population by selection ., In such exapted sequences , selection will act to maintain the functionality of the EVE sequence ., Many of the EVEs identified in this study were highly mutated and/or fragmented and these likely represent non-functional , neutrally evolving pseudogenes ., However , several EVEs encoded intact ORFs , and some also express RNA ( Figure 2a , Figure 3a , Figure 5a ) ., For most of these EVEs , the time since insertion is unknown , and intact ORFs could reflect recent insertion rather than a long-standing history of purifying selection within the host genome ., In primates , however , orthology of the bornavirus-derived insert EBLN-1 , which is intact in several species , demonstrates an insertion date predating the divergence of strepsirhine primates ( ∼54 million years ago ( MYA ) ) ( Figure 7 ) ., Simulations in which a consensus derived from all EBLN-1 sequences was allowed to neutrally evolve over this time period indicated the probability of maintaining an intact ORF in the absence of purifying selection was <0 . 00001 ( 100 , 000 replicates , mean number of stop codons =\u200a15 . 57 , 95% confidence range 7 . 9–23 . 3 ) ., This analysis provides more robust support for purifying selection than classical tests based on the ratio of synonymous to non-synonymous mutations ( which are weakly significant for EBLN-1 11 ) , strongly indicating that EBLN-1 has been exapted in the primate genome , at least during part of its evolutionary history ., Curiously , however , EBLN-1 has not retained coding capacity in all primate species ., Perhaps selection to maintain it has recently been lost across all primates , and all the inserts may become inactivated in future ., In this report , systematic screening revealed that sequences derived from a broad range of non-retroviral mammalian virus groups occur as endogenous elements in the genomes of mammals , birds and insect vectors ., We describe the first EVEs derived from the rtDNA and dsRNA groups , thereby establishing that the complete range of known animal virus replication strategies ( see Figure 1 ) are represented by endogenous elements in animal genomes ., Richer sampling of animal genomes is likely to reveal an even greater diversity of EVEs ., While EVEs that are very ancient ( i . e . that inserted prior to the divergence of major host lineages ) can be identified by selectively screening a small number of host species , identification of more recent insertions will often require richer sampling within orders and genera ., Sampling of mammalian species for whole genome sequencing has generally been across , rather than within orders ( primates are an exception ) ., Consequently the majority of mammal species sampled in this study diverged more than 50 million years ago ( Figure 6 ) ., Any mammal species that was not sampled , and diverged more recently , could contain uncharacterized EVEs ., Sampling of avian and insect vector genomes has so far been quite limited , and these may also harbor a rich virus fossil history ., Furthermore , the vast majority of EVE insertions never reach fixation , and there are likely many unfixed EVEs present within species gene pools at a given time ( known examples of unfixed EVEs include Israeli acute paralysis virus ( IAPPV ) insertions in honey bees ( Apis mellifera ) 15 , koala endogenous retrovirus ( KoRV ) in koala bears 47 , and human herpesvirus 6 ( HHV-6 ) and HERV-K HML-2 insertions in humans 18 , 48 ) ., Identification of such unfixed EVEs will often require population-level screening ., The in silico screening strategy employed here likely underestimates the actual diversity of EVEs for several reasons ., Firstly , only low-coverage , incomplete genome data were available for most species ., Furthermore , EVEs within the data we screened could have been overlooked because, ( i ) screening was based on similarity searches , and is thus dependent on current ( limited ) knowledge of viral diversity , and, ( ii ) more ancient EVEs may not be identified due to the divergence in both host and virus lineages subsequent to insertion ( this may also result in a bias toward detecting more conserved genes ) ., Certain groups of ( non-retroviral ) viruses appear to be better represented in the genomic fossil record than others ( e . g . Parvoviridae , Mononegavirales ) ., This likely reflects a predisposition for germline integration among viruses with particular patterns of replication and infection ., Notably , viruses that establish persistent infections and/or replicate within the nucleus are particularly well represented among the EVEs identified in this study ., Nevertheless , these characteristics do not appear to be prerequisites for germ line integration ( Table 1 ) ., Indeed , since retroelements are ubiquitous in animal genomes , and replication of all known viruses requires the expression of RNA , retroelement activity in germ line cells 49 may present a general mechanism for mediating insertion of virus genes into animal germ lines ( see Figure 1 ) ., The discovery that a broad range of viruses are represented by EVEs in animal genomes indicates that viral ‘molecular fossils’ can provide the basis for robust , time-scaled , macroevolutionary studies across a range of animal and virus groups ., For example , EVE sequences can be combined with phylogenetic data of extant host species to reveal patterns of inter-class virus transmission ( Figure 5 ) 50 ., In this study , orthologous EVEs derived from the Borna- , Filo- Circo- , and Parvoviridae provided direct evidence for the ancient origins of these families ( Figure 6 ) ., These findings also indicate that more recent dates of origin obtained for other virus families using molecular clock-based extrapolations are artifacts 30 ., The diversity represented by known virus isolates represents a tiny fraction of the total viral diversity ., Indeed , given their likely ancient origins , many virus families may be broadly distributed across mammalian hosts ., This was reflected in viral phylogenies containing a mixture of EVEs and exogenous viruses - closely related exogenous relatives could often not be identified , or had only been recently characterized 37 , 38 , 51 ( Figure 2 , Figure 3 , Figure 5 ) ., These findings suggest that EVEs can inform viral surveillance efforts by revealing novel virus diversity and indicating the likely host range of virus groups ( particularly if they inserted relatively recently ) ., For example , a strong association between filoviruses and marsupials ( Table 1 , Figure 2 ) unexpectedly highlighted this group as a potential filovirus reservoir ., The potential presence of EVEs may also be an important consideration in studies where bulk sequencing of environmental samples is used to identify novel virus groups 51–53 ., EVEs that reach fixation in the host germ line may do so fortuitously , or because they are exapted by the host genome ., Monte Carlo simulations provided robust statistical support for a history of purifying selection in the primate EVE EBLN-1 , indicating this sequence has been exapted by the primate genome ., However , selection on EBLN-1 has clearly relaxed in some primates and may also have relaxed in humans ( Figure 7 ) ., Such transient co-option may be expected for EVEs that function as restriction factors in their hosts by conferring resistance to infection by exogenous viruses ., Several examples of this phenomenon have been described in animals 15 , 54 , 55 , and it is likely one of the most common exaptations of viral genes by host organisms 56 , 57 ., In these cases , counter-adaptation in a rapidly evolving virus population may eventually render the EVE restriction mechanism non-functional 55 , causing selection to relax ., Importantly , the rate at which EVEs are exapted as restriction factors in animals could greatly exceed their rate of fixation in animal genomes ., The diverse EVE sequences described in this report demonstrate an extensive history of gene flow from virus to animal genomes ., Animal genomes are a living document of virus and host interaction , and genomic studies have an important role to play in advancing understanding of virus and host evolution ., Chromosome assemblies and whole genome shotgun assemblies of 44 species ( Table S1 ) were screened in silico using tBLASTn and a library of representative peptide sequences derived from mammalian virus groups with genomes <100 Kb in total length ( selected from the 2009 International Committee on Taxonomy of Viruses ( ICTV ) master species list ( Table S2 ) ) ., Host genome sequences spanning high-identity ( i . e . e-values <0 . 0001 ) matches to viral peptides were extracted , and a putative viral ORF was inferred using BlastAlign 58 and manual editing ., Putative EVE peptides were then used to screen the Genbank non-redundant ( nr ) database in a reciprocal tBLASTn search ., Matches to retroviruses , viral cloning vectors , and non-specific matches to host loci were filtered and discarded ., The remaining sequences were considered viral if they unambiguously matched viral proteins in the Genbank and PFAM databases as shown in Tables S3 , S4 , S5 , S6 , S7 ., Genetic structures for these elements were determined by comparison of the putative EVE peptide sequence to the nucleotide sequence of a viral type species representing the most closely related viral genus recognized by ICTV ., Boundaries between viral and genomic regions were identified by analysis of sequences flanking matches to viral peptides , the genomes of the host species , and closely related host species ., Sequences that flanked viral insertions were considered genomic if they;, ( i ) were present as empty insertion sites in a related host species;, ( ii ) disclosed highly significant similarity ( i . e . e-values <1×10−9 ) to host proteins; or, ( iii ) non-viral and highly repetitive ( >50 copies per host genome ) ., Insertions were considered endogenous when >100 bp of genomic flanking sequence could be identified either side of a viral match ., Insertions for which >100 bp of unambiguous ( i . e . >80% nucleotide identity ) flanking sequence was identified in host sister taxa were considered orthologous insertions ., PERL scripts were used to automate BLAST searches and sequence extraction ., Putative EVE peptide sequences , and alignments of EVEs and exogenous retroviruses , are available online ( http://saturn . adarc . org/paleo/ ) ., Putative EVE sequences inferred using BlastAlign were aligned with closely related viruses using MUSCLE and manually edited 59 ., Maximum likelihood ( ML ) phylogenies were estimated using amino acid sequence alignments with RAXML 60 , implementing in each case the best fitting substitution model as determined by ProtTest 61 ., Support for the ML trees was evaluated with 1000 nonparametric bootstrap replicates ., The best fitting models for the datasets were: Parvoviridae: dependovirus NS1 gene ( JTT+Γ , 332 amino acids across 17 taxa ) , Parvoviridae: parvovirus NS1 gene , ( JTT+Γ , 293 amino acids across 13 taxa ) , Circoviridae: Rep gene ( Blosum62+Γ+F , 235 amino acids across 14 taxa ) , Hepadnaviridae: polymerase gene ( JTT+Γ+F , 661 amino acids across 9 taxa ) , Orthomyxoviridae: GP gene ( WAG+Γ+F , 482 amino acids across 5 taxa ) , Reoviridae: VP5 gene ( Dayhoff+Γ+F , 171 amino acids across 4 taxa ) , Bunyaviridae: phlebovirus NP gene ( LG+Γ , 247 amino acids across 12 taxa ) , Bunyaviridae: nairovirus NP gene ( LG+Γ , 446 amino acids across 5 taxa ) , Flaviviridae: mostly NS3 gene ( LG+Γ+F , 1846 amino acids across 8 taxa ) , Filoviridae: NP gene ( JTT+Γ , 369 amino acids across 29 taxa ) , Filoviridae: L gene ( LG+Γ+F , 517 amino acids across 9 taxa ) , Bornaviridae: NP gene ( JTT+Γ , 147 amino acids across 73 taxa ) , Bornaviridae: L gene ( JTT+Γ+F , 1243 amino acids across 12 taxa ) , Rhabdoviridae: NP gene ( LG+Γ , 220 amino acids across 34 taxa ) , Rhabdoviridae: L gene ( LG+Γ+F , 383 amino acids across 26 taxa ) ., A Monte Carlo simulation procedure was employed to determine the probability that the bornavirus-derived element EBLN-1 has retained coding capacity over 54 . 1 million years under neutral evolution ( i . e . not under purifying selection ) ., A consensus EBLN-1 sequence was inferred , and the effects of neutral evolution were simulated using seq-gen 62 for a branch length equivalent to the minimum amount of time that EBLN-1 orthologs have resided in primate genomes , based on the primate divergences estimated by Bininda-Emonds et al 42 , and given a neutral rate of evolution of 2 . 2×10–9 12 ., The number of stop codons accrued was counted for 100 , 000 iterations of the simulation ., The probability that the reading frame could have remained open under neutrality is given by the number of replicates under which no stop codons have evolved , divided by the number of iterations ., Parvoviridae; AAV2 ( NC_001401 ) ; Minute virus of mice ( NC_001510 . 1 ) ; AMDV ( NC_001662 ) ; Goose parvovirus ( EU583390 . 1 ) ; Muscovy duck parvovirus ( X75093 . 1 ) ; Porcine hokovirus ( EU200671 . 1 ) ; Snake parvovirus ( AY349010 . 1 ) ; Avian AAV ( AY629582 . 1 , AY629583 . 1 , GQ368252 . 1 ) ; AAV1 ( AF063497 . 1 ) ; AAV4 ( U89790 ) ; AAV2 ( AY695375 . 1 ) ; Bovine AAV ( AY388617 . 1 ) ; Caprine AAV ( DQ335246 . 2 ) ; Bocavirus ( M14363 . 1 ) ; Erythrovirus ( AB126265 . 1 ) ; Aleutian mink disease virus ( M20036 . 1 ) ; Porcine parvovirus ( EU790642 . 1 ) ; Feline panleukopenia virus ( EF988660 . 1 ) ; Canine parvovirus ( EU310373 . 2 ) ; Rat parvovirus ( AF036710 . 1 ) ; Hamster parvovirus ( U34255 . 1 ) ; Minute virus of mice ( DQ196317 . 1 ) ; Kilham rat virus ( U79033 . 1 ) ; Circoviridae; Porcine circovirus 1 ( NC_006266 ) ; Porcine circovirus 2 ( GU325757 ) ; Cyclovirus PK5006 ( GQ404856 . 1 ) ; Cyclovirus NG14 ( GQ404855 . 1 ) ; Human stool-associated circular virus NG13 ( GQ404856 . 1 ) ; Beak and feather disease virus ( AY450436 . 1 ) ; Columbid circovirus ( AF252610 . 1 ) ; Hepadnaviridae; duck HBV ( NC_001344 ) ; Stork HBV ( AJ251937 . 1| ) ; Heron HBV ( NC_001486 ) ; Ross Goose HBV ( AY494849 . 1 ) ; Crane HBV ( AJ441113 . 1 ) ; Sheldgoose HBV ( AY494852 . 1 ) ; Snow goose HBV ( AF111000 . 1 ) ; Woodchuck HBV ( AF410861 . 1 ) ; Flaviviridae; Kamiti river virus ( NC_005064 ) ; Aedes flavivirus ( NC_012932 ) ; Quang binh virus ( NC_012671 ) ; Culex flavivirus ( NC_008604 ) ; Nakiwogo virus ( GQ165809 ) ., Reoviridae; Liaoning virus ( NC_007736 - NC_007747 ) ; Kadipiro virus ( NC_004199 , NC_004205-NC_004210 , NC_004212-NC_004216 ) ; Banna virus ( NC_004198 , NC_004200-NC_004204 , NC_004211 , NC_004217-NC_004221 ) ., Bunyaviridae; Crimean-Congo hemorrhagic fever virus ( NC_005300 , NC_005301 , NC_005302 ) ; Uukuniemi virus ( NC_005214 , NC_005220 , NC_005221 ) ; Uukuniemi virus ( M33551 . 1 ) ; Catch-me-cave virus ( EU274384 . 1 ) ; Sandfly fever Naples virus ( EF201832 . 1 ) ; Massilia virus ( EU725773 . 1 ) ; Punta Toro virus ( EF201834 . 1 ) ; Buenaventura virus ( EF201839 . 1 ) ; Rift Valley fever virus ( DQ380156 . 1 ) ; Phlebovirus sp ., ( EF201818 . 1 ) ; Icoaraci virus ( EF076014 . 1 ) ., Orthomyxoviridae; Quaranfil virus ( FJ861694 . 1 ) ; Johnston Atoll virus ( FJ861696 . 1 ) ; Thogoto virus ( M77280 . 1 ) ; Dhori virus ( M34002 . 1 ) ., Bornaviridae; Borna disease virus ( NC_001607 ) ; Avian BDV ( FJ169441 ) ., Filoviridae; Reston ebola virus ( NC_002549 ) ; Zaire ebola virus ( NC_002549 ) ; Lake Victoria marburgvirus ( NC_001608 ) ., Rhabdoviridae; vesicular stomatitis virus ( NC_001560 ) ; Wongabel virus ( NC_011639 ) ; Kotonkon virus ( DQ457099 ) ; Adelaide river virus ( U10363 . 1 ) ; Obodhiang virus ( DQ457098 . 1 ) ; Bovine ephemeral fever virus ( AF234533 . 1 ) ; Rochambeau virus ( DQ457104 . 1 ) ; Mount elgon bat virus ( DQ457103 . 1 ) ; Oita rhabdovirus ( AB116386 ) ; Kern canyon virus ( DQ457101 . 1 ) ; Sandjimba virus ( DQ457102 . 1 ) ; Kolongo virus ( DQ457100 . 1 ) ; Tupaia rhabdovirus ( AY840978 . 1 ) ; Spring viremia of carp ( DQ491000 . 1 ) ; Pike fry rhabdovirus ( FJ872827 . 1 ) ; Cocal virus ( EU373657 . 1 ) ; Vesicular stomatitis Indiana virus ( AF473865 . 1 ) ; Isfahan virus ( AJ810084 . 2 ) ; Chandipura virus ( AY614728 . 1 ) ; Ngaingan virus ( FJ715959 . 1 ) ; Wongabel virus ( EF612701 . 1 ) ; Flanders virus ( AF523194 . 1 ) ., Nyaviridae; Midway virus ( NC_012702 ) ; Nyamanini virus ( NC_012703 ) .

Introduction, Results, Discussion, Materials and Methods

Integration into the nuclear genome of germ line cells can lead to vertical inheritance of retroviral genes as host alleles ., For other viruses , germ line integration has only rarely been documented ., Nonetheless , we identified endogenous viral elements ( EVEs ) derived from ten non-retroviral families by systematic in silico screening of animal genomes , including the first endogenous representatives of double-stranded RNA , reverse-transcribing DNA , and segmented RNA viruses , and the first endogenous DNA viruses in mammalian genomes ., Phylogenetic and genomic analysis of EVEs across multiple host species revealed novel information about the origin and evolution of diverse virus groups ., Furthermore , several of the elements identified here encode intact open reading frames or are expressed as mRNA ., For one element in the primate lineage , we provide statistically robust evidence for exaptation ., Our findings establish that genetic material derived from all known viral genome types and replication strategies can enter the animal germ line , greatly broadening the scope of paleovirological studies and indicating a more significant evolutionary role for gene flow from virus to animal genomes than has previously been recognized .

The presence of retrovirus sequences in animal genomes has been recognized since the 1970s , but is readily explained by the fact that these viruses integrate into chromosomal DNA as part of their normal replication cycle ., Unexpectedly , however , we identified a large and diverse population of sequences in animal genomes that are derived from non-retroviral viruses ., Analysis of these sequences—which represent all known virus genome types and replication strategies—reveals new information about the evolutionary history of viruses , in many cases providing the first and only direct evidence for their ancient origins ., Additionally , we provide evidence that the functionality of one of these sequences has been maintained in the host genome over many millions of years , raising the possibility that captured viral sequences may have played a larger than expected role in host evolution .

virology/virus evolution and symbiosis, evolutionary biology/paleontology, computational biology/genomics

journal.pcbi.1004309

A Computational, Tissue-Realistic Model of Pressure Ulcer Formation in Individuals with Spinal Cord Injury

Pressure ulcers ( PU ) affect 2 . 5 million US acute care patients and cost up to $1 billion per year 1 ., They are a significant source of morbidity in both hospitalized patients and community-dwelling individuals with impaired mobility ., PUs are especially common in individuals with spinal cord injury ( SCI ) , occurring in up to 80% of this population at some point during their lifetime 2 ., Spinal cord injury is a condition associated with decreased functional mobility , acutely increased oxidative activity in leukocytes , and chronic elevation of systemic inflammatory markers 3–5 ., Pressure ulcers are thought to arise from pressure-induced ischemia , reperfusion injury , and/or deformation-induced cellular damage 6 ., The pathogenesis of PU involves activation of the acute inflammatory response 7 , 8 , a highly conserved cascade of events mediated by a set of specialized cells ( e . g . platelets , mast cells , macrophages , and neutrophils ) and molecules ( inflammatory cytokines , free radicals , and Damage-Associated Molecular Pattern molecules DAMPs ) that demarcate stressed or damaged tissue , and alert and recruit other cells and molecules ., The inflammatory response can either restore the tissue to equilibrium ( healing ) and resolve , or become self-maintaining inflammation that causes and is caused by ancillary tissue damage ., This excessive inflammation prevents the body from initiating wound healing and can lead to PU incidence 9 ., The dynamic interplay of the inflammatory and healing cascades determines the ultimate success or failure of the healing process ., These intracellular signaling networks and their products , including diffusible molecular mediators , are possible targets for diagnosis or therapeutic intervention ., However , the complexity of the process as a whole , and the dependence of any given pathway or mediator on timing and context , complicates such translational approaches ., The “translational dilemma” centers on the inability of traditional , reductionist approaches to yield better diagnostics and novel drug targets for complex diseases 10–12 ., Despite increased understanding of the underlying mechanisms and improved clinical vigilance , PUs remain a prevalent problem in hospitalized patients and people with chronic conditions such as diabetes and SCI 13–15 ., While wound healing is well studied in animal systems 16–18 , these animal models generally do not recapitulate the complex etiology of impaired wound healing such as that which occurs in PU ., Only recently have experimental methodologies emerged that may allow for the study of the time courses of wound healing in humans 17 , but these approaches are limited in that time courses of primary samples from humans with chronic wounds are difficult to collect without disturbing the very process being measured ., As an alternative diagnostic approach , digital photographs of developing ulcers are , in theory , both plentiful and non-invasive—unlike wound biopsies ., Given the dire need for new therapeutic avenues for complex diseases , a platform combining in silico approaches with easily- and inexpensively-obtained clinical samples ( such as photographic images ) , may yield novel diagnostic and therapeutic modalities ., We hypothesize that wound images could be used to calibrate mechanistic simulations of inflammation and healing that could then be interrogated to predict how and when a small irritation might resolve or progress to become a chronic ulcer ., Agent-Based Models ( ABMs ) allow for the investigation of both space- and time-dependent dynamics of complex systems via mechanistic simulations ., The modeler provides behavioral rules that allow the model to proceed stepwise through discrete space and time ., Unlike differential equation models , ABM simulations are stochastic and thereby able to replicate the randomness of biological processes ., Furthermore , ABMs produce visual outputs whose morphological features evolve throughout a simulation and provide a rich set of spatio-temporal data that can be leveraged to probe underlying dynamics 19 ., We report herein on the creation of an ABM of post-SCI PU formation ( the Pressure Ulcer Agent Based Model , or PUABM ) that incorporates key inflammation mechanisms ., Explicitly included is the forward feedback loop of inflammation to damage to inflammation that has served as the core motif of our prior simulations of inflammation in both systemic and local contexts 9 , 20 , 21 ., The PUABM replicates visual morphology associated with the development and resolution of post-SCI PU by simulating vascularized soft tissue overlaying a bony prominence ( the clinically recognized “pressure points” at which PU typically develop ) and the effects of repeated ischemia/reperfusion ( representing the turning of a person with SCI in bed ) on such an area of tissue ., Also recapitulating clinical outcomes , the model reaches two distinct endpoints when simulated from the same initial parameters ., We leverage thousands of model simulations to explore the root of this phenomenon and predict at what time an ulcers fate is determined ., We also demonstrate the utility of the PUABM as a platform for in silico clinical trials of strategies for prevention and therapies for treatment of PU post-SCI ., This model suggests that the reason treatments thus far have been ineffective is that they have been applied too late ., This study was approved by the University of Pittsburgh Institutional Review Board , approval number PRO08010011 ., Written informed consent was obtained from subjects participating in the study ., If the subject was unable to provide written consent due to medical condition , proxy consent was obtained from the subject’s authorized representative , and the subject provided written consent for continued research participation as soon as they were able to do so ., Photographic data were obtained from pressure ulcers sustained by patients enrolled in the Rehabilitation Engineering Research Center on Spinal Cord Injury at the University of Pittsburgh ., Subjects were recruited from a single tertiary care center if they were 18 years of age or older and had sustained a traumatic spinal cord injury ., Subjects were excluded if they had a history of pre-existing diseases that affect the inflammatory response to SCI ( e . g . autoimmune disease , demyelinating diseases ) or a history of previous SCI or other neurological diseases affecting motor and sensory function ., Pressure ulcers were initially photographed when identified , and serial photographs were taken three times per week while patients remained in the acute care and/or weekly while in the inpatient rehabilitation hospital to monitor progression and/or healing , and during their outpatient visits or 6 month intervals after discharge ., Photos were obtained using a Canon Power Shot SD 750 camera with 3X optical zoom lens with flash enabled , at a distance of 12 inches from the ulcer , resolution of 2048 x 1536 pixels ., Pressure ulcer site , severity ( stage 1–4 , unstageable , or deep tissue injury ) , general appearance , size and shape were also recorded ., The PUABM was built using an iterative approach ( Fig 1A ) ., Hereafter , when we use words or phrases that usually specify biological reality , such as “pressure , ” “neutrophil , ” “TNF-α , ” “inflammatory mediators , ” “cell , ” “oxygen , ” “ischemic , ” etc . we are referring to PUABM components or generated phenomena unless clearly stated otherwise ., A simpler model of pressure ulcer formation 22 was altered to increase mechanistic detail and create clinically relevant model output ., Rational improvements were based on domain knowledge and data from the literature ., First , the area of tissue simulated in the field was extended ., Whereas in the previous version of the model pressure was applied evenly across the entire field , in this version , it was applied maximally ( value determined by the parameter , pressure-intensity ) to a circular area in the center of the field and decreasing radially outward ., This allowed representation of pressure over a bony prominence and the surrounding tissue , which also experiences pressure , but to a lessening degree 23 , 24 ., Next , the PUABM code was altered so that neutrophils and macrophages enter the tissue in a resting state and can be activated by one of two circulating mediators ( TNF-α or TGF-β1 ) ., In the previous version of the model , all entering leukocytes were in the active state ., After this modification , the activation state of neutrophils and monocytes/ macrophages was determined by the local concentrations of inflammatory mediators ., These thresholds are explained further in the section below entitled , “Rules: Tissue Damage . ”, In the previous version of the ABM , all cells and mediators appeared in a single visualization window ., Epithelial cells ranged in color from green to red , depending on their level of health ., Other cells were rendered in layers over the grid of epithelial cells22 , 25 ., In the PUABM presented herein , circulating inflammatory cells and mediators were each removed to a separate viewing window , allowing for comparison of spatio-temporal pattern of each individual component with any other ., In the main window , stationary epithelial cells remained , but the colors indicating tissue health were altered to increase the realism: healthy tissue was now rendered as peach , mimicking the most common skin tones in the clinical cohort ., Unhealthy tissue continued to be rendered as red , and cells that died disappeared from the grid , leaving behind a white empty space ., Thus , each component of the model was represented according to the rules in the ABM , but by separating the viewing windows , we were able to view and compare spatial details of individual components ( damage , mediators , etc . ) ., Furthermore , we were able to compare epithelial damage patterns between simulations and clinical images , as described in later sections ., Mechanistic details in the PUABM were augmented by incorporating damage to tissue via ischemic injury and its counterpart , reperfusion injury ., Damage directly resulting from pressure ( as previously simulated 22 , 25 ) was removed and tissue health was penalized for tissue cells receiving inadequate oxygen levels ., ( In the previous version , oxygen had positive effects on tissue health , but lack of oxygen was not damaging 22 ) ., A mechanism based on the conversion of xanthine dehydrogenase to xanthine oxidase during ischemia 26 , 27 was also implemented ., The accumulation of xanthine oxidase in ischemic cells represents the potential of a cell to experience reperfusion injury , due to formation of ROS when oxygen reperfuses 26 ., The complexity of the inflammatory response was increased by creating two subpopulations of activated macrophages , one with pro-inflammatory ( M1 macrophages ) and another with anti-inflammatory ( M2 macrophages ) phenotype 28 , 29 ., A fourth mediator was also added ., Representing a canonical later-acting pro-inflammatory mediator , it is released by pro-inflammatory macrophages and labeled IL-1β ., After the mechanisms comprising the PUABM were set , model behaviors were explored over wide ranges of parameter values and default values were tweaked into ranges producing behavior that qualitatively matched the clinical data ., Finally , we further investigated dynamics encoded in the model by simulating hypothetical and existing therapies ., Corticosteroids were incorporated as a data layer that could be introduced to the tissue via blood vessels ( as in an intravenous injection ) , while antibodies to DAMPs were simulated as a topical cream applied to the entire field ., The dose and timing of administrations of each were varied to assess the viability of these treatment options ., The PUABM was built using SPARK , a platform designed for agent-based modeling of biological systems , freely available for download at: www . pitt . edu/~cirm/spark 25 , 30 , 31 ., SPARK models are written in a logo-like language called SPARK-PL and run on a java platform 32 ., SPARK models contain Agents—autonomous entities that interact with each other and the environment , called Space—and Data Layers , corresponding to individual species in the environment that can diffuse , etc ., The behavior of the model is determined by rules that govern how and when agents interact and react ., These rules are generally written to be interpreted by one agent at a time , and therefore are necessarily restricted in scope ( both time and space ) ., A rule specifies how much an agent should move , produce , change an internal variable , etc . when it encounters a certain amount or type of data layer or agent in its immediate neighborhood ., Model rules can be probabilistic in nature , allowing the model to evolve in a stochastic manner ., Therefore , the behaviors and patterns produced by simulating several ticks ( time steps ) of the model in succession arise as emergent phenomena resulting from the accumulated actions of a population of agents over time ., SPARK has several built-in standard methods that allow for convenient coding of common biological processes ., For example , diffusion is encoded by the function diffuse , which implements a simple discrete approximation in which each data layer cell shares a given percentage of its value with its eight neighbors ., Examples of other methods that were used in this model include wiggle and jump to approximate undirected random movement and sniff , for chemotaxis ., See pseudocode ( S1 Text ) for all methods used ., The components of the PUABM are described in Table 1 , and the operant biological mechanisms and tissue structures are shown graphically in Fig 1B ., In the PUABM , simulation time is linked to actual time using the lifespans of cellular agents ., A simulated macrophage lifespan ranges from 100–150 model time steps ( ticks ) , corresponding to 100–150 hours ( 4–6 days ) of real time ., Neutrophil lifespans range from 10–20 ticks ( hours ) in the tissue , but increase when neutrophils become activated ., Each simulation takes roughly five minutes to compute on a supercomputing node containing 32 process cores , each running at 4 . 7 GHz , making it possible to complete approximately 1000 simulations per day ., The model contains 68 numerical parameters that are set by the modeler and 11 random variables ( whose values are drawn from a corresponding uniform distribution when necessary during the course of a simulation ) ., At any given tick , the maximum number of agents computed in the model is on the order of 105 ., Pressure was a key factor in simulations of the PUABM ., In simulations of the acute inflammatory response in the absence of pressure , as in Fig 2C and 2D , the differences in outcomes were most dramatic , as they either did or did not result in an ulcer ., In contrast , when pressure was added , the differences in predicted outcome were more subtle: an ulcer always formed , but it was associated with varying degrees of overall damage ., To better define the mechanisms influencing this bifurcated outcome , we initially carried out simulations of acute inflammation in the absence of pressure ., Acute inflammatory dynamics were incited by an initial injury to the center of the tissue ., We first varied the intensity of the initial injury in order to determine over what range of injury both outcomes persisted ., Because we found that the frequency of ulcerative inflammation was correlated to the intensity of initial injury , we focused on simulations with a 30% initial injury—the level at which 50% of simulations resolved and 50% formed an ulcer ( see S2 Fig ) ., We approximated the distributions of total tissue damage in PUABM simulations using a Gaussian Mixture Model fit using the Expectation-Maximization algorithm , as implemented in the Matlab function , gmmfit ., We repeated this process varying the number of Gaussians in the model from 1 to 4 , and then compared the goodness-of fit for each one , using Aikikae and Bayesian Information Criteria ., This step allowed us to determine quantitatively the number of underlying Gaussian models that give rise to the pattern we see , an important step when the total damage from each outcome does not vary significantly ., We selected the model that was a mixture of two independent Gaussian distributions because this model yielded the lowest scores for both Aikikae Information Criteria ( AIC ) and Bayesian Information Criteria ( BIC ) , shown in S2 Table ., Following the work of Xing et al . 37 , we employed a 1-nearest neighbor ( 1NN ) approach to automatically segregate simulations ending in ulcers from those that displayed resolving inflammation ., The training set consisted of data from 100 simulations , labeled according to which endpoint was reached ( resolved or ulcerated ) ., For each simulation in the test set , a pairwise distance was computed between itself and every simulation in the training set ., The unlabeled test sequence was assigned the same class label as the training sequence that was closest in distance to it , its “nearest neighbor . ”, This method relies heavily on the choice of distance metric , in this case the Euclidean distance between sequences ., To take advantage of the time dependence of the features , we considered each simulation to be a sequence , wherein every entry in the sequence was a time point from a simulation ., Each of those entries consisted of either a single value ( e . g . total oxygen at tick t ) or a 13-dimensional vector of model component values ., For univariate time series , we calculated the Euclidean distance between each pair of vectors consisting of measurements of a single feature through time ., For multivariate time series , we calculated the Euclidean distance between two feature vectors at each time step and took the Euclidean norm of those distances to be the distance between the two simulations ., To equalize the contributions of all features , their values were normalized to fall into similar ranges before calculating distance ., We next sought to define the model parameters that most affected simulation outcomes in the PUABM ., Because there are more than 50 free parameters in the model , it was impractical to examine the sensitivity of key model outputs to all of parameter space at once ., Instead , modules of rules in the model that impacted tissue health were identified and used to create groups of parameters ., Parameters within these groups were then prioritized according to the mathematical degree of their effect on the system ., For example , a parameter that sets the value of an exponent produces a more dramatic effect than one that sets a scalar multiple ., Therefore , the first parameters varied were those controlling threshold values ., Simulations initially varied parameters over coarse-grained and then finer-grained threshold value ranges ( 100 simulations per parameter value ) ., Total tissue damage at time t served as a quantitative output measure ., From this analysis , a sensitivity index could be calculated for each parameter , taking the ratio of change in damage to change in threshold value ., A second level of sensitivity analysis was designed to examine the interplay between two potentially related parameter values ., This analysis allowed for a direct comparison of the sensitivities of two parameters , and also revealed any secondary effects that occurred when the two parameters changed in a combinatorial way ., Parameters were chosen from the same “damage module” in order to get a sense of the relationships between “sub-processes” in each module ., Snapshots of the tissue layer in the model at a fixed time point served as the output in order to assess overall damage , presence and size of ulcer , and other qualitative features ., Treatments with corticosteroids or neutralizing anti-DAMP antibodies were simulated as in silico clinical trials ., These trials consisted of sets of model simulations in which parameters controlling drug dose and timing and tissue response were varied , each independently ., Corticosteroid administration was simulated as an intravenous injection ., Therefore , steroid molecules ( implemented as a data layer ) were introduced to the tissue via blood vessels ( and were restricted when pressure was applied ) ., The mechanism of steroid action was to kill inflammatory cells , regardless of their state ( active/ resting ) , as illustrated in S6 Fig . When neutrophils were killed , additional ROS was released by the dying cell ( see S1 Text ) ., Anti-DAMPs were simulated as a topical cream administration ., A uniform layer of this molecule was introduced onto the field as a data layer at the tick specified by the parameter designating time of onset ., The method of action of the antibodies was controlled by a quenching reaction , wherein local concentrations of DAMPs were reduced by an amount proportional to the smaller concentration of the two molecules present: antibody or DAMPs ( see S1 Text for pseudocode and also S7 Fig ) ., The PUABM represents vascularized soft tissue ( skin , fat , muscle , or a composite thereof ) overlaying a bony prominence ., The central hypothesis of this model is that when this tissue is damaged ( which can happen in a variety of ways ) , further complications can evolve , including inciting further tissue damage ., See the Materials and Methods for a more detailed description of model rules ., Pseudocode is provided in the online supplement ( S1 ) ., Epithelial cell health in the PUABM depends on the continuous flow of oxygen to the field of simulated tissue ., Pressure—such as that created by shifting a person’s weight—is simulated as compressing the tissue and vasculature over a bony prominence , indirectly causing damage to the tissue by restricting the flow of oxygen to the tissue and secondarily by formation of oxygen species ( ROS ) upon reperfusion ( when pressure is relieved ) ., Injured tissue incites activation of inflammatory cells , which bring mediators to the field , some of which injure the tissue further ., As simulated cycles of pressure on/off are repeated , these mechanisms are sufficient to generate pressure ulcers in silico ., The molecular and cellular components of the model are enumerated in Table 1 ., In simulations of the PUABM , tissue health begins to decline within the first few hours of pressure cycling , first in the area where pressure intensity is greatest , directly over a simulated bony prominence ., This is manifested by a change in color from peach to red ( similar to the erythematous appearance of inflamed tissue; notice the area of redness developing over time in the simulations Fig 3B , top 2 rows ) ., While all of the patients in our cohort who had detectable ulcers at this early stage had light skin tones ( and therefore simulations did as well ) , model parameters could be adjusted to match darker skin tones ., The simulated tissue is able to recover some health in the very first rounds of pressure , during periods of reperfusion , but after a certain point , the PUABM tissue field remains red ( damaged ) despite reflow of oxygen and leukocytes to the region ., This is comparable to a region of un-blanchable skin erythema that is the diagnostic criteria for a Stage I PU 14 ., Tissue health declines further as the simulation progresses , increasing the intensity and radius of redness ., Eventually , tissue cells begin to die and exit the simulation , leaving behind a white patch to indicate lack of cellular activity at that position ( see 4th panel day 25 of the top 2 rows of Fig 3B ) ., The first instance of tissue cell death was considered to be the opening of a PU ., From this time onward in the simulation , the PU was observed to grow outward while cells near the edge continued to decline in health ., Simulations of the PUABM were initially compared to reference clinical images of PU severity ( see Fig 3A ) ., The National Pressure Ulcer Advisory Panel ( NPUAP ) has issued guidelines classifying ulcer severity by depth 41 , i . e . layer of tissue affected: skin , fat , muscle , or bone ., The prevailing notion is that some ulcers begin as deep tissue injury before opening to the epidermal surface 42 ., While the PUABM is two-dimensional , simulations were nonetheless able to capture the appearance of a variety of PU of various degrees of severity ., As shown in Fig 3A , simulations achieved appearances similar to all four stages of ulcer progression ., Specifically , the pattern of damaged tissue surrounding the ulcers was similar in simulations and clinical images ., In both simulations and clinical ulcers , tissue immediately surrounding the ulcer showed the most damage , while tissue farther away from the pressure point generally appeared healthier ( though not without some smaller areas of damage ) ., The simulations also recapitulated evolution of PU development observed in a prospective patient cohort of 49 patients ( see Materials and Methods ) ., The average time to ulceration post-injury was 20 ± 12 days among the RERC subjects ., With default parameters in the PUABM , the time to ulceration was 405 ± 6 hours ( 17 ± 1 days ) ; there was no statistically significant difference between the simulation predictions and the actual RERC ulceration times ( p = 0 . 606 by Mann-Whitney comparison on ranks ) ., We take this agreement to be one measure of validation of the mechanisms encoded in the PUABM and their associated parameter values ., The timing of ulceration in the PUABM was not encoded explicitly; in fact , ulceration itself was not encoded in the model ., Rather , the progression to ulcer occurred as a result of the accumulation of many smaller tissue-damaging events in a larger context ( field of tissue ) ., The parameter values that allowed this progression to agree with clinical data do not govern ulceration itself , but only control these smaller events ( e . g . macrophage lifespan , cytokine secretion rates , etc . ) ., When taken together across the whole simulation field , we conclude that PUABM-generated patterns are comparable to images of tissue in human patients ., The reference clinical images are static snapshots of a dynamic process; moreover , the images are from different individuals ., Accordingly , individual simulations of the PUABM were compared to the dynamics of PU formation in a single individual ., Note that clinical days are approximate , and that we sought to determine if the PUABM would result in qualitative concordance with clinical images ., Therefore , we allowed for alignment of simulation images to clinical snapshots within a several-day window , especially early in the process of ulcer formation ( when it is quite difficult for caregivers to define nascent ulcers ) ., Fig 3B shows snapshots from a single simulation in which PU reached visual appearances with striking similarities to time courses of individual patients ., Key characteristics of the PU arise in both simulations and patients ., Though the simulated bony prominence is circular while in reality the sacral pressure area is approximately a triangle , both simulated and actual patients develop ulcers with irregular shapes ., This is especially noticeable from day 33–34 onward in Fig 3B , for both simulations and clinical images ., Furthermore , in both simulations and clinical subjects , once an ulcer formed , a second nearby ulcer was more likely to develop ., Secondary ulcers are marked with open arrows; they are apparent in RERC subject showing a stage IV ulcer and in the last simulation panel of Fig 3A , though the ulcer also appears in the third simulation panel in a less severe state ., In our simulations , decreased oxygen and increased tissue damage surrounding the primary ulcer contribute to weakened surrounding tissue that is more vulnerable to ulceration than tissue that is farther away ., The model also recapitulated irregularly shaped ulcer boundaries: compare the last panel of Fig 3A and both rows of Fig 3B to images from stage IV NPUAP and the clinical time course in Fig 3B ., Jagged edges were also noted in both simulations and clinical images , which are indicated with solid arrows and appear in the rightmost panels of Fig 3A and across time courses in Fig 3B ., In order to gain insights that may at some point be of diagnostic utility , we sought to understand the range of behaviors that the PUABM can achieve , as well as the conditions that accompany these behaviors ., To inspire new treatment strategies , we investigated the likely mechanism by which those outcomes are reached ., While the structure of a given computational model can , of course , dictate its ultimate behaviors , much of the behavior of a model can be determined by the values of its parameters 43–49 ., Accordingly , model parameters ( including initial conditions , rates , thresholds , etc . ) were varied alone and in combination with each other in order to gain an understanding of how variations in parameters relate to variations in output ( time-evolving health of tissue ) ., Tissue health was quantified by assessing overall damage as indicated by the life score attributed to each tissue cell , as detailed in the Materials and Methods ., A second metric for tissue health was presence or absence and size of an ulcer , which was marked by the death of the first tissue cell ., Taken together , total tissue damage and presence/ size of an ulcer allowed us to compare one simulation to another and qualify whether the outcome of one was better or worse than the other ., We have previously put forward an approach involving extensive simulations validated against highly focused clinical data in order to streamline subsequent in vitro , in vivo , and clinical studies 62 ., Herein , we demonstrate how a tissue-realistic ABM of ischemia/reperfusion injury and inflammation in epithelial tissue could recapitulate morphological features of pressure ulcers in individuals with spinal cord injury , in a manner that is highly consistent with clinical images ., At the molecular level , the model was calibrated to in vivo studies of ischemia/ reperfusion injury in rat epidermis 38 as well as inflammatory dynamics reported in the literature ., The model output matched clinical ulcers qualitatively , generating irregular shapes and jagged edges , as well as distinct , secondary foci of inflammation that could progress to ulcers , despite initial conditions simulating a circular area of bony protrusion ., In further general agreement with clinical data , model simulations spontaneously reached endpoints of either pathogenic or resolving disease ., Interrogation of this stochastic phenomenon in silico revealed that ulceration outcomes were determined before the appearance of an open ulcer—suggesting that early diagnosis and intervention are critical ., From a diagnostic standpoint , our simulations suggest that the most important predictors of ulcer formation are tissue oxygen levels and the levels of pro-inflammatory mediators ., Predictions of the PUABM could be compared visually with easily-obtained images of patient skin; thus , we suggest that with further calibration and validation , this model could eventually be used as a diagnostic aid to determine which patients are at higher risk for ulcer formation before ulcers progress beyond stage I . From the standpoint of therapeutic interventions , in silico clinical trials using the PUABM suggested that after a relatively early point in time , inflammation-targeting intervention are unlikely to prevent ulceration or reduce tissue damage ., In our simulations , corticosteroids were incapable of preventing ulcer formation , though when applied early enough could reduce the amount of tissue damage surrounding the wound ., Sensitivity analysis revealed that ulceration in the model was correlated to tissue sensitivity to oxygen ., Thus , wound oxygenation may be a potential therapeutic avenue for post-SCI PU ., To date , studies of hyperbaric oxygen treatment have not pr

Introduction, Materials and Methods, Results, Discussion

People with spinal cord injury ( SCI ) are predisposed to pressure ulcers ( PU ) ., PU remain a significant burden in cost of care and quality of life despite improved mechanistic understanding and advanced interventions ., An agent-based model ( ABM ) of ischemia/reperfusion-induced inflammation and PU ( the PUABM ) was created , calibrated to serial images of post-SCI PU , and used to investigate potential treatments in silico ., Tissue-level features of the PUABM recapitulated visual patterns of ulcer formation in individuals with SCI ., These morphological features , along with simulated cell counts and mediator concentrations , suggested that the influence of inflammatory dynamics caused simulations to be committed to “better” vs . “worse” outcomes by 4 days of simulated time and prior to ulcer formation ., Sensitivity analysis of model parameters suggested that increasing oxygen availability would reduce PU incidence ., Using the PUABM , in silico trials of anti-inflammatory treatments such as corticosteroids and a neutralizing antibody targeted at Damage-Associated Molecular Pattern molecules ( DAMPs ) suggested that , at best , early application at a sufficiently high dose could attenuate local inflammation and reduce pressure-associated tissue damage , but could not reduce PU incidence ., The PUABM thus shows promise as an adjunct for mechanistic understanding , diagnosis , and design of therapies in the setting of PU .

A virtual pressure ulcer was created as a platform to test therapies and determine the mechanisms most correlated with unfavorable outcomes ., A layer of tissue fed with oxygen and diffusible molecules via blood vessels could develop an ulcer if pressure was applied , by simulating constriction of blood vessels in a circular region ., Simulated ulcers were visually similar to digital photographs of ulcers in individuals with spinal cord injury in their irregular shapes , jagged edges , and overall progression in time ., Statistical analyses of simulation outputs revealed that inflammation was an important determinant of ulcer severity and overall tissue damage ., However , simulated clinical trials revealed that blocking the negative effects of inflammation could not prevent ulceration , and in order to be beneficial at all for this specific type of ulcer , anti-inflammatory treatments must be applied during the earliest stages of ulcer formation—before many clinical signs of ulceration appear .

journal.pntd.0001811

Whole Genome Sequencing of Field Isolates Provides Robust Characterization of Genetic Diversity in Plasmodium vivax

Plasmodium vivax is the most widely distributed human malaria species and causes more illness than P . falciparum in many regions 1 ., Its global public health burden is estimated to be US$1 . 4 to 4 . 0 billion 2 ., Even in areas of low transmission , up to 20% of the population can have a symptomatic infection each year , with a cumulative experience of 10–30 episodes of malaria during a lifetime 3 ., Research on P . vivax is complicated by our inability to propagate the parasite in continuous in vitro cell cultures 4 ., This limits our ability to perform genetic crosses , to conduct in vitro functional assays on anti-malarial drug susceptibility or invasion mechanisms , and RNA-based investigations ., One alternative to understand phenotypic variations in P . vivax is to rely on purely genetic approaches and to statistically link genetic markers to traits of interests using linkage disequilibrium mapping ., A first step for developing genetic studies in P . vivax was achieved in 2008 with the completion of the reference genome sequence 5 generated from the Sal I strain ., This strain originated from a patient infected in El Salvador in 1972 and was propagated through infections of Aotus monkeys 5 , 6 ., A second milestone was cleared in 2010 with the first P . vivax genome sequenced directly from an infected patient 7 , demonstrating that it was possible to sequence P . vivax field isolates ., Currently , both P . vivax genome sequences have been generated from Central/South American parasites 5 , 7 ., While this is an important region of endemicity , where P . vivax consistently predominates in prevalence over P . falciparum 8 , these genomes only captured genetic diversity in a subset of the geographical range of P . vivax , and only its most recent expansion 9 ., To expand our understanding of genetic diversity in P . vivax , we have sequenced the genome of three field isolates from Cambodia where the parasite diversity is significantly different than in Central/South America and in closer proximity to its geographic origin 9 , 10 ., We have also sequenced two field isolates from Madagascar where we have recently identified P . vivax strains capable of infecting Duffy-negative erythrocytes 11 ., In addition , we have included another South American parasite , the monkey-adapted Belem strain 12 , and re-sequenced the Sal I strain 5 to rigorously assess the reliability of next generation whole genome sequencing for characterizing DNA polymorphisms ., Continuing advances in high-throughput sequencing technologies allowed us to generate high sequence coverage of these genomes , which circumvents most of the problems raised earlier 7 and provides reliable identification of single nucleotide polymorphisms ( SNPs ) ., This study was conducted according to the principles expressed in the Declaration of Helsinki ., Patient samples were obtained as part of on-going studies in accordance with human studies protocols IRB N°035-CE/MINSAN ( Comité dEthique du Ministère de la Santé de Madagascar , June 30th 2010 ) and IRB N°160 NECHR ( National Ethics Committee for Health Research – Cambodia , October 28th 2010 ) ., All patients provided written informed consent for the collection of samples and subsequent analysis ., We collected blood samples from two Malagasy ( M08 and M19 ) and three Cambodian patients ( C08 , C15 and C127 ) ., For each patient , we processed 5 ml of fresh blood collected in EDTA vacutainers through two consecutive CF11-packed columns to remove leukocytes and platelets ., We extracted parasite DNA directly from 200 µl of the remaining red blood cell fraction using DNeasy purification kit ( Qiagen ) ., For all samples , we confirmed P . vivax mono-species infection by Plasmodium species PCR-based diagnosis 13 ., We also analyzed DNA extracted from the monkey-adapted Belem and Sal I strains of P . vivax ( Text S1 ) ., For each sample , we sheared 144–518 ng of DNA into 250–300 bp fragments using a Covaris S2 instrument and used the fragmented DNA molecules to prepare sequencing libraries according to the Illumina protocol for genomic DNA ., Briefly , after end repair and A-tailing we ligated Illumina paired-end adapters to the ends of the fragmented DNA molecules and selected fragments of 300 bp ( i . e . P . vivax fragment size of ∼250 bp ) using an E-gel ( Invitrogen ) ., We then amplified the final products using 12 cycles of PCR and verified the quality and quantity of the libraries by Agilent BioAnalyzer and qPCR using the Illumina primers ., We sequenced each library on one lane of an Illumina HiSeq 2000 and generated between 79 and 230 million paired-end reads of 100 bp ., We mapped all reads to the human ( UCSC build hg18 , 14 ) and the P . vivax Sal I strain 5 reference genome sequences ., We used the program bwa 15 to independently map each end of all read pairs ., We considered as correctly mapped only reads mapped to a unique genomic location with, i ) less than 3 mismatches in the first 28 bases ,, ii ) 5 or less mismatches in the 100 bp sequence and, iii ) at most one insertion or deletion ., Only read pairs for which both ends fulfilled these criteria were included for further analyses ( Table 1 ) ., We identified read pairs that mapped to the exact same positions ( and could represent molecules amplified during the library preparation ) and randomly discarded all but one pair ., To identify SNPs , we focused on nucleotide positions covered by at least 20 reads with a quality score greater than 30 ., Since SNP identification is complicated in regions of high homology , we excluded from our analysis possible paralogous sequences ( see Text S1 for details ) ., In addition , we considered only read pairs that mapped in head-to-head configuration and within 1 , 000 bp of each other ., We identified consistent mismatches between reads generated from a given sample and the reference genome sequence using the samtool mpileup 16 with the extended base alignment quality computation ., Finally , we only considered a position variable in a given sample if at least 10% of the reads differed at this position from the reference nucleotide ( i . e . Reference Allele Frequency RAF <90% ) ., We characterized the function of each DNA polymorphism identified using the Sal I gene annotation downloaded from Ensembl ., We used perl scripts to annotate whether each polymorphism occurred in an intergenic region or a protein-coding gene and for the latter , whether it resulted in an amino acid change or a premature termination ., We reconstructed individual haplotypes ( i . e . the haploid DNA sequence of each P . vivax strain present in a sample ) from the short read sequences mapped at the Duffy Binding Protein ( DBP ) locus ., We retrieved , for each sample , all reads mapping to 1 kb upstream and downstream of the DBP gene ( PVX_110810 , chr6:384 , 498–390 , 259 ) and recorded all co-occurrences of consecutive alleles on read pairs: since read pairs are generated from the sequencing of the ends of individual DNA molecules , alleles observed on the same read pair are carried by the same haplotype ., For this analysis we focused on common haplotypes and only analyzed polymorphisms with a minor allele frequency of at least 5% in the sample studied ( i . e . only sites variable within a given sample are considered ) ., We inferred the haplotype using direct information when available , or allele frequency ., The final haplotype sequences were generated by substituting , in the Sal I reference sequence , the variable positions ( i . e . the haplotype polymorphisms inferred as well as the alleles at positions where all strains of a sample differed from the reference sequence , i . e . RAF <5% ) ., For the Belem strain none of the mismatches reached an allele frequency of 5% , consistent with a single strain being present in the infected monkey ., For five of the samples sequenced in this study , we amplified the Duffy Binding Protein region II from genomic DNA and , after cloning the PCR products , sequenced 12–91 clones per sample by traditional Sanger technology ., We reconstructed the haploid DNA sequence of the major strain across the entire genome using , at each nucleotide position , the most frequently observed allele ., Analytical and re-sampling approaches ( see Text S1 for details ) showed that this method performs well when one strain represents more than 80% of the parasite DNA ( using a minimum coverage of 20 X ) ., We analyzed allele sharing across samples by comparing the haploid DNA sequences of C08 , C127 , M15 , Sal I and Belem ., For each annotated protein coding DNA sequence , we calculated the number of nucleotide differences between each pair of samples and determined which haplotypes were closest ( i . e . lowest number of differences ) ., We looked for signals of local selection across the entire genome by searching for nucleotide positions where one allele was fixed in one population and the other allele fixed in the other populations ., To deal with multiple infections , we considered that all strains in one sample had the same allele if >90% of the reads carried this allele ( we used 90% instead of 100% to account for possible sequencing errors ) ., Studies of malaria parasites obtained from blood samples of infected patients are complicated by the presence of human genomic DNA 7: due to the difference in genome size , if only one leukocyte is present per 10 parasite cells , more than 95% of the extracted DNA will be of human origin ., This is particularly problematic in studying P . vivax as its parasitemia is typically less than 10 , 000 parasites per µl of blood ., Here , we analyzed blood samples from five malaria patients ( Table S1 ) , two from Madagascar ( M08 and M19 ) and three from Cambodia ( C08 , C15 and C127 ) ., We processed blood samples through cellulose columns to remove leukocytes and platelets 17 , 18 and extracted DNA from the red blood cell fraction ., In addition , we analyzed P . vivax DNA from the monkey-adapted Belem strain ( Text S1 ) as well as from the Sal I strain used for generating the P . vivax reference genome sequence 5 ., After library preparation , we sequenced each sample on one lane of an Illumina HiSeq 2000 to generate between 79 and 231 million paired-end reads of 100 bp ( Table 1 ) ., We mapped all reads to the P . vivax ( Sal I , 5 ) and human 14 reference genome sequences ., The Belem strain showed 71 . 09% of the reads mapping to the P . vivax genome ( Table 1 ) , resulting from the extensive effort to remove leukocytes from the blood sample ( Text S1 ) ., By contrast , only 1 . 32% of the reads generated from Sal I DNA mapped to the P . vivax genome ., This figure can be explained by the absence of leukocyte depletion prior to DNA extraction of the Sal I sample and illustrates the benefits of processing fresh blood samples on cellulose columns to remove host DNA ., Despite its relatively low coverage ( 20 X ) , we included Sal I in our analyses since comparison with the reference genome sequence generated from the same strain provided an opportunity to estimate the false positive rate of our SNP calling approach ., For the field isolates , a variable proportion of reads ( 4 . 91%–58 . 06% ) could be mapped to the human genome ( Table 1 ) , consistent with incomplete leukocyte depletion from the blood samples ., Despite this residual human DNA contamination and stringent quality controls , which eliminated between 27 and 62% of the reads mapped to P . vivax ( Table 1 ) , the amount of DNA sequence generated provided high coverage of the P . vivax genomes ( between 70 X and 407 X , Table 1 and Figure 1 ) ., However , the average genome coverage does not accurately represent the quality of the sequencing data ., Dharia et al . 7 sequenced the first P . vivax field isolate at an average genome coverage of 30 X . This 30 X coverage translated into 24 . 89% of the Sal I nucleotides being covered by more than 20 high-quality reads , and only 3% of the genes ( 158 out of the 5 , 050 annotated genes in the Sal I genome ) having more than 90% of their coding region sequenced at this coverage ., By contrast , owing to continuing advances in massively parallel sequencing , the Malagasy and Cambodian samples analyzed here had at least 93% of their genome sequenced by 20 reads or more , and between 84 and 97% of the genes covered ( Table 1 ) ., To determine whether we could use whole genome sequence data to identify DNA polymorphisms , we first compared the reads generated from Sal I DNA to the reference genome previously sequenced from the same strain 5 ., Out of ∼11 . 6 million nucleotide positions covered by 20 reads or more , only 121 nucleotides differed from the reference nucleotides in more than 10% of the reads covering those positions , and none were supported by more than 90% of the reads ( see Figure 2A and Text S1 ) ., These results highlighted both the high quality of the assembled P . vivax reference genome sequence and the suitability of high coverage genome sequence data for identifying SNPs with low false positive rates ., For all further analyses , we focused on positions of the Sal I reference genome that were covered by at least 20 reads in the Belem strain and each field isolate ( i . e . , all samples we sequenced excluding the low coverage Sal I ) ., We also excluded from our analyses potentially paralogous sequences that could generate spurious SNP calls ( Text S1 ) ., Overall , 19 , 533 , 315 nucleotides ( 86 . 35% of the Sal I reference genome ) were included in the SNP analysis ., For each sample , we recorded the percentage of reads differing from the reference nucleotide at each position ., For the monkey-adapted P . vivax Belem strain , at any given nucleotide position , all reads carried the reference allele ( i . e . 100% Reference Allele Frequency RAF ) or all reads differed from the Sal I reference ( 0% RAF ) ( Figure 2A ) ., This is consistent with P . vivax being haploid in the human/monkey host and with the presence of a single strain in the Saimiri monkey ., Note that the RAF at some positions differed slightly from 0 or 100% ( typically by less than 5% ) due to sequencing errors ., The distribution of RAF was strikingly different for the P . vivax field isolates: in these samples , we consistently observed two alleles at many positions ( Figure 2B ) ., This pattern suggested that multiple strains of P . vivax were present in each patient blood sample ., For example , a SNP with an RAF of 20% ( as it was frequently observed in C08 ) could occur if two strains of P . vivax were present in the patient blood and the major strain ( accounting for 80% of the parasites ) differed from the reference allele at this position while the minor strain ( making up the remaining 20% of the parasites ) carried the Sal I reference nucleotide ., The peaks at 0% RAF in Figure 2B represent positions where all strains present in a sample differed from the Sal I reference allele ., We independently validated a subset of the SNPs by cloning and Sanger sequencing the Duffy binding protein region II ( DBPII ) for five of our samples ., The Sanger sequencing results were consistent with whole genome sequence findings and validated 17 out of the 17 SNPs identified by Illumina sequencing in this region ( Table S2 ) ., In addition , analysis of the cloned sequences revealed distinct haplotypes amplified from a single blood sample , confirming the presence of multiple strains in two of the samples ( Figure S1 ) ., Overall , we identified 80 , 657 nucleotide positions where at least 10% of the reads differed from the reference sequence in one or more of the samples ., The 80 , 657 SNPs were distributed throughout the genome with an average of 4 . 13 SNPs per kb ., However , there was extensive variation in SNP density among genomic regions ( Figure 1 ) ., Most notably , the extent of genetic diversity was highly dependent on the gene context: intergenic regions showed a much higher diversity than coding regions ( 6 . 98 vs . 2 . 96 SNPs per kb , respectively ) with intronic sequences harboring an intermediate level of diversity ( 4 . 29 SNPs per kb ) ., This observation was similar to the results of a study of 100 kb of contiguous DNA sequence 19 and consistent with purifying selection maintaining the DNA sequence at most genes in the P . vivax genome by removing deleterious mutations ., 48 , 224 SNPs occurred in intergenic regions ., SNPs in annotated protein coding regions included 13 , 203 synonymous polymorphisms ( sSNPs , 16 . 37% of all SNPs ) , 19 , 191 non-synonymous polymorphisms ( nsSNPs , 23 . 79% ) and 39 substitutions ( 0 . 05% ) introducing an early stop codon ., We only observed 1 . 5-fold more nsSNPs than sSNPS , while based on the composition of the P . vivax genome we would expect by chance ∼4-fold more nsSNPs ., This observation also suggested that the evolution of most protein-coding sequences in P . vivax genome is driven by purifying selection ., We attempted to assign allelic variants observed within a sample to individual P . vivax parasites for the Duffy binding protein locus ., For each sample , we recorded the co-occurrence of consecutive alleles on individual read pairs: since read pairs were generated by sequencing the ends of single DNA molecules , alleles observed on the same read pair were carried by the same parasite ., Using this procedure , we were able to reconstruct haplotypes for the most prevalent strain for all samples as well as for a second strain for the C15 , C127 and M08 field isolates ( Figure 3 and Figure S2 ) ., The inferred haplotypes were identical to the consensus DBPII sequences generated by cloning and Sanger sequencing from the same samples ( Figure S1 ) , validating our haplotype reconstruction approach ., The haplotypes inferred from the same patient blood sample were not closely related to each other and represented unrelated P . vivax parasites ., It is important to note here that while sequencing data allows identifying genetically distinct parasites , it does not differentiate related parasites derived from parental strains by mutations or recombination ( see e . g . 20 ) ., Haplotype reconstruction , when several strains are present in a single sample , depends on several factors , including the SNP density and the extent of linkage disequilibrium ., This currently hampers extending our approach to the entire genome ., However , the analysis of Duffy binding protein haplotypes was consistent with the distribution of allele frequencies shown on Figure 2 and indicated that , in all field samples , 2–4 strains contributed to more than 95% of the P . vivax DNA ( as opposed to a scenario where dozen of strains would be equally abundant in a patient ) ., Two strains were equally abundant in C15 , while for M19 three strains dominated ( with roughly 50% , 25% and 25% frequency ) ., In three samples , M15 , C08 and C127 , one single strain largely dominated all others and represented , respectively , 80% , 80% and >90% of the parasites present ., Given the high sequence coverage generated here , for these samples , the most frequently observed allele at each nucleotide position was very likely carried by the dominant strain ( see Text S1 for details ) ., We therefore inferred for these three samples the haploid sequence of the dominant strain for the entire genome by considering the major allele at each variable position ., Combined with the single strain sequences of Belem and Sal I , these sequences provided five haploid genome sequences for P . vivax from three continents ., Studies of the global P . vivax genetic diversity have been limited by the lack of informative markers and essentially based on a few loci ( e . g . the mitochondrial DNA 9 , 21 and the Duffy binding protein 22 ) ., This has greatly limited our understanding of the P . vivax population structure and history since diversity at these loci may be influenced by natural selection ., The five haploid genome sequences generated here provided an opportunity to preliminarily assess the global genetic diversity of P . vivax ., Identification of likely neutral sequences in the P . vivax genome is complicated by its gene density: less than half of the genome sequence is intergenic and there are few long stretches ( e . g . ≥10 kb ) of DNA sequences without annotated genes ., We therefore focused on the analysis of four-fold degenerate sites ( i . e . , nucleotide positions where substitutions do not change the amino acid sequence ) that are less likely to be directly affected by natural selection ., Among 98 , 393 four-fold degenerate sites sequenced , we observed 2 , 193 variable sites , including 1 , 769 sites that differed in only one sample ., This represented a significant excess of singletons compared to the number expected under a neutral model of a random mating population of constant size and could indicate that P . vivax population has recently expanded in size , or alternatively , that the parasite population is heterogeneous and composed of many sub-populations ., Consistent with previous reports 22 , our analysis of Duffy binding protein sequences showed a star-like phylogeny with no apparent geographic stratification ( Figure S3 ) ., This pattern could reflect the actual structure of the P . vivax population or simply indicate the action of natural selection on the DBP gene ., To further investigate population structure in P . vivax , we compared the five haploid sequences and determined , for each annotated gene , the geographical origin of the closest haplotype to a given sample ( similar to the nearest neighbor approach described in 23 ) ., Our results showed that , while strains from the same location tended to be more similar to each other than to a strain from a different continent , there was considerable allele sharing across continents ( Figure 4 ) ., For example , the haplotype sequence for the Cambodian sample C08 was most similar to the Cambodian C127 haplotype for 586 genes but most similar to the Malagasy M15 or South American Belem haplotypes at 463 genes ., Consistent with this observation , a tree reconstructed using the total number of nucleotide differences among whole genome haploid sequences showed that strains from a same continent clustered together but with very long external branches ( Figure S4 ) , indicating that most diversity is observed among samples rather than between geographical locations ., Previous studies based on mitochondrial DNA 24 and microsatellites 25 have also highlighted that similar haplotypes are often shared across continents ( but see also 26 ) ., This observation of extensive allele sharing across continents is unexpected as we may have expected to observe consequences of local adaptation , and therefore greater population differentiation , as P . vivax spread across the world and encountered new environments ( e . g . , different mosquito species , different hosts immune response ) ., We screened the P . vivax genome for evidence of adaptive selection by looking for SNPs for which one allele was fixed in all parasites from one geographical area while the other allele was fixed in all other parasites ., For comparison , this is the situation at the Duffy locus in humans where Duffy-negativity is fixed in Sub-Saharan Africans and absent in non-African populations ., Among the 80 , 657 SNPs identified , we only observed 96 SNPs with such dramatic allele frequency differences ( not statistically different from the number expected by chance due to the small number of samples analyzed ) ., In addition , while all three Cambodian-specific alleles occurred in close proximity ( within 20 bp from each other ) , the 92 Malagasy-specific alleles were distributed across the 14 chromosomes suggesting that chance rather than natural selection was responsible for these results ., This analysis was consistent with our observation of allele sharing across continents and suggested that P . vivax population is not highly differentiated ., In conclusion , we showed that continuing advances in sequencing technology allow the robust characterization of genetic diversity in P . vivax genomes ., The SNPs identified here will be valuable for vivax malaria research to design population studies ( e . g . studying the diversity of P . vivax in one region ) and to identify the genetic basis of disease-related traits by association studies ., In this regard , it is important to note that we identified multiple parasites in each patient blood sample analyzed , which will complicate these studies and will need to be rigorously accounted for ., Finally , our analysis of P . vivax genomes from three continents revealed allele sharing across continents and little evidence of local adaptations ., While our analysis includes , for the first time , genetic diversity estimates across the entire genome , the number of samples analyzed here is limited ., We conducted population genetic analyses using approaches robust to small sample sizes but our results will need to be confirmed as more genome sequences become available for this parasite ., One possible explanation for our observations is that the P . vivax population originated recently and dispersed rapidly across the world without major loss of diversity or much influence of natural selection ., Alternatively , allele sharing could be due to continuous gene flow in the present P . vivax population: P . vivax is now a cosmopolitan parasite that can be easily spread throughout the world by way of dormant hypnozoites ., If this second hypothesis is true , it holds bleak prospects for vivax malaria elimination: with high level of gene flow , genetic polymorphisms conferring drug resistance 27 , 28 or novel invasion mechanisms 11 could spread across the world and further complicate control strategies .

Introduction, Materials and Methods, Results and Discussion

An estimated 2 . 85 billion people live at risk of Plasmodium vivax transmission ., In endemic countries vivax malaria causes significant morbidity and its mortality is becoming more widely appreciated , drug-resistant strains are increasing in prevalence , and an increasing number of reports indicate that P . vivax is capable of breaking through the Duffy-negative barrier long considered to confer resistance to blood stage infection ., Absence of robust in vitro propagation limits our understanding of fundamental aspects of the parasites biology , including the determinants of its dormant hypnozoite phase , its virulence and drug susceptibility , and the molecular mechanisms underlying red blood cell invasion ., Here , we report results from whole genome sequencing of five P . vivax isolates obtained from Malagasy and Cambodian patients , and of the monkey-adapted Belem strain ., We obtained an average 70–400 X coverage of each genome , resulting in more than 93% of the Sal I reference sequence covered by 20 reads or more ., Our study identifies more than 80 , 000 SNPs distributed throughout the genome which will allow designing association studies and population surveys ., Analysis of the genome-wide genetic diversity in P . vivax also reveals considerable allele sharing among isolates from different continents ., This observation could be consistent with a high level of gene flow among parasite strains distributed throughout the world ., Our study shows that it is feasible to perform whole genome sequencing of P . vivax field isolates and rigorously characterize the genetic diversity of this parasite ., The catalogue of polymorphisms generated here will enable large-scale genotyping studies and contribute to a better understanding of P . vivax traits such as drug resistance or erythrocyte invasion , partially circumventing the lack of laboratory culture that has hampered vivax research for years .

Plasmodium vivax is the most frequently transmitted and widely distributed cause of malaria in the world ., Each year P . vivax is responsible for approximately 250 million clinical cases of malaria and its global economic burden , placed largely on the poor , has been estimated to exceed US$1 . 4 billion ., In contrast to P . falciparum , P . vivax cannot be propagated in continuous in vitro culture and this limits our understanding of the parasite’s biology ., In this study , we sequenced the entire genome of five P . vivax isolates directly from blood samples of infected patients ., Our data indicated that each patient was infected with multiple P . vivax strains ., We also identified more than 80 , 000 DNA polymorphisms distributed throughout the genome that will enable future studies of the P . vivax population and association mapping studies ., Our study illustrates the potential of genomic studies for better understanding P . vivax biology and how the parasite successfully evades malaria elimination efforts worldwide .

genome sequencing, medicine, infectious diseases, plasmodium malariae, neglected tropical diseases, biology, genomics, malaria, parasitic diseases, genetics and genomics

journal.pcbi.1002907

Structure-based Molecular Simulations Reveal the Enhancement of Biased Brownian Motions in Single-headed Kinesin

Time dependent structural information is of central importance to understand detailed mechanisms of biomolecular systems ., In particular , biomolecular machines dynamically transit many structurally and chemically distinct states making cycles in state space , by which they fulfill their functions ., Unfortunately , no single experimental technique provides sufficient spatio-temporal resolution for them ., X-ray crystallography and others provide structural information at high resolution , but this is primarily static ., Biochemical and single molecular experiments tell us kinetic and dynamic behaviors , but their spatial resolution is limited ., To fill the gap among them , molecular dynamics ( MD ) simulations have been playing important roles ., Yet , due to their size and long time scale involved , atomistic MD cannot cover an entire cycle of molecular machines at the moment 1 ., To overcome this limitation , recently , we initiated to use structure-based coarse grained MD ( CGMD ) methods 2 , 3 to mimic the cycle of machines for the case of F1-ATPase and others 4 , 5 ., Notably , most of these machines contain more than one ATPase domains and their coordinated dynamics are crucial to understand the mechanisms 6 , 7 , 8 , 9 ., This is an interesting issue , but at the same time , makes the cycle unavoidably complicated ., Thus , for the simplicity and clarity , it is good to study those that contain only one ATPase domain and that have much of crystallographic information ., In this sense , a single-headed kinesin , KIF1A , is an ideal target system , for which here we performed CGMD simulations mimicking an entire ATP hydrolysis cycle ., Kinesin is a family of molecular motors that move unidirectionally along microtubule ( MT ) using ATP hydrolysis free energy 10 ., In the family , the conventional kinesin , kinesin-1 , was experimentally characterized to move toward the plus ends of MT processively with discrete 8-nm steps per one ATP hydrolysis reaction , where the coupling between ATP hydrolysis reactions and 8-nm steps is rather tight 11 , ., The conventional kinesin is a two-headed motor and has been shown to “walk” in a hand-over-hand fashion by coordinated motions of the two heads 6 , 9 , 17 , 18 ., In this sense , it is a surprise that even though KIF1A , a member of kinesin family , is a single-head motor , it still can move processively and directionally along MT , as observed by single molecule experiments 19 , 20 , 21 ., In particular , the mechano-chemical coupling of KIF1A is loose: KIF1A can move back and forth stochastically with an average biased towards the forward direction , with step sizes in multiples of 8-nm ., This is in contrast to conventional kinesin that seldom shows backward steps without a large load and that shows a uniform step size of 8-nm per one ATP hydrolysis 19 , 20 , 21 ., Thus , KIF1A must use a different mechanism from the conventional kinesin to achieve the overall unidirectional motions ., How KIF1A , with only one head , can generate the unidirectional movements driven by ATP-hydrolysis reaction is unclear in terms of structural dynamics , which we address in this paper by structure-based CGMD ., Various molecular simulations have been applied to kinesin as well as other molecular motors 4 , 8 , 22 , 23 , 24 ., For the molecular simulations of KIF1A movements , structural information on nucleotide-dependent conformational change is indispensible ., X-ray crystallography provides KIF1A structures in two major conformations; ATP and ADP bound forms 25 , 26 , 27 ., The two forms share the overall fold of the head domain with some changes ., One crucial change is in the helix α4; its orientation relative to the rest of the head is rotated by about 20 degrees between the two forms ( Fig . 1A , blue for ATP-form and red for ADP-form ) ., Another major change is in the so-called neck-linker region , which is the C-terminus of the head domain: the neck-linker is ordered and tightly docked to the core of the head in the ATP form ( magenta in Fig . 1B ) , while it is disordered and thus invisible in the ADP form ., This neck-linker docking/undocking has been implicated as a source of the power-stroke in the kinesin family 28 , 29 ., The K-loop ( L12-loop ) and L11-loop , which flank the α4 helix , also show some changes between the two forms ( Fig . 1A ) ., Cryo-electron microscopy ( cryo-EM ) of the KIF1A-MT complex together with X-ray structures of building blocks led to structural models for the KIF1A-MT complex in the two major forms of KIF1A ( the ATP and ADP forms ) 30 , 31 ., The modeled complexes show that , in both of the forms , the key interaction sites of KIF1A with MT is the α4 helix , which fits to a groove located between α-tubulin and β-tubulin ., In the two forms of KIF1A , the orientation of the α4 helix relative to MT is mostly unchanged , which leads to the 20-degrees rotation of the core domain relative to the long axis ( z-direction in this article ) of MT depending on the bound nucleotide states ( Fig . 1C ) : In the ATP-form , the core adopts the “upright” docking ( blue in Fig . 1C left ) , while in the ADP form the core is rotated about 20 degrees and adopts the “tilted” docking ( red in Fig . 1C right ) 31 ., This core rotation has been suggested to be important for KIF1A movement 26 , 31 ., The KIF1A-MT complex models provide a clue for the processivity of KIF1A ., Regardless of the nucleotide states , the positively-charged K-loop of KIF1A is close to the negatively charged E-hooks , disordered C-terminus regions of α/β-tubluin 26 , 31 ., Thus , the long-ranged electrostatic attractions between K-loop and E-hooks are assumed to prohibit the KIF1A from completely leaving from MT . This idea is supported by a mutational experiment , in which charge reduction of K-loop decreased the processivity 19 ., The ATP hydrolysis cycle and its correlation with KIF1A head motion have been investigated previously 19 , 20 , 21 , 26 ( see Fig . 2A ) ., The ATP form of the KIF1A head binds strongly to MT ( T-phase in Fig . 2A ) , whereas the direct contact of the ADP-form of KIF1A to MT is weak ., Thus , after the ATP hydrolysis and Pi release , the KIF1A head can detach from MT ( still loosely bound to MT via electrostatic interactions of the K-loop and E-hooks ) ., The detached head starts diffusion along MT under the constraints generated by the interaction of the K-loop and E-hooks ., After the long one-dimensional diffusion along MT , KIF1A can finally find a binding site located at the groove between α- and β-tubulins ( D-phase in Fig . 2A ) ., The contact between tubulin and KIF1A induces ADP dissociation from KIF1A leading to the nucleotide free state ., In this state , KIF1A binds MT tightly ( Φ-phase in Fig . 2A ) ., At the final stage , ATP binding induces neck-linker docking and the rotation of the core ( T-phase in Fig . 2A ) ., The above knowledge , however , does not tell us the mechanism of how KIF1A can generate directional movement towards the plus end of MT . To address the mechanism of directional movement , we designed and conducted a series of molecular simulations employing structure-based CG protein models ., The structure-based CG protein models have proven to be useful to study mechanical aspects of kinesin 8 , and other molecular motors 4 , ., Based on the energy landscape view of proteins 32 , 33 and structural data for the two forms of KIF1A , we set up single and/or two-basin energy landscapes of KIF1A for every phase of an ATP cycle 34 , 35 ., Then , the ATP hydrolysis cycle was mimicked by dynamically switching the energy functions of KIF1A in different phases of the cycle ( Fig . 2B ) 4 , 36 ., While the full-length KIF1A has a rather long C-terminal tail , we here concentrate on a truncated KIF1A ( C351 ) that was used in in vitro single-molecule assays 19 , 20 , 21 ., We note that , by employing the structure-based CG simulations , our purpose here is not to conduct a single simulation that most accurately approximate the real molecular system , as some parameters in the CG simulations are not accurately derived from atomic interactions ., Instead , taking advantage of the speed of the structure-based CG simulations , we systematically conduct a series of simulations for a broad range of these parameter values ., These comparative computer experiments are useful for a mechanistic understanding ., We designed a simulation system for one ATP hydrolysis cycle of KIF1A that induces KIF1A motions along MT . The simulation system contains 7 protein subunits: a KIF1A molecule that moves dynamically and three copies of tubulin αβ dimers that were fixed in the form of a segment of single protofilament of MT ( Fig . 1D ) ., All the proteins were modeled at a one-bead-per-residue resolution ( each amino acid was represented by a bead located at the Cα position ) ., For KIF1A , we employed structure-based CG models that concisely represent the energy landscape , which is a globally funnel-like shape where the bottom of the funnel can have more than one basin 33 ., We focused on a truncated KIF1A C351 ( unless otherwise mentioned ) since the motility of this type KIF1A is intensively investigated in the single-molecular assay of Okada et al 19 , 20 , 21 ., Conformational changes of KIF1A upon chemical reactions were simulated by the multiple-basin model 35 , while the long-time dynamics that do not involve chemical reactions , such as diffusion process , were simulated by a single-basin perfect-funnel ( i . e . , Go ) model 34 , 37 ( see below and Materials and Methods for more details ) ., Protein dynamics was simulated by stochastic differential equation , i . e . , the Langevin equation ( see below and Materials and Methods for more details ) ., The crystal structures of ATP-bound KIF1A ( designated as KIF1A ( T ) hereafter ) and ADP-bound KIF1A ( KIF1A ( D ) ) are available from the Protein Data Bank and were used in the CG models as reference structures of the corresponding states ., For the KIF1A-MT complex structures , the cryo-EM-based models for the ATP- and ADP-bound KIF1A-MT complexes are also available and were used ( we designate XT and XD respectively ) ., These models explain the high and low affinities in ATP-bound and ADP-bound forms of KIF1A , respectively , by the number of direct contacts ., The structure for nucleotide-free KIF1A ( KIF1A ( Φ ) ) is currently unavailable; we assumed that the neck linker is disordered based on experiments , and that the KIF1A ( Φ ) -MT complex structure XΦ except the neck linker to be the same as that of XT because both states have a similarly high affinity to MT . Using these complexes , we modeled the interactions between KIF1A and MT as a Go-like pair potential ( unless otherwise mentioned ) ., In the current CG model , the interaction strength between KIF1A and MT is a key parameter ., First , the interaction strength parameter had to be tuned so that KIF1A ( T ) can stably bind to MT while KIF1A ( D ) can detach from MT during the affordable simulation time ., This tuning was easy because , as mentioned above , the modeled complex structures of KIF1A-MT have more residue-contacts in the ATP form than in the ADP-form ., A more delicate tuning was necessary for the affinity of KIF1A ( D ) to MT because KIF1A ( D ) is expected to detach from MT and later reattach ., Obviously , a too weak interaction does not lead to attachment of KIF1A to MT , whereas a too strong interaction does not allow the detachment from MT . Via many preliminary runs , we found a certain range of the interaction strength parameters that satisfy these conditions ( described in the next subsection ) ., Our simulation started from the XT ., KIF1A was bound to the central tubulin αβ dimer ( Fig . 1D ) ., We simulated the KIF1A ( T ) state for 5×105 τ , where τ is the unit of time in CG-simulation , using the multiple-basin potential with two basins: a stable basin at XT and a meta-stable basin at XD structures ( Fig . 2B top ) ., The unit of time τ can be mapped to ∼0 . 128 ps in real time scale based on the diffusion constant of the KIF1A head ( see Materials and Methods for the detail information ) ., Then , we induced the conformational change to the ADP-bound form by switching the potential so that the XD structure becomes more stable than XT ( see the second row and left cartoon of Fig . 2B ) ., With this setting , we simulated the system for 4×106 τ , which is long enough to complete the conformational change to ADP-form ., For many samples , KIF1A ( D ) detached from MT during this period ., We note that , throughout the simulations , a constraint potential was applied that represents long-range loose interactions between the K-loop and E-hooks , by which KIF1A cannot move far away from MT ( see Materials and Methods for details ) ., Then , we conducted a long simulation ( 2×108τ ) with the single-basin Go potential for the XD ( the second row and central cartoon in Fig . 2B ) ., The switch from the multiple-basin potential to the single-basin Go potential saves computer time and is done solely for technical reasons ., During this period , many trajectories showed KIF1A re-attachment to MT . Once KIF1A attached to MT , we continued the run for another ∼1×107τ and then moved to the next stage ., The next stage is a preparation to the subsequent conformational change to the nucleotide-free ( Φ ) state and uses the multiple-basin model with the stable basin at XD and the meta-stable basin at XΦ for 5×105τ ., After that , corresponding to the release of ADP , we induced the conformational change to the nucleotide free form by switching the potential so that the XΦ structure is more stable than XD ( the third row right in Fig . 2B ) ., Subsequently , for a long time dynamics , we used the single-basin potential for the Φ state for 1×107τ ., Finally , ATP-binding is mimicked by switching the potential to the single potential for XT ., We simulated the T state for ∼2×108τ , which completes the XTXDXΦXT cycle ., We now analyze KIF1A movement during one ATP cycle ., As in Fig . 2A , it is expected that KIF1A detaches from MT and attaches to MT both in the D-phase ., Thus , modeling of the interaction between KIF1A ( D ) and tubulin is very delicate ., Since the CG modeling is unavoidably less accurate , instead of deciding one “correct” interaction strength , we scanned the strength over a certain range ., In a strong interaction case ( designated as stand-alone/strong , εgoKIF1A-MT\u200a=\u200a0 . 225 ) ( Throughout the paper , the energy unit corresponds to kcal/mol ( ∼1 . 7 kBT\u200a=\u200a∼6 . 95 pN . nm ) although the mapping is rather approximate ) , we saw KIF1A cannot detach from MT for 99 of 100 trajectories ( Fig . 3A ) within the simulated time ., Whereas , with a weak interaction ( stand-alone/weak , εgoKIF1A-MT\u200a=\u200a0 . 153 ) that was carefully tuned after trial-and-errors , we found that KIF1A can detach from MT and attach to MT ( the first three cases in Fig . 3B ) for 186 of 235 samples ( ∼80% ) ., The rest 49 samples did not show detachment ( bottom in Fig . 3B ) ., The first , second , and third cases in Fig . 3B illustrate the one forward step ( +8 nm ) , the zero-step ( 0 nm ) , and the one backward step ( −8 nm ) within one ATP hydrolysis cycle , respectively ( For an example of stand-alone KIF1A movements for stand-alone/weak , see Supporting Information Video S1 ) ., Of the 186 cases that KIF1A detached from and attached to MT within one ATP chemical cycle ( TDΦT ) , the positions of KIF1A at the end of simulations were +8 nm ( the forward step ) for 44 cases , 0 nm ( zero-step ) for 92 cases , and −8 nm ( the backward step ) for 50 cases ( For statistics , Table 1 ) ., We note that the system contained only 3 pairs of tubulin αβs that correspond to kinesin binding sites of +8 nm , 0 nm , and −8 nm so that possibilities of two steps were out of the scope here ., On average , no significantly biased move was observed ., Apparently , this does not explain the in vitro single molecule experiments that found forward biased moves ., The simulations above did not consider electrostatic interactions at all , which may have affected the results ., Indeed , recent work by Grant et al reported forward bias of two-headed kinesin landing due to electrostatic interactions 22 ., We thus added the electrostatic interactions between KIF1A and MT by the Debye-Huckel formula and repeated the same set of simulations for 80 samples for the case of stand-alone/weak/DH ., We set the salt concentration of 50 mM , and put +1 charges to all the Lys , Arg , and His residues and −1 charges to all the Asp and Glu residues in the simulated system ., Of 80 , 6 samples did not show detachment , 12 samples showed one forward-step ( 8 nm ) , 16 samples showed one backward step , and 46 samples returned to the original site ( see Fig . S1 ) ., Thus , inclusion of the simple electrostatic interactions did not produce forward-biased movements although it changed the trajectories to some extents ( see Figs S2 , S3 , S4 , S5 , S6 ) ., We further tried simulations with many different sets of parameters never finding biased motions ., Our results is apparently inconsistent with the biased binding mechanism proposed in 21 ., There can be two possibilities ., 1 ) Some fine effect which is not included in our CG simulations , such as more accurate electrostatic treatment by Grant et al , is responsible for the forward biased binding ., 2 ) The forward-biased binding is not realized ., Further work is necessary to solve the issue ., In struggling for search of models/situations that exhibit the forward biased move of KIF1A , we came up with a situation that a large cargo-analog is attached to the C-terminus of the neck-linker of KIF1A ., The cargo-analog is modeled as a large sphere of ∼1 µm radius , and thus has very small diffusion constant ., There are some in vitro experiments for myosin , as well as another kinesin mutant , that suggest the importance of diffusion anchor linked at the end of motor proteins for processive and directional movements 38 , 39 ., Technically , we added a mass point with large friction coefficient to the C-terminus of the neck linker ., With a large cargo-analog , we first used a strong interaction between KIF1A and MT ( cargo/strong , εgoKIF1A-MT\u200a=\u200a0 . 225 , the same strength as the case of stand-alone/strong ) , and simulated one ATP cycle for 109 samples ., We modeled the cargo as a sphere of radius 3000 times as large as the radius of an amino acid , which is ∼1 µm ., Assuming the same density as amino acids , the mass of the cargo scales as 30003 times as large as that of an amino acid ., The Stokes-Einstein law D\u200a=\u200akBT/6πηr , where η is water viscosity: ∼0 . 8 m Pa s and r is the radius of the particle , gives that the diffusion constants Dcargo for the cargo is 3000 times smaller than the diffusion constant of an amino acid ., ( See Materials and Methods for the detailed information ) ., In the simulations , we found most samples either moved one-step forward ( 52 of 109 cases , an example in the upper panel of Fig . 4A top and Video S2 ) or re-bound to the original site ( 56 of 109 cases , the upper panel of Fig . 4A bottom ) , while almost no case of the backward step was found ( Table 1 ) ., In ATP-bound state ( t<5105τ ) , KIF1A head kept binding to MT firmly and the cargo-analog did not move significantly at 4 nm in front of the head corresponding to the length of the neck-linker ( a snapshot in Fig . 4B top , a histogram in Fig . 5 left ) ., Immediately after the ATP hydrolysis , KIF1A head detached from MT quickly ., After the detachment , KIF1A head exhibited quasi-one dimensional diffusion along MT , while , due to the large friction , the cargo-analog did not move significantly ., Thus , the fluctuation of the KIF1A head was restricted around the almost-fixed cargo located 4 nm in front ( Fig . 4B and Fig . 5 left ) ., The cargo-analog played a role of an anchor ( or a cane ) ., After some diffusion , the detached head finally re-bound to MT . Because of the limited range of diffusion , the re-binding site was either the forward site ( +8 nm ) or the original site ( 0 nm ) ., After the attachment on MT , we changed the state of the system from ADP-state to Φ-state , which did not lead to any marked difference in the movement of the cargo or the head ., After that , ATP binding to KIF1A induced the neck-linker docking that moved the position of the cargo-analog , which is about 8 nm in case of the forward step ( Fig . 5 left ) ., Thus , after one ATP cycle ( TDΦT ) , the 8-nm or 0-nm displacements of the cargo-analog as well as the head were realized stochastically ., With a weak interaction between KIF1A and MT ( cargo/weak , εgoKIF1A-MT\u200a=\u200a0 . 153 ) , we still found clear forward bias ( the bottom panel of Fig . 4A ) although the details were different ., In particular , due to a weaker interaction , the average time for the head diffusion increased , which resulted in larger exploration by one-dimensional diffusion and appearance of the one backward step ( 26 of 150 samples ) as well as the one forward ( 43 of 150 ) step , and the zero step ( 81 of 150 ) ( Fig . 5 middle ) ., As noted before , our simulation system included only the three binding sites and thus two forward or backward steps were not realized by design ., For comparison , Fig . 5 right shows the histogram of the move for the case of stand-alone/weak , confirming that no significant bias is observed ., We now focus on the detachment process of the KIF1A head from MT after the ATP hydrolysis and Pi release ., Upon the potential switch from ATP- to ADP-state at t\u200a=\u200a5105 τ ( Fig . 2B top to the second row left ) , the decrease in the number of residue-contacts between KIF1A head and MT led to the reduced binding energy , which could induce the detachment of KIF1A head ., Interestingly , with the strong interaction , the stand-alone KIF1A simulation showed the KIF1A head detachment with the probability 1% , whilst the simulation with the cargo-analog promptly induced the head detachment with the probability 100% ( Fig . 6A ) ., Thus , clearly , the cargo-analog enhanced the KIF1A head detachment from MT . Even with the weak interaction between KIF1A and MT , the detachment probability was 79% for the stand-alone KIF1A ( Fig . 6A ) ., With the strong interaction , we tested the detachment process with three cargo sizes ( and thus three frictions and masses ) ( the inset in Fig . 6A ) ; the small ( light-green ) , the middle-size ( red , the default one ) and the large ( purple ) cargoes correspond to the radii of 2000 , 3000 , and 4000 times of one amino acid , respectively ., Technically , for given radii , masses and frictions were scaled according to Stokes-Einstein law in the same way as described ., We see that KIF1A did not detach from MT with the probability 11% for the case of the small cargo , while the detachments probabilities were 100% for the system with the middle or the large cargo ., Thus , the relatively large friction/mass cargo promoted the detachment of the KIF1A head ., In the complex of KIF1A-MT , the α4 helix of KIF1A fits into a groove of MT . When the ATP hydrolysis occurs in the KIF1A head bound to MT , the head tends to make conformational change from ATP-form to ADP-form ., With the constraint on the α4 helix , the conformational change would induce about 20 degree clockwise rotation of the head relative to the microtubule ( viewed from the top as shown in Fig . 1C left to right ) , which increases the distance between C-terminal of the head and the cargo rapidly ., Then , a tag-of-war between the head and the cargo takes place ., When the cargo is sufficiently large , the cargo is less mobile and wins the tag-of-war , thus finally pulling the KIF1A head out of MT . Fig . 6B illustrates time series of the binding energy for the strong interaction case ., For the case of stand-alone/strong ( orange ) , the binding energy was weakened from ∼−30 kcal/mol in T-state to ∼-20 kcal/mol in D-state , but the latter was strong enough to hold the KIF1A head stably ., For the large cargo case ( purple ) , upon ATP hydrolysis , KIF1A promptly detached from MT . For the cases of small ( light-green ) and the middle-size ( red ) cargoes , TD switch immediately weakened the binding energy to ∼−12 . 5 kcal/mol , which were followed either by detachment or by the relaxing to the binding energy ∼−20 kcal/mol in D-state ( light-green ) ., This transient intermediate state with the binding energy ∼−12 . 5 kcal/mol corresponds to the frustration imposed by the immobile cargo ., Similar behavior was seen in the case of the weak interaction ( Fig . 6C ) ., We found it interesting to plot the trajectories in the plane ( zrelative , EB ) zrelative: the relative position of the cargo ( zcargo-zhead ) , EB: the binding-energy both for the cases with and without the cargo-analog ( Fig . 6D ) ., Trajectories start from the right-lower area in ( EB , zrelative ) plane ., With the large cargo ( red and blue ) , after the relaxation of the binding energy from the initial condition to 0 kBT ( the detachment ) , the cargo-analog moved ., Whereas , without the cargo-analog , the C-terminus fluctuation occurred first and then KIF1A head may or may not detach from MT ( orange and dark-green ) ., The difference comes from the different time scale for the mobility of the cargo-analog ., Experimentally , the binding free energy of KIF1A head with MT was estimated from the dissociation constant as ∼−20 kBT in the ADP bound state 19 ., In the current simulations , the binding energies in the D-phase are −35 kBT for the strong interaction case ( see Fig . 6B ) and −17 kBT for the weak interaction case ( Fig . 6C ) ., Note that the experimental estimate is the free energy about the standard state , while the estimates from simulations are merely interaction energies ., Thus these numbers should not be quantitatively compared ., With the uncertainty in mind , perhaps , the real binding strength may fall in between the strong and the weak interaction cases ., Next , we analyze the diffusion and the attachment processes of KIF1A head after the detachment in ADP-state ( Fig . 7 ) ., The attachment rate for cargo/strong is larger than that for cargo/weak , as expected ., Interestingly , the attachment rate for cargo/weak was much smaller than that for stand-alone/weak , probably due to the restricted motions anchored by the large cargo-analog ., Thus , the large cargo-analog enhanced the detachment , but retarded the attachment ., Fig . 7B shows a transient histogram for the z-coordinates of the KIF1A head and of the cargo-analog soon after the detachment from MT . With the cargo-analog ( Fig . 7B left and middle ) , its positions were nearly fixed , whereas the head fluctuated broadly ( ∼4 nm in both directions ) , which coincides with the length of neck linker ., We note that , since we measure the diffusion after the detachment from MT , the histograms for cargo/strong and for cargo/weak are nearly the same: The diffusion process itself ( up to the attachment ) was not affected by the interaction strength ., As the diffusion time increases , the cargo-analog slowly moves , which enables the head to reach the backward site , as well as the forward site ., For the stand-alone case ( Fig . 7B right ) , C-terminus position diffused quickly after the detachment from MT , and the distributions of the C-terminus and the head were nearly symmetric about the starting point ( 0 nm ) ., In the simulations , the average times τattachment for attachment of the KIF1A head to MT for the system cargo/strong and cargo/weak were ∼0 . 2×108 τ ( ∼2 . 5 µs ) and ∼0 . 5×108 τ ( ∼6 . 4 µs ) , respectively ., A rough estimate of the diffusion length in this time scale is ∼1 . 2–1 . 9 nm , which is small ., After ATP binding , the neck-linker docked to the head core ., The neck-linker docking moves the cargo-analog by about +8 nm when the head landed to the forward site ( Fig . 8 ) ., The docking rate depends on the size of the cargo-analog , as expected ., Only in the cases of the weak interaction , after the attachment of the head onto MT , occasionally the head re-detached from and then re-attached to MT ( Fig . 9 ) ., This extra processes , being not coupled with ATP cycle , did not produce significant bias in the KIF1A move ., In the above simulations , the cargo was always placed at z∼4 . 25 nm based on the ATP-form reference structure , which may raise a concern that this specific initial positioning may affect the stepping statics ., To address this concern , we performed the same type of one-ATP cycle simulations with various initial cargo positions z; z\u200a=\u200a4 . 75 , 4 . 25 , 3 . 75 , 3 . 25 , 2 . 75 , 2 . 25 , 1 . 75 , and 1 . 25 nm especially for the cargo/strong case ., This range corresponds to the range of cargo found at the end of original simulations ( see Fig . S7 A the upper panel which shows the distribution of the probability density for the relative position of the cargo at the end of original simulation ) ., From each of these cargo ( initial ) positions , 10 simulations were conducted ., From the initial position of the cargo: z>3 nm , we found clear forward-biased moves , whereas z<3 nm , the head seldom detached from MT ( the lower panel of Fig . S7A ) ., Overall , by distributing the initial cargo positions , the forward bias is somewhat reduced on average ., Importantly , however , we still clearly see , on average , forward-biased moves of KIF1A with the cargo ., In this paper , we primarily focused on the specified molecular construct ( C351 ) used in in-vitro motility assay experiments 19 , 20 , 21 , in which the length of neck-linker except for His-tag is 22-residues ., Here , to test robustness of our results , we investigated the stepping statics of another construct that has 5-residue longer neck-linker ., The 5-residue segment is modeled as a flexible chain ( by Modeller ) ., In a similar way to the above sub-section , we estimated the range of the cargo position ( Fig . S7B upper panel which shows the distribution of the probability density for the position of the cargo ) , and repeated simulations ( 10 runs each ) with the initial cargo position at z\u200a=\u200a6 . 25 , 5 . 75 , 5 . 25 , 4 . 75 , 4 . 25 , 3 . 75 , 3 . 25 , 2 . 75 , 2 . 25 , 1 . 75 , and 1 . 25 nm ., From the initial position of the cargo: z>4 . 5 nm , we found clear forward-biased moves , whereas z<4 . 5 nm , the head seldom detached from MT ( Fig . S7B lower panel ) ., Thus , although the bias is weakened , we still see clear forward-biased moves of KIF1A with the cargo linked by 5-residue longer linker ., Conventional kinesin is dimeric and “walks” in a hand-over-hand fashion , akin to human walking by two legs ., Extending the analogy to human walking , the current simulations suggest that the large cargo-analog can play the role of a cane for the walk of single-headed kinesin; with a cane , we can walk even with one leg ., Although in this work , we focused on a truncated KIF1A which is used in the single-molecular assay of Okada et al 19 , 20 , 21 , we should note that the cellular function of KIF1A in vivo is markedly more complicated than the situation we considered here ., Several experiments 40 , 41 showed that KIF1A may be dimerized by virtue of being bound to a single cargo-analog in some case ., Our model does not straightforwardly apply to the dimeric KIF1A system in vivo ., Next , we discuss in vitro experiments of related systems ., First of all , forward-biased movements were observed for single-head kinesins , both KIF1A and a single-head construct of conventional kinesin mutant , with latex beads linked to C-terminus , where the size of beads are sub-µm to µm 13 , 21 ., Thus the current simulations are perfectly consistent with these results ., In a study of myosin VI , single-molecule experiments reported very similar phenomenon to our simulations 3

Introduction, Results, Discussion, Materials and Methods

Kinesin is a family of molecular motors that move unidirectionally along microtubules ( MT ) using ATP hydrolysis free energy ., In the family , the conventional two-headed kinesin was experimentally characterized to move unidirectionally through “walking” in a hand-over-hand fashion by coordinated motions of the two heads ., Interestingly a single-headed kinesin , a truncated KIF1A , still can generate a biased Brownian movement along MT , as observed by in vitro single molecule experiments ., Thus , KIF1A must use a different mechanism from the conventional kinesin to achieve the unidirectional motions ., Based on the energy landscape view of proteins , for the first time , we conducted a set of molecular simulations of the truncated KIF1A movements over an ATP hydrolysis cycle and found a mechanism exhibiting and enhancing stochastic forward-biased movements in a similar way to those in experiments ., First , simulating stand-alone KIF1A , we did not find any biased movements , while we found that KIF1A with a large friction cargo-analog attached to the C-terminus can generate clearly biased Brownian movements upon an ATP hydrolysis cycle ., The linked cargo-analog enhanced the detachment of the KIF1A from MT . Once detached , diffusion of the KIF1A head was restricted around the large cargo which was located in front of the head at the time of detachment , thus generating a forward bias of the diffusion ., The cargo plays the role of a diffusional anchor , or cane , in KIF1A “walking . ”

It is one of the major issues in biophysics how molecular motors such as conventional two-headed kinesin convert the chemical energy released at ATP hydrolysis into mechanical work ., While most molecular motors move with more than one catalytic domain working in coordinated fashions , there are some motors that can move with only a single catalytic domain , which provides us a possibly simpler case to understand ., A single-headed kinesin , KIF1A , with only one catalytic domain , has been characterized by in vitro single-molecule assay to generate a biased Brownian movement along the microtubule ., Here , we conducted a set of structure-based coarse-grained molecular simulations for KIF1A system over an ATP hydrolysis cycle for the first time to our knowledge ., Without cargo the simulated stand-alone KIF1A could not generate any directional movement , while a large-friction cargo-analog linked to the C-terminus of KIF1A clearly enhanced the forward-biased Brownian movement of KIF1A significantly ., Interestingly , the cargo-analog here is not merely load but an important promoter to enable efficient movements of KIF1A .

physics, statistical mechanics, theoretical biology, biophysics theory, biology, biophysics simulations, biophysics

journal.pgen.1000091

Genome-Wide Analysis Reveals a Complex Pattern of Genomic Imprinting in Mice

Genomic imprinting refers to the phenomenon of monoallelic gene expression that depends on the parent-of-origin of alleles , where either the maternally or paternally inherited copy is expressed while the other copy is silenced 1 , 2 ., The uniparental expression of paternally and maternally derived genes is usually caused by an epigenetic mark of differential methylation set during gametogenesis 2 ., Imprinted genes have been shown to crucially affect the development and expression of complex traits such as growth and development 3 throughout life , ranging from early embryonic stages to postnatal and adult phenotypes , and often with tissue specific expression e . g . 4 , 5 ., For example , studies have demonstrated that imprinted genes affect cognitive abilities 4 , 6 , several major human disorders ( e . g . Prader-Willi and Angelman syndrome ) , and possibly obesity 7–11 ., In the quest to investigate the genetic basis of such complex traits , genome-wide association and linkage studies have become a powerful tool e . g . 12 , 13 where regions of the genome are identified that contain sequence variants associated with phenotypic variation or the presence of a specific disorder ., However , very few of such studies have attempted to include investigations of epigenetic variation caused by genomic imprinting despite the known significant effects on complex traits e . g . 14 , 15 ., Most of our current knowledge about the number , distribution and effects of imprinted genes comes from studies using gene-targeting techniques focusing on regions of the genome with chromosomal aberrations 1 ., Under this methodology , several large-effect imprinted genes and their gross phenotypic effects have been described e . g . 3 , 4 ., Most prior studies on imprinting assume the traditional phenotypic pattern resulting from the monoallelic expression of the paternal or maternal allele e . g . 10 , 16 ., Yet , more complex patterns exist such as the callipyge phenotype described in sheep , where one of the two heterozygotes shows muscular hypertrophy while the other three genotypes have normal appearance and do not differ from each other 17 ., Moreover , other studies demonstrated that loci can deviate from the typical imprinting patterns of uniparental expression where loci may show differential expression of the two parental alleles e . g . partial expression: 18 ., In addition , recent work suggests that many more loci may be imprinted than previously assumed 19 ., However , little is known both about the quantitative effects of genomic imprinting and the diversity of patterns of expression ., Furthermore , while both large and small additive and dominance effects have been successfully mapped for a wide variety of traits 20 , relatively little empirical research has been conducted into the nature and effects of allelic diversity on quantitative trait variation at imprinted loci ., Data on alleles with relatively minor quantitative effects could potentially have important implications for normal physiological and behavioral variation and the expression of complex disease-related traits ., Most studies of genomic imprinting have focused on complete knock-outs of a specific locus and , thus , reveal limited information on the effects of different less severe alleles ., Using a three-generational intercross between two inbred strains originally selected for divergent adult body size LG/J and SM/J; 21 , we scanned the genome for loci showing significant parent-of-origin effects on body size and growth traits ., We present a hypothetical model for the functional origin of these complex effects and demonstrate that further dissection of imprinted quantitative trait loci ( iQTL ) is likely to yield a more comprehensive understanding of the complex patterns and likely evolutionary origins of imprinting ., In a genome-wide scan for iQTL , we detected ten loci on six chromosomes showing significant parent-of-origin dependent effects that were characterized by a diversity of genomic imprinting patterns ( Table 1 ) ., Five of these loci exceeded the genome-wide significance threshold and four were significant at the chromosome level ., The remaining locus was identified as having an imprinting effect ( i ) that was significant at the chromosome level , but the overall test for the locus was not significant ., However , because of the strong parent-of-origin effect at the locus , it was included as a suggestive iQTL ., Significance tests using the fit of the various possible forms of imprinting ( see below ) suggest that 8 of the 10 loci were significant at the genome-wide threshold , with the other two being significant at the chromosome level , lending additional support to our findings that these loci represent true iQTL ., Post-hoc analyses tested whether the parent-of-origin effect appeared in heterozygous offspring of heterozygous mothers and confirmed that the parent-of-origin dependent effects of these ten loci were indeed caused by genomic imprinting and not maternal genetic effects ., These analyses revealed that the parent-of-origin dependent effects of five other loci were caused by maternal genetic effects , and consequently , these loci are discussed elsewhere 22 ., Genomic imprinting was found to affect all weekly weights and growth traits ., The effects of the iQTL are generally pleiotropic , with effects on weight at different stages in development ., The effects of four of these loci show a change in the imprinting pattern over time ( Table 1 ) ., The dashed line in Table 1 indicates the weaning age , and it is noteworthy that almost all iQTL effects occurred after weaning ., Imprinted QTL for weight and growth are identified as WtiX . Y where Wt specifies body weight , i specifies an imprinting effect , X specifies the chromosome and Y specifies the iQTL on the chromosome that is being referred to ., On chromosome 3 we detected several iQTL , with the proximal QTL ( Wti3 . 1 ) being more than 50cM away from the distal ( Wti3 . 2 ) and the two are thus regarded as independent from each other ., In addition , two-QTL mapping analysis ( see Methods ) revealed a third QTL on distal chromosome 3 ( Wti3 . 3 ) ., Across all loci , we discovered five different patterns of imprinting: paternal expression , maternal expression , polar over- and underdominance imprinting and bipolar dominance ., Figure 1 and Material and Methods contain a description of the forms of imprinting , which are defined by the relationship of the imprinting genotypic value i to the additive and dominance genotypic values ( a and d ) as well as their sign ., Six loci showed paternal expression at some point during development ( e . g . Wti1 . 1; Figure 2A ) , with four of these showing exclusively paternal expression through development ., Only one locus ( Wti3 . 1 ) showed maternal expression ., Three loci showed a hitherto undescribed pattern which we refer to as bipolar dominance ( e . g . Wti3 . 2; Figure 2B ) where the two heterozygotes are significantly different from each other but the two homozygotes have similar phenotypes and are not different from each other ., Four loci showed polar dominance imprinting ( e . g . Wti5 . 1; Figure 2C ) , with three of the four showing polar overdominance ., Most loci maintained the same pattern over ontogeny , however , four loci showed a change in expression pattern through time ., For example , Wti5 . 1 showed paternal expression in week 4 ( Figure 3A ) , but the pattern gradually changed to polar overdominance through time ., By week 7 , the best fit model was for polar overdominance , and by week 10 the pattern was very clearly polar overdominance ( Figure 2C ) ., The change in the pattern of the ordered genotypes was caused by polar overdominance for growth after weaning ( Figure 3B ) , with the LS heterozygote growing faster than the other three genotypes ., Similarly , locus Wti2 . 1 showed bipolar dominance early in development ( with the bipolar pattern being the best fit pattern from week 1 to week 6 ) , but by week 7 the pattern had shifted to maternal expression ., Likewise , Wit2 . 1 changed from polar underdominance to paternal expression from week 1 to week 3 ., Many of the loci showed patterns consistent with partial imprinting , where the difference between the two homozygotes is larger than the difference between the two heterozygotes ., This can be seen in the relationships between the additive or dominance genotypic values to the imprinting genotypic value ( a/i or d/i; see Table S1 ) , which deviate from the values expected for a particular form of imprinting ., For example , most loci showing paternal expression have much larger additive effects than imprinting effects; resulting in a/i ratios larger than 1 ., In many cases of paternal expression , the additive effect is more than twice the imprinting effect; this is illustrated in Figure 4 for the effect of locus Wti7 . 1 on week 6 weight , where the additive effect is just over twice the imprinting effect ., The effects of the iQTL ( including their additive , dominance and imprinting effects ) together accounted for between 1 and 15 . 8% of the phenotypic variance in age-specific weights and weight gain , with the imprinting effects alone accounting for between 0 . 8 and 5 . 7% of the phenotypic variance ., The overall effects of individual loci ( i . e . , their additive , dominance and imprinting effects together ) explained between a 0 . 34 and 6 . 5% of the phenotypic variance ( with an average of 1 . 6% ) , with the imprinting effects alone accounting for between 0 . 2 and 1 . 7% of the phenotypic variance ., Surprisingly , the strongest effect of genomic imprinting on weight occurred between weeks 6 and 10 ( Table 1 ) ., Turning to the duration of imprinting effects and their onset we found that most imprinted loci showed significant effects over long periods during development , with only a single locus ( Wti2 . 1 ) showing effects limited to the pre-weaning period ., The QTL with imprinting effects over the longest period is located on proximal chromosome 3 ( Wti3 . 1 ) with a significant effect on most weights from week 1–9 ., This study advances research on genomic imprinting in several ways ., First , by using genotypes from two rather than one generation we can assign parent-of-origin of alleles with near certainty without invoking probabilities to calculate likelihoods of allelic parent-of-origin ., Thus , we have been able to examine in detail the pattern of phenotypic variation caused by genomic imprinting , and found previously unknown patterns of imprinting ., Second , these results suggest that imprinting patterns may be more diverse and , consequently , the traditional view of predominantly paternally or maternally expressed loci should be replaced with a picture of multiple imprinting patterns ( Figure 1 ) ., Indeed , most iQTL detected in our study display patterns other than simple paternal or maternal expression , with three loci showing the new bipolar dominance imprinting pattern ., Third , an important implication of our results is that the effects of alleles may change sign depending on their parent-of-origin ( see below ) ., These parent-of-origin-dependent allelic effects may also be akin to dominance in that the effect of an imprinted allele not only depends on its parent-of-origin , but also on the allele it is paired with at a locus ., Finally , the results of this study demonstrate that imprinting effects can vary over time both in their patterns ( Figure, 3 ) and the proportion of variance explained , and may arise or persist well into adulthood ., The latter highlights that imprinting effects are not necessarily most influential at early stages in development as currently viewed e . g . 16 ., The processes underlying the diversity of imprinting patterns found in our study are likely due to different mechanisms 23 many of which may involve differentially methylated DNA elements called imprinting centres regulating multiple genes in a region 24 ., Wood & Oakey 23 discuss three different mechanisms that may explain uniparental expression patterns ., While the enhancer-blocker model invokes an imprinting centre between reciprocally expressed genes with shared enhancer elements ( e . g . Igf2/H19 ) , a second model for the maternally expressed Igf2r gene utilizes cis-mediated silencing of maternally expressed genes by non-coding paternally expressed RNA ( e . g . Igf2r ) ., Finally , at microimprinted domains oocyte-derived methylation in the promoter region of protein-coding genes is assumed to be the key mechanism ., In addition to the ‘traditional’ imprinting patterns we found four loci with polar dominance effects causing a pattern equivalent to that described for the callipyge ( CLPG ) locus in sheep and pig homologues DLK1-GTL2 17 , 25 where one of the two heterozygotes is different from the other three genotypes 26 ., For the callipyge locus , the observed pattern is caused by a paternally inherited mutation in the CLPG locus that results in the expression of a number of core group genes in cis in addition to an interaction in trans between reciprocally imprinted genes 26 ., The authors proposed that inhibition of DNA methylation or altered histone modification may be causal to the callipyge phenotype ., To our knowledge , a pattern of polar dominance has only once been reported previously for any known murine gene 15 , and in that case the locus effect was lethality , not trait expression ., Furthermore , while polar overdominance has been found for one locus in sheep and pigs , no prior studies in any system have observed a pattern of polar underdominance imprinting affecting trait expression as demonstrated by our results ., We suggest that the pattern of bipolar dominance may be explained by a model where the sign of the allelic effect changes depending on the parent-of-origin ., This might occur when two differentially imprinted genes are in close linkage ( e . g . callipyge ) , such that the alternative alleles are composed of variants at both the maternally and the paternally expressed loci ., This scenario is illustrated in Figure 5 , showing a hypothetical case in which a QTL with alleles 1 and 2 is comprised of two variable sites , A and B , that are in close linkage , with gene A being paternally expressed while B is maternally expressed ., The effect of the paternally derived QTL copy will be determined by variation at site A while the effect of the maternally inherited copy will be determined by variation at site B . In this scenario , allele 1 of the QTL may have a positive effect on a trait when paternally inherited but a negative effect when maternally inherited whereas allele 2 may show the opposite pattern ., When the same allele ( 1 or, 2 ) is inherited from both the father and the mother the effects cancel out , yielding no difference between the two homozygotes ., However , if two different alleles are inherited from the parents then the joint effects of the paternal and maternal copy do not cancel and , as a result , produce a pattern of bipolar dominance ( Figure 1 ) ., As with the callipyge locus in sheep 26 , this pattern of ‘interference’ between closely linked maternally and paternally expressed loci could potentially be a signature of conflict , where concerted counter-evolution of maternally and paternally expressed alleles results in linked alleles that negate each others effect ., The currently known number of imprinted loci in mice ( about 80; www . geneimprint . com ) may in part reflect a research bias toward regions of the genome with chromosomal aberrations and loci with large phenotypic effects , especially in the light of recent research showing that as many as 600 genes are predicted to be imprinted 19 ., First , comparing the locations of iQTL found in our study with those of currently known imprinted genes ( www . geneimprint . com ) , we find that most of our loci are likely to be novel ., No currently known imprinted genes are located on chromosomes 1 , or 3 , where we detected a total of four iQTL ., There are known imprinted genes on chromosomes 2 and 5 , but they all lie well outside of the confidence intervals 27 for the iQTL locations ( ca . 100Mb away on chr . 2 , 40Mb away on chr . 5 ) ., Chromosome 12 has a number of imprinted genes that are located close to but outside of either end of the confidence interval , with Mirn337 more than 20Mb proximal and several genes ( Dlk1m , Gtl2 , Rtl1 , Dio3 ) more than 10Mb distal to the confidence interval ., The iQTL with the strongest effect ( Wti7 . 1 ) is located on Chromosome 7 , which contains nearly half of the currently confirmed imprinted genes in mice ( www . geneimprint . com ) ., The confidence interval for the location of this locus includes 17 known imprinted genes ( ca . 20–24% of all currently confirmed or putatively imprinted genes in mice; www . geneimprint . com ) , including for example , Peg3 , Peg4 ( Snrpn ) , Peg6 ( Ndn ) and Peg12 ., More than half of these ( 10 of 17 ) and 10 of 14 loci with a known imprinting pattern are characterized by paternal expression , which matches the pattern we identified for the iQTL in this region ., On chromosome 9 , the imprinted gene Rasgrf1 falls within the confidence region of Wti9 . 1 ., Rasgrf1 was the first imprinted gene found to affect postnatal growth only 5 and has been described as paternally expressed ., While the postnatal effect of Rasgrf1 is congruent with the effect on postweaning growth found for Wti9 . 1 , we found a bipolar pattern for growth and polar overdominance for weekly weights affected by this locus in contrast to paternal expression reported for Rasgrf1 ., However , we note it is unclear whether a bipolar pattern would emerge for Rasgrf1 for the post-weaning growth period of 3–10 weeks ., Further studies both on the growth trajectories of Rasgrf1 mutants and on fine-mapping our identified iQTL are required to determine whether Rasgrf1 could be a candidate gene ., Finally , the confidence interval for the location of Wti14 . 1 includes a single known imprinted gene , Htr2a , which is known to be maternally expressed in mice , in contrast to the pattern of paternal expression seen for the iQTL , suggesting it is an unlikely candidate gene ., Turning to the results of Luedi et al . 19 whose simulation study predicted 600 imprinted genes across the genome , we found that a total of 50 predicted genes are within the confidence regions of our iQTL ( Table 2 ) ., While one may expect some congruence of our confidence regions and the list of predicted genes from Luedi et al . by chance ( 5 per QTL locus ) , several confidence regions contain a large number of predicted imprinted genes ( and the predicted imprinted genes are not uniformly distributed across the genome ) ., Confidence intervals for all loci contain multiple genes predicted to be imprinted , providing potential candidate genes for all iQTL that could be explored in future fine mapping and methylation-status studies to search for imprinted genes ., Somewhat surprisingly , we found only two loci that show imprinting effects during the pre-weaning period from 1 to 3 weeks and only one of these loci ( Wti2 . 1 ) has effects that are restricted entirely to the preweaning period ., Many more loci show effects that do not appear until week 5 or later and many of these extend to mature adult weight at 9 or 10 weeks ., Additive and dominance QTL mapping in this population of mice has shown that different sets of QTL affect variation in growth and body weight between 1 and 3 weeks of age and between 4 and 10 weeks 28 , 29 , paralleling the known differences in the physiology of mammalian growth over these periods 30 ., These previous analyses of additive and dominance effects have shown that the number of QTL affecting weekly weights and growth does not vary greatly before and after weaning , but that dominance effects tend to be more important earlier ( peaking around weaning ) while additive effects tend to increase in magnitude with age ., Following these results we have no reason to assume that a bias exists for finding imprinting effects before or after weaning ., Overall , our results suggest that the quantitative analysis of imprinting effects using allelic variation can identify genomic regions showing novel imprinting effect patterns ( e . g . , bipolar dominance ) ., Moreover , by not restricting our analysis to traits expressed early in life we demonstrate that imprinting effects can appear and often be stronger later in life ( and notably , after the cessation of maternal care ) , and may also change their pattern of effect during growth and development ., More generally , our investigation provides a framework for classifying the diversity of patterns that imprinted loci may show ( Figure 1 ) ., Further investigation into the proximate causes of the underlying processes that generate these novel imprinting patterns may ultimately provide important insights into the evolutionary origin of imprinting and multiple pathways in which imprinting contributes to quantitative trait variation ., We used the F2 and F3 generation of an intercross between the inbred mouse strains Large ( LG/J ) and Small ( SM/J ) 31 ., These strains were established over 60 years ago and were originally under artificial selection for either large or small body weight at 60 days of age 21 , 32 , 33 and have been inbred for over 120 generations prior to their use in this study ., Due to this extended period of inbreeding , these strains are essentially devoid of within-strain variation ., The strains differ by 6–8 standard deviations in size and growth related traits 31 , making this an ideal model system to study imprinting effects arising from genes regulating growth and development ., To generate the study population , ten males of the SM/J strain were mated to ten females of the LG/J strain ., The resulting F1 population consisted of 52 individuals , which were randomly mated to produce 510 F2 animals , representing the parental generation in our study ., These F2 animals , again , were randomly mated to produce 200 full-sibling families of the F3 generation with a total of 1632 individuals ., Males were removed from the cages when females were visibly pregnant ., Half litters were reciprocally cross-fostered at random between pairs of females that gave birth on the same day ., In total , offspring in 158 families were cross-fostered in this way ., Pups were weaned at 21 days of age and randomly housed with three or four other same sex individuals ., Further details of the husbandry are given in 28 , 29 ., Pups were weighed weekly starting at one week of age through week 10 using a digital scale with an accuracy of 0 . 1g ., Growth was calculated as the difference between weekly weights such that , for example , the growth from week 1 to week 3 is the difference between week 3 weight and week 1 weight ., The traits analysed in this study are weekly individual bodyweights corrected for sex and litter size beginning with weight at week 1 and ending with weight at week 10 ., Growth traits were obtained for preweaning growth from week 1 to 3 and for the postweaning growth from week 3 to 10 ., DNA was extracted from livers of the F2 and F3 individuals using Qiagen DNeasy tissue kits ., After standardizing DNA concentration , the samples were scored for 384 SNPs using the Golden Gate Assay by Illumina , San Diego , USA ., These markers were previously found to be polymorphic between LG/J and SM/J as part of the Oxford/CTC genotyping collaboration ( http://www . well . ox . ac . uk/mouse/INBREDS/ ) ., After further testing , 15 loci were found not to have been reliably scored and were excluded from the analysis ., Sixteen loci were scored on the X chromosome and are not included in this analysis because the genome structure and the statistical model for the X are complex and unresolved ., This leaves 353 loci across the 19 autosomes for analyses ., A genetic map of these markers based on Haldanes centiMorgans ( cM ) was produced using R/QTL 34 and validated against the genome coordinate locations in the Ensembl database ( www . ensembl . org ) ., The average map distance between markers in the F2 generation is 4 cM ., Markers are evenly placed throughout the genome except for regions in which LG/J and SM/J have been found to be monomorphic 35 ., A list of the markers along with their physical and map positions are given in Table S2 ., The combined genotypes of parents and offspring were used to reconstruct haplotypes for all animals with the program PedPhase 36 , 37 , which uses several algorithms to infer haplotype configurations for all individuals that minimize the number of recombination events in the whole pedigree i . e . , it solves the ‘minimum-recombination haplotypes configuration problem’; 38 ., We used the ‘block-extension algorithm’ to reconstruct haplotypes , which produced a set of unordered haplotypes for the F2 animals and a set of ordered haplotypes ( i . e . , ordered by parent-of-origin of alleles ) for the F3 animals ., The ordered genotypes of the F3 allowed us to distinguish between the four possible genotypes at a given locus , LL , SL , LS or SS ( L being the LG/J allele and S the SM/J allele ) where the first allele refers to the paternally derived allele and the second to the maternally derived allele ., The four ordered genotypes at the marker loci ( LL , LS , SL and SS ) were assigned additive , dominance and imprinting ( parent-of-origin ) genotypic index values following Mantey et al . 15 ., These index values ( slightly modified from those used by Mantey et al . ref . 15 ) can be written in matrix form , where the vectors of genotypic means ( i . e . , genotypic values ) ( ) are defined by: ( 1 ) which yields estimates of the parameters: ( 2 ) Where r is the reference point for the model ( the mid-point between homozygotes ) , a is the additive genotypic value ( half the difference between homozygotes ) , d is the dominance genotypic value ( the difference between the mean of the heterozygotes and the mid-point of the homozygote means ) , and i is the parent-of-origin or imprinting genotypic value ( half the difference between heterozygotes ) cf . ref . 15 ., These index values in equation ( 1 ) were used to build a model to scan the genome in the F3 generation ( parent-of-origin of alleles cannot be directly assigned in the F2 because their F1 parents are all genetically identical , making it impossible to unambiguously assign haplotypes to parents ) to detect loci showing significant parent-of-origin-dependent effects ( i . e . significant i effects ) ., A mixed general linear model was used to estimate the overall significance of a locus as well as the significance of the additive , dominance , and imprinting effects ., To test the overall significance of a locus , a model with ordered genotype class as a fixed effect and family as a random effect was fitted using restricted maximum likelihood ( REML ) as implemented in the Mixed Procedure of SAS ( SAS version 9 . 1; SAS Institute , Cary , NC , USA ) ., The significance of the individual genetic effects was determined using a mixed model with the a , d and i index values as fixed regression effects and family as a random effect ( fitted again using REML in the Mixed Procedure of SAS ) ., Family was included as a random effect to account for variation among families not attributable to the effects of the locus in question ., A power analysis combined with a simulation to determine significance thresholds ( see below ) showed that the inclusion of family greatly improved power while also removing any bias in significance tests introduced by family structure ., The mixed model with the fixed genetic effects and random family effect was used to scan the genome to produce a probability distribution for the overall effect of the locus as well as the additive ( a ) , dominance ( d ) and imprinting ( i ) effects ., These probability values were then transformed to a logarithmic probability ratio ( LPR ) in order to make them comparable to the LOD scores typically seen in QTL analyses ( LPR\u200a=\u200a−log10 ( probability ) ) ., Significance thresholds were determined using a Bonferroni correction , which was calculated using the effective number of markers method 39 , which has been demonstrated to be less artificially conservative than a simple Bonferroni correction ., This analysis showed that , due to correlations between linked markers , the genome has 133 effective markers , which results in a Bonferroni threshold LPR at the 5% level ( i . e . , α\u200a=\u200a0 . 05 ) of 3 . 41 ., Chen & Storey 40 have shown that , where several QTL can be expected to affect traits , a modified genome-wide error rate should be applied as opposed to the traditional genome-wide error rate or the false discovery rate ., This is achieved by applying the significance criterion to the highest LPR on each chromosome and yields overall the best results by increasing the discovery of true positives while at the same time avoiding problems using the false discovery rate in gene mapping experiments ( with 19 autosomes , we would expect only about 1 false positive result using the chromosome level thresholds ) ., Our collective chromosome-wide significant results across the genome greatly surpass this expectation providing confidence in the overall set of results ., Therefore , for each chromosome we used the effective number of markers on the chromosome 40 to generate a chromosome-level significance threshold ., The thresholds for individual chromosomes are given in the Supplementary Table ., Once a QTL was identified , we used post-hoc tests , with a LPR significance threshold of 1 . 3 ( i . e . , p<0 . 05 ) to determine whether the locus also had additive , dominance or imprinting effects , or affected more than one trait ., Confidence intervals for the positions of iQTL were determined using a one LOD drop ( using LPR values ) following Lander & Botstein 27 ., Because apparent parent-of-origin effects at a locus can also be caused by a maternal effect of that locus , rather than genomic imprinting , we tested all loci with a significant i effect to confirm that the appearance of a parent-of-origin effect could not be attributed to maternal effects 22 ., If the apparent imprinting effect is due to a genetic maternal effect , the differences between reciprocal heterozygotes born of homozygous mothers will be much larger than the differences between those born of heterozygous mothers , which are all exposed to the same maternal environment ., Therefore , we confirmed the existence of imprinting by testing the i effect using only the offspring from heterozygous mothers ., This approach adequately accounts for the potential confounding patterns of maternal effects since maternal effects only lead to the appearance of a parent-of-origin dependent effect at the locus that has the maternal effect ., The occurrence of non-genetic maternal effects cannot lead to the appearance of parent-of-origin dependent effects and , likewise , the presence of maternal genetic effects attributable to other loci in the genome will not lead to the appearance of a parent-of-origin effect at other loci ., To determine the relative proportion of variance explained by the loci overall and by genomic imprinting effects , we calculated the approximate variance contributed by a locus ( Vg ) using the expectation: ( 3 ) which is the genetic variance of a locus in a population with two alleles at approximately equal frequency in Hardy-Weinberg equilibrium ., The analytical expectation was used because REML does not compute sums of squares and the corresponding R2 ., The proportion of variance explained would , therefore , be Vg/Vp ( Vp being the phenotypic variance ) ., To obtain the variance explained by the parent-of-origin effect alone we calculated ., When chromosomes contained more than one significant QTL , we assumed that all loci more than 50cM apart represented separate iQTL ., For cases where loci were closer than 50cM apart , we te

Introduction, Results, Discussion, Material and Methods

Parent-of-origin–dependent gene expression resulting from genomic imprinting plays an important role in modulating complex traits ranging from developmental processes to cognitive abilities and associated disorders ., However , while gene-targeting techniques have allowed for the identification of imprinted loci , very little is known about the contribution of imprinting to quantitative variation in complex traits ., Most studies , furthermore , assume a simple pattern of imprinting , resulting in either paternal or maternal gene expression; yet , more complex patterns of effects also exist ., As a result , the distribution and number of different imprinting patterns across the genome remain largely unexplored ., We address these unresolved issues using a genome-wide scan for imprinted quantitative trait loci ( iQTL ) affecting body weight and growth in mice using a novel three-generation design ., We identified ten iQTL that display much more complex and diverse effect patterns than previously assumed , including four loci with effects similar to the callipyge mutation found in sheep ., Three loci display a new phenotypic pattern that we refer to as bipolar dominance , where the two heterozygotes are different from each other while the two homozygotes are identical to each other ., Our study furthermore detected a paternally expressed iQTL on Chromosome 7 in a region containing a known imprinting cluster with many paternally expressed genes ., Surprisingly , the effects of the iQTL were mostly restricted to traits expressed after weaning ., Our results imply that the quantitative effects of an imprinted allele at a locus depend both on its parent of origin and the allele it is paired with ., Our findings also show that the imprinting pattern of a locus can be variable over ontogenetic time and , in contrast to current views , may often be stronger at later stages in life .

For certain genes , individuals express only the copy of the gene they inherit from either their mother ( “maternally expressed” genes ) or their father ( “paternally expressed” genes ) ., This “parent-of-origin–dependent” pattern of gene expression is known as genomic imprinting and has been shown to play an important role in modulating a variety of traits ranging from developmental processes to cognitive abilities and associated disorders ., While various molecular techniques have allowed for the identification of many imprinted genes , very little is known about the contribution of imprinting to variation seen among individuals in continuously varying traits such as body size ., Here we address this issue by using a genome-wide analysis aimed at finding regions of the genome that show an effect of imprinting on body weight and growth in mice ., We identified ten loci that displayed complex and diverse patterns of effect , including four loci with effects similar to the unusual callipyge mutation found in sheep and three that displayed a new phenotypic pattern that we refer to as bipolar dominance ., Surprisingly , most imprinting effects were strongest during the post-weaning period , and many showed shifts in the pattern of imprinting over ontogenetic time .

genetics and genomics/genomics, genetics and genomics/epigenetics, genetics and genomics/animal genetics, genetics and genomics/functional genomics

journal.pcbi.1004497

A Gene Gravity Model for the Evolution of Cancer Genomes: A Study of 3,000 Cancer Genomes across 9 Cancer Types

Cancer development and progression are mediated by the accumulation of genomic alterations , including point mutations , insertions and deletions , gene fusions , amplifications , and chromosomal rearrangements 1 , 2 ., The majority of the somatic mutations found in tumor cells are ‘passenger’ rather than ‘driver’ mutations 3 ., In 1976 , Peter Nowell wrote a landmark perspective for the clonal evolution model of cancer and applied evolutionary models to understand tumor growth and treatment failure 4 ., He proposed that most neoplasms arise from a single cell , and tumor progression results from acquired genetic variability within the original clone , allowing sequential selection of more aggressive sublines ., He also noted that genetic instability , occurring in tumor cells during disease progression , might enhance this process ., This view now has been widely accepted 4 , 5 ., Somatic cell evolution leads to adaptive cancer cell survival , including increased proliferative , angiogenic , and invasive phenotypes 2 ., However , understanding how somatic cell evolution drives tumorigenesis remains a great challenge in cancer research ., Genome instabilities , such as chromosomal instability and microsatellite instability , have been well studied in cellular systems 2 , 6 , 7 ., For example , Teng et al . found that in yeast a mutation on a single gene may cause genomic instability , leading to adaptive genetic changes 8 ., Whether and how human tumor genomes are genetically unstable , induced by single gene alterations , has been debated for decades 9–12 , but has recently gained much support ., For instance , Emerling et al . found an amplification of PIP4K2B in HER-2/Neu-positive breast cancer with its co-occurrence with mutations in TP53 11 ., They showed that a subset of breast cancer patients had a high level of gene expression of PIP4K2A and PIP4K2B and provided evidence that these kinases are essential for growth in the absence of p53 ., Liu et al . found that POLR2A ( encoding the largest and catalytic subunit of the RNA polymerase II complex ) was deleted together with TP53 in cancer cell lines and primary tumors in human colon cancer 13 ., Additionally , the DNA cytidine deaminase APOBEC3B-catalyzed genomic uracil lesions are responsible for a large proportion of both dispersed and clustered mutations in multiple distinct cancers 12 ., These lines of evidence show that single gene alterations may induce the mutations of other genes in a cancer genome that drive tumorigenesis and tumor progression 9–13 ., Thus , a quantitative assessment of whether the perturbation of any single gene in a cancer genome is sufficient to drive genetic changes would help us better understand tumorigenesis and tumor evolution through genomic alterations ., However , distinguishing functional somatic mutations from massive passenger mutations and non-genetic events is a major challenge in cancer research ., Massive genomic alterations present researchers with a dilemma: does this somatic genome evolution contribute to cancer , or is it simply a byproduct of cellular processes gone awry 14 ?, Cells consist of various molecular structures that form complex , dynamic , and plastic networks 15 ., In the molecular network framework , a genetic aberration may cause network architectural changes through affecting or removing a node or its connection within the network , or changing the biochemical properties of a node ( protein ) 16–18 ., The abundance of next-generation sequencing data of cancer genomes provides biologists with an unprecedented opportunity to gain a network-level understanding of tumorigenesis and tumor progression 15 , 19–22 ., However , how to integrate large-scale molecular networks with cancer genomic aberrations is highly challenging 9 , 10 ., The development of a mathematical model will be helpful to understand how genetic aberrations perturb the molecular network architecture and manifest the effects during tumorigenesis ., In this study , we proposed a novel mathematical model , namely gene gravity model , derived from Newton’s law of gravitation to study the evolution of cancer genomes ., The gene gravity model detects a gene-gene pair that two genes are co-mutated and highly co-expressed simultaneously in a given cancer type based on several previous evidences 8 , 11 , 13 ., As proof of principle , we applied the model to approximately 3 , 000 tumors’ transcription and somatic mutation profiles across 9 cancer types from The Cancer Genome Atlas ( TCGA ) project ., We found that cancer driver genes may shape somatic genome evolution by inducing mutations in other genes during tumorigenesis ., We identified six putative cancer genes by quantifying the gene gravity model ., Furthermore , we found a higher somatic mutation density related to cancer driver genes on the X chromosome in comparison to the whole autosomes , suggesting that hypermutation in inactive X chromosomes is a general feature in females ., In summary , this study would provide new insights into adaptive cancer genome evolution shaped by somatic mutations in cancer ., The gene gravity model postulates that if two genes have high mutation density and strong gene co-expression in a given cancer type , they should have a higher G score and related to a higher risk of inducing mutations to other genes; this postulation is based on several previous observations 8 , 11 , 13 ., We developed the gene gravity model by incorporating ~3 , 000 tumors’ transcription and somatic mutation profiles across 9 cancer types from TCGA under molecular network architecture knowledge ( Fig 1 ) ., These 9 cancer types consist of breast invasive carcinoma ( BRCA ) , colon adenocarcinoma ( COAD ) , glioblastoma multiforme ( GBM ) , head and neck squamous cell carcinoma ( HNSC ) , kidney renal clear cell carcinoma ( KIRC ) , lung adenocarcinoma ( LUAD ) , lung squamous cell carcinoma ( LUSC ) , ovarian serous cystadenocarcinoma ( OV ) , and uterine corpus endometrial carcinoma ( UCEC ) ., First , we collected 3 , 487 tumor transcription profiles ( RNA-Seq ) for the 9 cancer types ., Then , we constructed 9 co-expressed protein interaction networks ( CePINs ) for the 9 cancer types ( S1 Table ) respectively by incorporating the transcription profiles into a large-scale protein interaction network ( PIN ) in S2 Table and Fig 1A ., Each CePIN contained ~100 , 000 edges connecting ~12 , 000 genes ., Second , we collected 277 , 370 nonsynonymous somatic mutations identified from 2 , 946 tumor exomes across 9 cancer types from TCGA ( S1 Table ) ., For each cancer type , we projected the somatic mutations onto PIN to construct a somatic mutation PIN via a network propagation algorithm ( Fig 1B and 1C ) ., We then derived a G score for each gene-gene pair in the 9 cancer types , using Newton’s law of gravitation ( Fig 1C ) ., Then , we examined the G score for seven gene sets: cancer driver genes , cancer gene census ( CGC ) genes ( experimentally validated cancer genes ) , tumor suppressor genes ( TSGs ) , oncogenes , DNA repair genes , chromatin regulation factors ( CRFs ) , and essential genes ( Fig 1D ) ., Finally , we investigated the pattern of hypermutation of the inactive X chromosome in female versus male cancer genomes by quantifying cancer genome evolution using the gene gravity model ( Fig 1E ) ., To verify the gene gravity model , we investigated the enrichment of somatic mutations on protein-protein interaction ( PPI ) pairs as well as unfiltered interactions relative to the same number of random pairs based a previous study 23 ., We found that PIN is significantly more enriched for high mutation density than random pairs across the 9 cancer types ( q < 2 . 2 × 10−16 , Wilcoxon rank-sum test corrected by Benjamini-Hochberg multiple testing , S1 Fig ) ., We first examined the distribution of G score for two benchmark gene sets: DNA repair genes and CRFs ., The CRFs modulating the epigenetic landscape have emerged as potential gatekeepers and signaling coordinators for the maintenance of genome integrity 24 ., The enzymes encoded by DNA repair genes continuously monitor chromosomes to repair damaged nucleotide residues generated by exposure to carcinogens and cytotoxic agents ( e . g . , anticancer drugs ) 25 ., Thus , both CRFs and DNA repair genes are of critical importance for the maintenance of the genetic information in the cancer genome ., In this study , we collected two high-quality gene sets: 153 DNA repair genes 26 and 176 CRFs 27 ( S3 Table ) ., We defined a DNA repair gene-gene pair gravitational interaction as one or two genes in a pair is/are DNA repair genes ., A non-DNA repair gene-gene pair gravitational interaction was defined as neither of the two genes in a pair is a DNA repair gene ., We applied the same definition for the remaining 6 gene sets: cancer driver genes , CGC genes , TSGs , oncogenes , CRFs , and essential genes ., We then investigated the complementary cumulative G score ( S2–S10 Figs ) ., We found that the DNA repair gene cumulative G score is higher than that of non-DNA repair genes in 8 cancer types , except BRCA ., Furthermore , the CRF cumulative G score is higher than that of non-CRFs in all of the 9 cancer types ( S2–S10 Figs ) ., Collectively , these observations demonstrated that we could use the gene gravity model to quantitatively examine how perturbations of a single gene shape subsequent evolution of the cancer genome based on evidence in several previous biological studies 8 , 11 , 13 ., We investigated “high somatic evolutionary pressure” for a particular gene that tends to be co-mutated and highly co-expressed with other genes in a given cancer type ., We hypothesized that if a gene has a higher somatic evolutionary pressure , this gene may increase subsequent genetic changes 8 , 11 , 13 ., We compiled a high-quality , mutated cancer driver gene set ( 614 cancer driver genes , S3 Table ) from four pan-cancer genomic analysis projects 3 , 28–30 ., We found that the cancer driver gene cumulative G score is significantly higher than that of non-cancer driver genes in all of the 9 cancer types ( q < 2 . 2 × 10−16 , Wilcoxon rank-sum test , S2–S10 Figs ) ., These observations suggest that cancer driver mutations may increase subsequent genetic changes based on the previous studies 8 , 11 , 13 ., We also studied CGC genes , which are well curated and have been widely used as a reference cancer gene set in many cancer-related studies 31 , 32 ., As expected , we found that the CGC gene cumulative G score is higher than that of non-CGC genes in 6 cancer types: BRCA , COAD , GBM , HNSC , KIRC , and UCEC ( S2–S6 and S10 Figs ) ., However , the CGC gene cumulative G score is slightly higher than that of non-CGC genes in 3 cancer types: LUAD , LUSC , and OV ( S7–S9 Figs ) ., A previous study indicated that an average mutation frequency in smokers is more than 10-fold higher in never-smokers in non-small cell lung cancer 33 ., We next separated TCGA patients into smokers and never-smokers in LUAD and LUSC , and reexamined the CGC gene cumulative G score ., As expected , the CGC gene cumulative G score is significantly higher than that of non-CGC genes in LUAD and LUSC never-smokers ( q < 0 . 05 , S11 Fig ) ., However , the CGC gene cumulative G score is slightly higher than that of non-CGC genes in LUAD and LUSC smokers ( S11 Fig ) ., Thus , heterogeneous mutation frequencies and gene transcription profiles in the combined smokers and never-smokers in LUAD or LUSC may influence the performance of the gene gravity model 33 ., For OV ( S9 Fig ) , high genomic instability of the ovarian cancer genome may cause this slight gene cumulative G score between CGC and non-CGC genes 34 ., Finally , we considered essential genes ., We compiled 2 , 719 essential genes ( S3 Table ) from the Online GEne Essentiality database 35 ., S2–S10 Figs showed that the essential gene cumulative G score is higher than that of non-essential genes across 9 cancer types ., Remarkably , the cancer driver gene-gene G score is higher than that of essential genes ( q < 0 . 01 ) in all of the 9 cancer types ( S2–S10 Figs ) ., Tumorigenesis is dependent on the accumulation of one or multiple driver mutations that activate oncogenic pathways or inactivate tumor suppressors 36 , 37 ., Oncogenes often positively co-expressed with interacting partners due to gain-of-function mutations; while TSGs often negatively co-expressed with interacting partners due to lose-of-function mutations 38 ., Thus , we defined attractive gravitation ( AG ) as two genes that have positive gene co-expressed correlation and repulsive gravitation ( RG ) as two genes that have negative gene co-expressed correlation in a specific cancer type ., We compiled 477 oncogenes and 1 , 040 TSGs ( S3 Table ) , and then examined the AG and RG score for oncogenes and TSGs , respectively ., We found that the oncogene AG cumulative distribution is higher than that of non-oncogenes in 5 cancer types: BRCA , COAD , KIRC , OV , and UCEC ( S12 Fig ) ., However , as shown in S13 Fig , the oncogene RG cumulative distribution is similar or slightly higher than that of non-oncogenes in all of the 9 cancer types ., Additionally , we examined the AG and RG score for TSGs ., We found that both AG and RG cumulative distribution for TSGs is higher than that of non-TSGs in 7 cancer types , except LUSC and OV ( S14 and S15 Figs ) ., Taken together , our gene gravity model can distinguish one important tumor biological characteristics , oncogenic potential altered by oncogenes , very well ., However , our model fails to distinguish caretaker or gatekeeper roles altered by TSGs ., One possible reason is that some TSGs have both tumor suppressor and oncogenic activities in different cancer types or cell types ., For example , p21 , encoded by CDKN1A , plays both tumor suppressor activities and paradoxical tumor-promoting activities in cancer 39 ., In addition , it is partially because TSGs have truncated mutations that may scattered in the gene region ., Thus , further study will be needed for systematic investigation of the AG and RG score for TSGs , which we hope will be prompted by the findings herein ., We calculated the gene average gravitation ( aveG ) score using ( ρ ) i = ∑j Gij / n between gene i and gene j ( j belongs to the set of gene i’s interacting partners ( n ) in PIN ) ., We found that the aveG score of cancer driver gene is significantly higher than that of DNA repair , CGC , and essential genes in all of the 9 cancer types ( Fig 2 and S4 Table ) ., For BRCA , the cancer driver gene aveG score ( 0 . 47 ± 0 . 02 ) is significantly higher than that of DNA repair genes ( 0 . 30 ± 0 . 03 , q = 1 . 9 × 10−4 ) , CGC genes ( 0 . 35 ± 0 . 02 , q = 1 . 1 × 10−4 ) , and essential genes ( 0 . 26 ± 0 . 01 , q = 2 . 3 × 10−32 , S4 Table ) ., However , the cancer driver gene aveG score is similar to that of CRFs ( 0 . 42 ± 0 . 04 , q = 1 . 0 ) in BRCA ., Similar trends were observed in the remaining 8 cancer types ( S4 Table ) ., Thus , chromatin regulation might play an important role in tumorigenesis ., We further investigated whether genetic or epigenetic alterations have combinatorial effects that shape cancer genome evolution ., Since CRFs represent the epigenetic landscape 27 , we divided cancer driver genes into two subgroups: CRF cancer driver genes and non-CRF cancer driver genes ., We found cancer driver genes are significantly enriched in CRFs ( 38 out 176 CRFs versus 176 CRFs from 20 , 462 human protein-coding genes collected from National Center for Biotechnology Information NCBI database , p = 3 . 0 × 10−21 , Fisher’s exact test , Fig 3A ) ., Furthermore , the CRF cancer driver gene aveG score is higher than that of non-driver CRFs across 9 cancer types ( q < 0 . 10 , Fig 3A and S5 Table ) ., For KIRC , the CRF cancer driver gene aveG score ( 1 . 6 ± 0 . 48 ) is significantly higher than that of non-CRF cancer driver genes ( 0 . 76 ± 0 . 04 , q = 4 . 2 × 10−3 ) and non-driver CRFs ( 0 . 72 ± 0 . 09 , q = 3 . 3 × 10−3 , S5 Table ) , respectively ., However , we did not find a significant aveG difference between non-CRF cancer driver genes and non-driver CRFs in any of the 9 cancer types ( q = 1 . 0 , Fig 3A and S5 Table ) ., We next divided CGC genes into two subgroups: CRF CGC genes and non-CRF CGC genes ., We found that CGC genes are significantly enriched in CRFs as well ( p = 1 . 2 × 10−15 , Fisher’s exact test , S16A Fig ) ., As expected , we did not observe a significant aveG difference between non-CRF CGC genes and non-CGC CRFs in 7 cancer types ( q > 0 . 05 , S6 Table ) , with the exception of OV ( q = 0 . 04 ) and KIRC ( q = 0 . 04 ) ., Put together , the cancer genome evolution might be shaped by the combinatorial synergy between cancer driver genes and CRFs ., We next divided cancer driver genes into two subgroups: DNA repair cancer driver genes and non-DNA repair cancer driver genes ., Fig 3B showed that DNA repair genes tend to be cancer driver genes as well ( 18 out 153 DNA repair genes versus 153 DNA repair genes from 20 , 462 human protein-coding genes collected from NCBI database , p = 1 . 1 × 10−6 ) ., However , CRFs are more likely to be cancer driver genes than DNA repair genes ( p = 0 . 02 ) ., The DNA repair cancer driver gene aveG score is similar to that of non-DNA repair cancer driver genes in 6 cancer types ( q > 0 . 1 ) , except of HNSC ( q = 0 . 02 , S7 Table ) , KIRC ( q = 0 . 08 ) , and LUAD ( q = 0 . 08 ) ., However , the DNA repair cancer driver gene aveG score is significantly higher than that of non-driver DNA repair genes ( q < 0 . 01 , S7 Table ) in all of the 9 cancer types ( Fig 3B ) ., For BRCA , the DNA repair cancer driver gene aveG score ( 0 . 73 ± 0 . 13 ) is marginally higher than that of non-DNA repair cancer driver genes ( 0 . 46 ± 0 . 02 , q = 0 . 12 ) , while significantly higher than that of non-driver DNA repair genes ( 0 . 24 ± 0 . 03 , q = 4 . 4 × 10−4 , S7 Table ) ., Furthermore , the non-DNA repair cancer driver gene aveG score is significantly higher than that of non-driver DNA repair genes in all of the 9 cancer types as well ( q < 0 . 01 , Fig 3B and S7 Table ) ., We further divided CGC genes into two subgroups: DNA repair CGC genes and non-DNA repair CGC genes ., We found CGC genes are significantly enriched in DNA repair genes as well ( p = 2 . 7 × 10−18 , Fisher’s exact test , S16B Fig ) ., S8 Table indicated that DNA repair CGC gene aveG score is not significantly higher than that in both non-DNA repair CGC genes ( q > 0 . 50 ) and non-CGC DNA repair genes ( q > 0 . 10 ) in 8 cancer types with an exception of OV ( q = 0 . 03 ) ., Moreover , the non-DNA repair CGC gene aveG score is higher than that of non-CGC DNA repair genes in COAD ( q = 0 . 04 ) and OV ( q = 0 . 02 , S8 Table ) ., Collectively , the cancer genome evolution shaped by cancer driver genes may have additional mechanisms ( i . e . , chromatin regulation ) , except DNA repair ., We found that the top 100 genes with the highest aveG scores tend to be cancer driver genes ( q < 0 . 01 , Fisher’s exact test , Fig 4A and S9 Table ) or CGC genes ( q < 0 . 05 , S10 Table ) in all of the 9 cancer types ., In addition , the top 100 genes with the highest aveG scores are more likely to be CRFs ( q < 0 . 05 , S11 Table ) in 7 cancer types with the exception of COAD ( q = 0 . 12 ) and LUSC ( q = 0 . 12 ) ., However , the top 100 genes are not significantly enriched in DNA repair genes in all of the 9 cancer types ( q > 0 . 05 , Fig 4A and S12 Table ) ., We further examined the tumor exome mutation density ( the average number of mutations per Mb ) for the top 10 genes with the highest aveG score via the genome-wide mutation rate analysis ( S13 Table ) ., By examining mutation density data of ~3 , 000 tumor exomes from Kandoth et al . 29 , we found that patients having nonsynonymous somatic mutations on any of four genes ( FAT4 , SYNE1 , AHNAK , or COL11A1 ) often showed a higher cancer genome mutation density at the whole genome level compared to that of wild-type ( WT ) patients in 4 cancer types: COAD , LUAD , LUSC , and UCEC ( Fig 4B ) ., FAT4 ( protocadherin fat 4 ) , a member of the cadherin super-family , is a key component in the Hippo signaling pathway , playing a candidate tumor suppressor role in cancer 40 ., In COAD , 40 patients harbored FAT4 nonsynonymous mutations ., The average number of mutations per Mb for 40 FAT4 mutated COAD samples ( 43 . 3 ± 12 . 8 ) are significantly higher than that of FAT4 WT samples ( 5 . 0 ± 0 . 57 , q = 1 . 1 × 10−5 , Fig 4B ) ., Similarly , the average number of mutations per Mb for 43 FAT4 mutated LUAD samples ( 26 . 3 ± 8 . 4 ) are significantly higher than that of FAT4 WT samples ( 7 . 5 ± 0 . 48 , q = 2 . 8 × 10−9 , Fig 4B ) ., Using genome-wide association studies , Berndt et al . found FAT4 to be a candidate gene for spontaneous pulmonary adenomas 41 ., Using exome sequencing , Zang et al . found that the somatic inactivation of FAT4 might be a critical tumorigenic event in a subset of gastric cancers 42 ., In this study , FAT4 was identified as a putative cancer gene involved in lung and colorectal cancer , which is consistent with previous studies 40–43 ., SYNE1 , encoding spectrin repeat containing , nuclear envelope 1 , is involved in nuclear organization and structural integrity , function of the Golgi apparatus , and cytokinesis ., Herein , we found that the average number of mutations per Mb for 49 SYNE1 mutated COAD samples ( 35 . 8 ± 8 . 4 ) are significantly higher than that of SYNE1 WT samples ( 7 . 5 ± 0 . 48 , q = 6 . 8 × 10−9 , Fig 5B ) ., Doherty et al . found that SYNE1 polymorphism relates to an increased risk of invasive ovarian cancer 44 ., Collectively , SYNE1 may be a candidate cancer mutated gene in COAD ., AHNAK ( neuroblast differentiation-associated protein ) , also known as desmoyokin , is essential for tumor cell migration and invasion 45 ., In this study , the average number of mutations per Mb ( 12 . 1 ± 2 . 6 ) for 22 AHNAK mutated samples is significantly higher than that of AHNAK WT samples in HNSC ( 4 . 5 ± 0 . 21 , q = 1 . 5 × 10−5 , Fig 4B ) ., Dumitru et al . found that AHNAK was associated with poor survival rates in laryngeal carcinoma , a major subtype of head and neck cancer 46 ., COL11A1 and COL6A3 , encoding collagen proteins , are two main structural proteins of the various connective tissues in animals ., In LUAD , the average number of mutations per Mb ( 25 . 3 ± 7 . 9 ) for 46 COL11A1 mutated samples is significantly higher than that of COL11A1 WT samples ( 7 . 4 ± 0 . 47 , q = 1 . 1 × 10−9 , Fig 5B ) ., Additionally , for LUSC , the average number of mutations per Mb ( 16 . 5 ± 0 . 59 ) for 32 COL11A1 mutated samples is significantly higher than that of COL11A1 WT samples as well ( 8 . 5 ± 0 . 40 , q = 4 . 9 × 10−5 ) ., Furthermore , COL6A3 ( q = 3 . 1 × 10−4 , COAD ) and COL5A2 ( q = 1 . 5 × 10−4 , LUAD ) mutations are significantly associated with a high mutation density in colorectal and lung cancer , respectively ., The over-expression of COL11A1 reportedly correlates with lymph node metastasis and poor prognosis in non-small cell lung cancer and ovarian cancer 47–49 ., The expression level of COL6A3 is involved in pancreatic malignancy 50 , 51 ., Collectively , AHNAK , COL11A1 , and COL6A3 may be potential candidates for therapeutic and diagnostic biomarkers in head and neck cancer and lung carcinoma ., However , the mutation status of each of aforementioned genes is associated with the genome-wide mutation rate ., Mutations in these genes could be either the cause of the mutation-rate increase or simply a consequence of an elevated global mutation rate ., Thus , further experimental validation of these genes in the specific cancer type is warranted ., When examining cancer driver gene aveG score across chromosomes in each of 9 cancer types , interestingly , we found that the X chromosome has an unusually higher cancer driver gene aveG scores compared to autosomes in BRCA , GBM , and UCEC using the total 22 autosomes as background ( Fig 5 ) ., In BRCA , cancer driver gene aveG score ( 0 . 66 ± 0 . 09 ) on the X chromosome is higher than that of the whole set of 22 autosomes ( 0 . 46 ± 0 . 02 , q = 0 . 06 p = 7 . 9 × 10−3 , Wilcoxon rank-sum test , Fig 5A ) ., Similarly , in GBM , the cancer driver gene aveG score ( 1 . 2 ± 0 . 18 ) on the X chromosome is higher than that of the whole set of 22 autosomes ( 0 . 80 ± 0 . 05 , q = 0 . 07 p = 9 . 9 × 10−3 , Fig 5B ) ., And the cancer driver gene aveG score ( 0 . 92 ± 0 . 15 ) on the X chromosome is also higher than that of the whole set of 22 autosomes in UCEC ( 0 . 56 ± 0 . 03 , q = 0 . 04 p = 5 . 3 × 10−3 , Fig 5C ) ., As a control , we repeated the aforementioned analyses for all genes and essential genes , respectively ., We did not find the higher aveG score on the X chromosome for all genes or essential genes in any of the 9 cancer types ( Fig 5 and S17 Fig ) ., Thus , the high gene aveG score on the X chromosome is unique for cancer driver genes ., The X chromosome is largely functionally haploid in both males and females ., A recent study showed that hypermutation of the inactive X chromosome is a frequent event in cancer 52 ., Both BRCA and UCEC ( Fig 5 ) are female-specific cancer , while GBM is not ., To explore the hypermutation of inactive X chromosome in the female versus male cancer genomes , we separated GBM patients as males and females , and performed the same analysis ., Interestingly , we found that the cancer driver gene aveG score ( 0 . 66 ± 0 . 13 ) on the X chromosome is significantly higher than that of the whole set of 22 autosomes ( 0 . 43 ± 0 . 04 , q = 0 . 04 , Fig 6A and 6C ) in the female GBM genomes ., However , the cancer driver gene aveG score ( 0 . 68 ± 0 . 17 ) on the X chromosome is similar to that of the whole set of 22 autosomes ( 0 . 72 ± 0 . 07 , q = 0 . 68 , Fig 6B and 6C ) in the male GBM genomes ., Furthermore , similar aveG scores for all genes ( q = 0 . 09 ) or essential genes ( q = 0 . 18 ) were observed between the X chromosome and the whole set of 22 autosomes in the female GBM genomes ., In contrast , we found a lower aveG score on the X chromosome for all genes ( q = 4 . 4 × 10−9 , Fig 6C ) or essential genes ( q = 0 . 06 ) compared to that on the whole set of 22 autosomes in the male GBM genomes ., We then examined the top 10 driver genes with the highest aveG scores on the X chromosome in BRCA , GBM , and UCEC ., Two putative cancer drivers ( DDX3X and STAG2 ) stood out ( Fig 6D and 6E ) ., We found that the patients harboring DDX3X or STAG2 nonsynonymous mutations have a higher genome mutation density in uterine cancer during the genome-wide mutation rate analysis ( Fig 6D ) ., For instance , the average number of mutations per Mb for 15 DDX3X mutated uterine tumors is 144 . 1 ± 34 . 0 , 11-fold higher than that of DDX3X WT tumors ( 13 . 1 ± 2 . 8 , q = 2 . 5 × 10−5 ) ., A previous study indicated that somatic mutations of DDX3X were associated with medulloblastoma 53 ., Additionally , the average number of mutations per Mb for 26 STAG2 mutated uterine tumors ( 144 . 5 ± 26 . 0 ) is significantly higher than that for STAG2 WT samples ( 10 . 2 ± 2 . 2 , q = 1 . 9 × 10−10 ) ., STAG2 belongs to cohesin protein family , playing an important role in mediating sister chromatid cohesion 54 ., Solomon et al . found that the inactivation of STAG2 causes aneuploidy in human glioblastoma cell lines 55 ., Lawrence et al . recently identified STAG2 as one of the 12 genes that were mutated at a substantially high frequency in at least four cancer types through examining the exome sequencing data of 4 , 742 human cancer samples across 21 cancer types 30 ., Taken together , we provided statistical evidence in that hypermutation of the cancer driver genes on the inactive X chromosome may be a general feature in the female cancer genomes 52 ., Further investigation on this feature is warranted ., Several previous studies showed several lines of strong biological evidences in that a single gene may shape subsequent evolution of the human cancer genome 8 , 11 , 13 ., Such evidence motivated us to develop a mathematical model that can quantitatively measure a gene-gene pair to be co-mutated and highly co-expressed simultaneously in a given cancer type ., Here , we proposed the gene gravity model based on Newton’s law of gravitation to study the cancer genome evolution by the systematic integration of ~3 , 000 cancer genome transcription and somatic mutation profiles from TCGA under molecular network architecture knowledge ., It is worth noting that some factors , such as gene length , network topology ( e . g . connectivity ) , high mutation rate on the cancer driver genes , and high PCC value for the particular genes , may affect the performance of the gene gravity model ., Longer genes would be more likely to harbor mutations , increasing the false positive rate during cancer genomic analysis 28 , 32 ., We investigated the correlation of the gene aveG score with gene cDNA length collected from Tamborero et al . 56 ., We removed two longest human genes ( TTN and MUC16 ) because no evidence has been found in cancer yet 28 , 32 ., We observed a moderate correlation between gene aveG score and cDNA length in the 9 cancer types ( S18 Fig ) ., For BRCA , the correlation is 0 . 21 between gene aveG score and gene cDNA length ( p < 2 . 2 × 10−16 ) ., In addition , we recalculated the aveG score by using the average mutation density ( M/L , here M is the number of mutations for a given gene in a specific cancer type ) per base pair in each cancer type normalized by gene cDNA length ( L ) ., We could reproduce the results ( S19 Fig ) , since the new results are nearly the same to those presented in S2–S10 Figs ., We next examined whether the gene connectivity and gene average co-expression correlation , such as “party hub” in the network 57 , contribute to the performance of the gene gravity model ., We found that gene aveG score significantly correlates with gene connectivity in all of the 9 cancer types ( S20 Fig ) ., For BRCA , the correlation is 0 . 40 between the gene aveG score and gene connectivity in PIN ( p < 2 . 2 × 10−16 , F-statistics , S20 Fig ) ., Thus , a gene with high connectivity may create a higher cancer genome evolution rate ., Additionally , we investigated the relationship between the gene aveG and the average gene co-expression coefficient ( avePCC ) ., We calculated a gene avePCC using ( ρ ) i = ∑j PCCij / n between gene i and gene j ( j belongs to the set of gene i’s interacting partners ( n ) in PIN ) based on the absolute value of PCC for each gene-gene pair ., We found a moderately positive correlation between gene aveG score and its avePCC across 9 cancer types ( p < 2 . 2 × 10−16 , S21 Fig ) ., Finally , we further examined whether we could reproduce the results using 4 features: high connectivity , high avePCC , gene length , and high mutation rate ., For comparison , we separated genes into 3 categories based on the range of the aveG score ., As shown in S22 Fig , for each of these 4 features , the distribution of aveG score cannot simply separate 3 different aveG categories: low , middle , and high groups ., In a previous study , we found a positive correlation of protein connectivity with the number of nonsynonymous somatic mutations across 12 cancer types 23 ., Thus , the current observation is consistent with our previous study that network-attacking perturbations due to somatic mutations occurring in the network hubs of the cancer interactome play important roles during tumor emergence and evolution 23 ., There are some ultra-mutated tumor samples in various cancer types , such as UCEC or COAD ., For example , a small number of tumor samples can contribute to a large proportion ( e . g . , 40% ) of total somatic mutations observed in the whole cancer cohort 29 ., We removed 18 ultra-mutated tumor samples in UCEC and 31 ultra-mutated tumor samples in COAD based on a previous study 29

Introduction, Results, Discussion, Materials and Methods

Cancer development and progression result from somatic evolution by an accumulation of genomic alterations ., The effects of those alterations on the fitness of somatic cells lead to evolutionary adaptations such as increased cell proliferation , angiogenesis , and altered anticancer drug responses ., However , there are few general mathematical models to quantitatively examine how perturbations of a single gene shape subsequent evolution of the cancer genome ., In this study , we proposed the gene gravity model to study the evolution of cancer genomes by incorporating the genome-wide transcription and somatic mutation profiles of ~3 , 000 tumors across 9 cancer types from The Cancer Genome Atlas into a broad gene network ., We found that somatic mutations of a cancer driver gene may drive cancer genome evolution by inducing mutations in other genes ., This functional consequence is often generated by the combined effect of genetic and epigenetic ( e . g . , chromatin regulation ) alterations ., By quantifying cancer genome evolution using the gene gravity model , we identified six putative cancer genes ( AHNAK , COL11A1 , DDX3X , FAT4 , STAG2 , and SYNE1 ) ., The tumor genomes harboring the nonsynonymous somatic mutations in these genes had a higher mutation density at the genome level compared to the wild-type groups ., Furthermore , we provided statistical evidence that hypermutation of cancer driver genes on inactive X chromosomes is a general feature in female cancer genomes ., In summary , this study sheds light on the functional consequences and evolutionary characteristics of somatic mutations during tumorigenesis by propelling adaptive cancer genome evolution , which would provide new perspectives for cancer research and therapeutics .

Cancer genome instabilities , such as chromosomal instability and microsatellite instability , have been recognized as a hallmark of cancer for several decades ., However , distinguishing cancer functional somatic mutations from massive passenger mutations and non-genetic events is a major challenge in cancer research ., Massive genomic alterations present researchers with a dilemma: does this somatic genome evolution contribute to cancer , or is it simply a byproduct of cellular processes gone awry ?, In this study , we developed a new mathematical model to incorporate the genome-wide transcription and somatic mutation profiles of ~3 , 000 tumors across 9 cancer types from The Cancer Genome Atlas into a broad gene network ., We found that cancer driver genes may shape somatic genome evolution by inducing mutations in other genes in cancer ., This functional consequence is often generated by the combined effect of genetic and epigenetic alterations ( e . g . chromatin regulation ) ., Moreover , we provided statistical evidence that hypermutation of cancer driver genes on inactive X chromosomes is a general feature in female cancer genomes and found a putative X-inactive specific gene STAG2 in uterine cancer ., In summary , this work illustrates the functional consequences and evolutionary characteristics of somatic mutations during tumorigenesis through driving adaptive cancer genome evolution .

journal.pcbi.1000915

Occurrence and Treatment of Bone Atrophic Non-Unions Investigated by an Integrative Approach

Atrophic non-unions , a class of non-healing fractures that display only limited external callus formation , were thought to occur as a result of impaired local blood supply 1 , 2 ., However , studies of human atrophic non-unions have shown that the gap tissues can be well vascularized 3–5 ., Recently several animal models were developed to investigate the etiology of atrophic non-unions 6–9 ., In these animal models , the periosteum disruption and reaming of the marrow canal is combined with adequate stabilization of the osteotomy site ., All models demonstrate established non-unions ., These animal models are a good representation of the clinical situation in which most atrophic non-unions develop 8 ., Previous non-union models often utilized large segmental bone defects 10–13 where non-unions developed due to the size of the defect rather than the altered biology of the fracture site 8 ., Based on previous , purely experimental , studies with these atrophic non-union models , it was hypothesized that in order to obtain successful fracture healing , blood vessels , growth factors and ( proliferative ) precursor cells all need to be present in the callus at the same time 6 , 7 ., In this study we will use a combined in vivo-in silico approach to look at these different aspects ( vasculature , growth factors , cell proliferation ) ., Furthermore , the in silico model is used to investigate the occurrence of and design possible treatment strategies for atrophic non-unions ., Experimental results ( original work and previously published ) are used to corroborate the numerical predictions for atrophic non-unions in particular and demonstrate the potential of both the in silico-in vivo approach and the treatment strategy of cell transplantation ., Finally , the mathematical model is applied to explain experimental observations and identify potentially crucial steps in the treatments ., Rats were kept in accordance with UK Home Office welfare guidelines and project license restrictions ., Animals and operative procedures were carried out as previously reported 7 ., In brief , 28 adult female Wistar rats were randomised into 2 groups of 14 ( ‘non-union’ and ‘healing’ ) and were sacrificed at 1 ( n\u200a=\u200a3 ) , 3 ( n\u200a=\u200a3 ) , 8 ( n\u200a=\u200a4 ) and 16 ( n\u200a=\u200a4 ) weeks and the right ( operated ) tibia was prepared for histological examination ., The animals were caged individually and allowed water and food ad libitum and unrestricted weight-bearing ., A standardised circular frame external fixator 7 was applied to the right tibia under general anaesthesia and with aseptic conditions ., An osteotomy was performed using a 1mm burr under constant irrigation with cold saline solution ., The fibula was fractured manually using a three-point bending method and a 1 mm gap introduced at the site of the osteotomy ., In 14 of the 28 animals the periosteum was stripped and the intramedullary canal was curetted , both proximally and distally , for a distance equivalent to 1 tibial diameter ., The wound was washed thoroughly and the skin was closed ., Two independent senior orthopaedic trainees assessed standardized radiographs obtained after operation and every two weeks thereafter ., They categorized the fractures as healing or not according to the criteria of the AO-ASIF manual 2 ., Atrophic non-union was induced in 8 WKY rats as previously described 7 ., Three weeks after operation , 4 rats received a 100 µl injection at the non-union site of cultured bone marrow cells and 4 rats received a 100 µl injection of carrier solution alone ., Bone marrow cells were obtained by aspiration of the hind limbs of WKY rats and processed for injection as described in ( a full description of this protocol is submitted for publication elsewhere ) ., Two oblique radiographs were taken of the right ( operated ) tibia post-operatively and at 2 weeks , 3 weeks ( immediately following cell/carrier injection ) and every week thereafter ., Radiographs were examined by two independent senior orthopaedic trainees ., Each animal was categorized as healing or not ( full report of results submitted for publication elsewhere ) ., Formation of callus was assessed by scanning radiographs into a Macintosh Quadra 650 computer and analyzing images ( Optilab Pro v2 . 5 , Graftek , France ) ., The callus outline was traced manually and the size of the outlined area was calculated ., The results were expressed as a percentage change in the amount of mineralized tissue from post-operative radiographs 7 ., The right lower limbs were fixed in neutral phosphate-buffered formalin ( 4% v/v ) for 48 hours , decalcified in neutral ethylenediaminotetra-acetic acid ( EDTA ) , embedded in paraffin wax and 6µm sections were cut and stained with haematoxylin and eosin ., Wax sections for immunohistochemistry were cut onto poly-l-lysine coated slides and immunostaining was performed with the following antibodies: Transforming growth factor beta ( TGF-β ) ( mouse monoclonal , AbD Serotec Ltd , Oxford , UK ) , Basic Fibroblast Growth Factor ( FGF basic ) ( goat polyclonal , Santa Cruz Biotechnology Inc . USA ) , Platelet-Derived Growth Factor ( PDGF ) ( goat polyclonal , R&D Systems Ltd , Abingdon , UK ) , Bone Morphogenetic Proteins 2 and 4 ( BMP 2/4 ) ( goat polyclonal , R&D Systems Ltd , Abingdon , UK ) , and Proliferating Cell Nuclear Antigen ( PCNA ) ( mouse monoclonal , Dako Ltd , Ely , UK ) ., Alkaline phosphatase-conjugated anti-mouse ( Dako , Ely , UK ) or anti-goat ( Sigma , Poole , UK ) secondary antibodies were used ., Cell transplant samples labelled with BrdU were prepared as above , followed by immunostaining with antibodies against BrdU ( mouse monoclonal , Dako Ltd , Ely , UK ) in conjunction with an animal research kit ( ARK , Dako Ltd , Ely , UK ) ., All antibodies were used according to the manufacturers instructions ., Sections were analyzed using a light microscope ., Cytoplasmic growth factor expression was semi-quantified using a four-value intensity score ( 0 , 1+ , 2+ , and 3+ ) 14 ., For proliferating cells , the numbers of positively and negatively stained cells were counted in randomly selected fields within the interfragmentary gap and the median positive staining index was calculated ., A Mann-Whitney U test was used to assess significance of the cell proliferation results ., All statistical analyses were performed using the Statview software package ( SAS Institute Inc . , USA ) and significance was assumed as p<0 . 05 ., The mathematical model used in this study was originally developed to describe normal fracture healing 15 ., It expresses the change of a number of continuum-type variables – growth factor concentrations , cell densities and matrix densities – as a function of time and spatial coordinates and is schematically represented in Figure 1 ., The model accounts for various key processes of bone regeneration ., Starting with a callus filled with granulation tissue , mesenchymal stem cells and growth factors quickly occupy the regeneration zone ., This is followed by mesenchymal stem cell differentiation into osteoblasts ( intramembranous ossification – close to the cortex away from the fracture site ) and chondrocytes ( central callus region ) ., Subsequently , endochondral ossification can take place during which VEGF , expressed by ( hypertrophic ) chondrocytes , attracts blood vessels and osteoblasts , resulting in cartilage degradation and bone formation ., Bone remodelling processes are not included in the model ., The effect of mechanical loading can also be incorporated , by making various biological processes dependent on local mechanical stimuli 16 ., Mechanical influences will however not be the subject of the current study ., The regeneration processes are described by calculating the spatiotemporal evolution of the density of mesenchymal stem cells , osteoblasts , chondrocytes , fibroblasts , endothelial cells , bone , cartilage , fibrous tissue and vascular matrix and the concentrations of three generic growth factor families ( osteogenic , chondrogenic and vascular growth factors ) ., The spatiotemporal dynamics is expressed by means of a system of 12 partial differential equations of the taxis-diffusion-reaction type ., The systems general structure is: ( 1 ) t represents the time , the space and the non-dimensional density of a migrating cell type ( mesenchymal stem cells , fibroblasts and endothelial cells ) ., represents the vector of the other nine non-dimensional cell concentrations , ECM densities and growth factor concentrations ( for which no directed migration is modelled ) ., and D ( non-negative diagonal matrix ) are the diffusion coefficients ( random motion ) ., represents the taxis coefficients for both chemotaxis ( movement up a gradient of growth factor concentration ) and haptotaxis ( movement up a gradient of matrix density ) ., and are reaction terms describing cell differentiation , proliferation and decay and matrix and growth factor production and decay ., The system ( 1 ) must be complemented by suitable initial and boundary conditions to ensure the existence , uniqueness and non-negativity of a solution ., The model equations are implemented in a customized finite volume code 17 ., Additional information , including the complete set of equations , boundary and initial conditions , parameter values and implementation details , is provided in the Supplementary Methods ( Text S1 ) and Geris et al . 15 ., For a detailed discussion of the models underlying assumptions , simplifications and shortcomings , we refer the reader to Geris et al . 15 , 16 ., Non-union cases were previously simulated by compromising the initial blood vessel or growth factor supply 15 or by applying increased mechanical loading ( instability ) on the fracture 16 ., In all cases , the simulations were able to capture observed experimental and clinical outcomes ., Moreover , in silico experiments were conducted to design potential treatment strategies for the various non-union models 15 , 18 , 19 however , these predictions were not corroborated by dedicated experiments as is the case in this study ., A simplified ( fixed ) geometrical domain of a fracture callus ( Fig . 2A , B ) was constructed based on the experimental set-up 7 ., For the atrophic non-union case the domain was extended at the distal end ( away from the fracture site ) over the distance that the periosteum was stripped and the marrow canal was reamed in the experiments ., The current implementation of the model assumes a constant callus size for both healing and non-union groups ( Fig . 2B ) ., The experimentally observed callus size for both healing and non-union groups is described in 7 ., For the non-union group there was a trend towards a decrease of the callus-size ( by 10% ) over time , though this was not a statistically significant decrease ., It was shown on a rabbit atrophic non-union model 6 that the ( lack of ) granulation tissue containing the necessary MSCs plays an important role in the formation of a non-union ., Therefore , in order to simulate the formation of an atrophic non-union , the boundary condition for the MSCs was adapted ( value of the Dirichlet boundary conditions was decreased with a factor 105 ) while all other model parameters and initial/boundary conditions were left unchanged with respect to the normal healing case ., As the initial mechanical conditions for both the healing and non-union group were equal , the influence of mechanical loading was not taken into account explicitly in this study ( the bioregulatory model presented in 15 was used rather than an extended mechanobioregulatory model presented in 16 ) ., After initial analyses , the values for the growth factor boundary conditions ( which were estimated in 15 ) were lowered by a factor 10 in this study ( for both the healing and non-union case ) to obtain a better correspondence with experimental results ( for both healing and non-union case ) ., This alteration only affected the average growth factor concentration during the first weeks of the healing process but had no effect on the amount and distribution of the cells and extracellular matrix ., For all simulations cartilage and bone matrix formation rates were equal to the values reported in 15 , apart from a case in which these values were lowered by a factor 10 ., This was done in order to investigate the effect on the healing rate ., For all the simulated treatment strategies , 1ml of MSCs was administered at a concentration of 106 cells/ml ., Tissue fractions were calculated from the bone , cartilage and fibrous tissue matrix densities in the central area of the domain , corresponding to the experimentally investigated region of interest ( i . e . excluding those parts of the domain stretching out to the right alongside the cortex , see insert in Fig . 2D ) ., The simulation results showed healing progression similar to experimental results 7 , for both the healing and the non-union group ( Fig . 2C ) ., For the non-union group , the mathematical model predicted the formation of small amounts of cartilage and bone by post-osteotomy week ( POW ) 8 ., For the healing group , both direct bone formation ( close to the undamaged cortical bone ) and cartilage formation ( central part of the callus ) were predicted to form by POW3 ( Fig . 2D ) ., By week 8 this cartilage was replaced by bone via endochondral ossification in the simulations ., Reducing the cartilage and/or bone matrix formation rate in the mathematical model resulted in a slower ossification process ( Fig . 2E ) ., As blood vessels are represented in the mathematical model by a continuum variable , the amount of vessels itself cannot be quantified from the simulation results ., Instead , the percentage of vascularized tissue was calculated ., Due to the absence of the vasculogenesis process ( i . e . process of blood vessel formation occurring by a de novo production of endothelial cells in contrast to angiogenesis where blood vessels are formed from pre-existing ones 20 ) in the mathematical model , blood vessel formation only appeared at the onset of osteogenesis ( week 1 ) ( Fig . 3A , B ) ., By 8 weeks , the non-union group reached the level of vascularisation that was present in the healing group at 3 weeks ( Fig . 3B ) ., In the experimental non-union group , cell proliferation was at its highest one-week post-osteotomy , with PCNA positive cells present throughout the interfragmentary gap ., The number of PCNA positive cells within the interfragmentary gap diminished at 3 and 8 weeks post-osteotomy ( Fig . 3C , E ) ., In the healing group , cell proliferation peaked at week 3 , where PCNA positive cells were present in the periosteum and at the edge of the ossification front ( Fig . 3E ) ., By 8 weeks the number of PCNA positive cells had diminished , and they were only evident in the periosteum at the periphery of the bridging callus ., In the numerical simulations , rapid proliferation of the MSCs at the onset of the healing process spiked the value for cellular proliferation in the first week for the healing group in contrast to the non-union group ( Fig . 3D ) ., After differentiation into chondrocytes in the central part of the callus and subsequently into osteoblasts , the cellular proliferation dropped in the healing group ., Fibroblasts are the predominant cell type in the non-union group and as the predicted fibrous tissue density was not as high as that of bone or cartilage ( leaving less space for cell proliferation ) for the healing group , the non-union group was predicted to have a higher proliferative capacity at POW 8 compared to the healing group , corresponding to experimental in vivo observations ., At POW 1 , the hematoma within the interfragmentary gap of the in vivo experimental healing and non-union groups stained positively for TGF-β , FGF-b , PDGF and BMP 2/4 , however , TGF-β and FGF-b staining of the hematoma in the non-union group appeared weaker than that of the healing group ( Fig . 4A and Figure S1 ) ., At POW 3 , staining of all four growth factors was evident in areas of endochondral ossification in the healing group , where both osteoblasts and chondrocytes were expressing these growth factors ., In the non-union group , however , TGF-β , FGF-b and BMP 2/4 staining had diminished in comparison to the one week time point ., At 8 and 16 weeks ( identical results were obtained for both time points ) , there was either weak or absent staining of all four growth factors in the healing group , due to bridging callus ., In the non-union group , weak staining of all 4 growth factors remained in the fibrous tissue of the interfragmentary gap ., In the mathematical model , generic , functional families rather than specific growth factors were implemented ., The average concentration for each generic growth factor group was calculated for the central area of the domain ( indicated on insert in Fig . 2D ) ., In Figure 4B , the experimentally measured growth factors for the healing group are depicted according to their classification in generic growth factor families by Cho et al . 21 , Pepper et al . 22 and Lienau et al . 23 ., Osteogenic growth factors are expressed early on in the healing process during intramembranous ossification ., Later on , an increase in their production is predicted during the endochondral ossification process ., For the vascular growth factors , after the initial decrease , upregulation is predicted during the endochondral ossification process taking place in the healing group , where VEGF is being expressed by ( hypertrophic ) chondrocytes ., For both the healing and non-union groups , the highest levels of chondrogenic growth factors are predicted by the mathematical model in the first week , similar to the experimental measurements ., For the non-union group ( Fig . 4C ) , after the initial growth factor release at fracture induction , some chondrogenic growth factor production remains present up to and after POW 8 ., Osteogenic growth factor production is predicted to rise between POW3 and POW8 in the non-union group , as experimentally observed ., After corroboration of the mathematical model , we wanted to test , both in silico and in vivo , the hypothesis that the onset of atrophic non-union could be prevented by the injection of cultured MSCs at three weeks post-osteotomy , i . e . when vascularity within the interfragmentary gap was sufficient to keep the injected cells alive ., In silico , after the injection of the cell transplant at POW 3 in the callus region , the amount of bone was predicted to gradually increase whereas the amount of fibrous tissue was predicted to decrease up to POW 16 ., The formation of a small amount of cartilage was also predicted with endochondral ossification still in progress at POW 16 ( Fig . 5Ai , Aii ) ., The amount of soft tissue present at POW 16 was strongly dependent on the exact location of injection of the cell transplant with excentral injection leading to unicortical bridging ( Fig . 5Bi , Bii ) ., A technique often used to administer growth factors to healing fractures is the administration of growth factors inside an injectable carrier close to but outside of the callus ( reviewed in 24 ) ., Alternatively , one could adopt such a carrier approach for cell delivery as well ., However , simulating such a treatment strategy predicted the formation of a layer of bone closest to the cell source , preventing other cells from further penetrating the callus ( Fig . 5c ) ., In the simulations , a combination of periosteal stripping and marrow canal curettage led to non-union formation , thereby predicting the observed in vivo outcome , where all animals that had periosteal stripping and curettage of the intramedullary canal went on to form an atrophic non-union at 8 and 16 weeks post-osteotomy and all animals where this was not performed went on to unite successfully 7 ., At 1 week there was no significant difference in tissue constituents between the two experimental groups ., However , from 3 weeks onwards , there was a significant increase in bone formation in the healing group when compared to the non-union group where the interfragmentary gap consisted predominantly of fibrous tissue ., The model did not predict the experimentally observed rounding of the cortical bone ends in the non-union group , nor the capping of the intramedullary canal ., The cortical bone falls outside of the modelling domain so changes in the cortical bone cannot be predicted by the model ., The mechanism which causes the capping of the intramedullary canal is not known and does not seem to be encompassed by the present model equations either ., Although the order of fracture healing events in the simulations corresponds to experimental observations , the normal healing enrolled faster in the simulations than in the experiments 7 ., As the predicted time frame of healing as well as the predicted tissue pattern correspond well to other rat fracture models 8 , 25 , we speculate that this difference might be attributed to a number of factors that are specific for this experimental set-up ., The initial delay between experimental observations and numerical simulations is one week ( Fig . 2D ) ., This could be due to a particularly lengthy inflammatory phase in this specific experimental set-up , a phase that is not incorporated explicitly in the mathematical model ., After POW 3 , the endochondral ossification process enrolls much faster in silico than in vivo increasing the time delay between experiments and simulations ., Decreasing the endochondral ossification speed in silico by e . g . reducing the cartilage and bone matrix production rate in the model leads to a better correspondence between experiments and simulations ( Fig . 2E ) ., For all other simulation results shown in this study the original values for cartilage and bone matrix production rates as determined in 15 were used ., Similar to the experimental observations 7 , by 8 weeks , the non-union group reached the level of vascularisation that was present in the healing group at 3 weeks ( Fig . 3B ) ., The slower vascularisation of the simulated non-union with respect to its experimental counterpart could be due to the absence of de novo blood vessel formation ( vasculogenesis ) in the mathematical model combined with the absence of vascular growth factor production by fibroblasts , leading to a slow build-up of the vascular network ., The experimentally observed decrease in number of blood vessels during bone/blood vessel remodelling at POW 8 ( which is often accompanied by an increase in vessel diameter , a parameter that is not included in the experimental measurements ) cannot be predicted as the model does not encompass the remodelling process ., Despite these limitations , the mathematical model does predict a substantial time difference in the vascularisation of the callus area as shown by the ( also experimentally observed ) lag in blood vessel formation between healing and non-union groups at POW 3 ., Cell proliferation observed in the in vivo healing group , concurred well with that observed by Iwaki et al . 26 where proliferating cells peaked between 1 and 3 weeks after fracture ., For the proliferation , no exact match to the experimental measurement could be obtained from the simulation results ., A general cellular proliferation value was calculated by multiplying the cells in the callus by their respective proliferation rates ( which are fixed parameters in the mathematical model ) , normalized to total cell amount and the maximal proliferation rate ., As all cells in the mathematical model are able to proliferate , providing there is sufficient space surrounding them ( i . e . ECM density is sufficiently low ) , this calculated value is merely a theoretical one ., The initial low density fibrous matrix present at the start of the healing simulations allows for rapid proliferation of the MSCs , spiking the value for cellular proliferation in the first week for the healing group in contrast to the non-union group where very few MSCs were present ( Fig . 3D ) ., The formation of differentiated tissue types such as bone and cartilage constrained the proliferation of the cells later on in the regeneration process for the healing group ., As the fibrous matrix that develops in the non-union group does not reach the density of the bone or cartilage in the healing group , the , mainly fibroblastic , cells in the former group had a higher proliferative capacity at POW 8 compared to the latter , as observed in vivo ., Growth factor expression observed experimentally during the early stages of bone healing correlated well with other studies reporting BMP 27 , TGF-β 28 , and FGFb 28 , 29 expression in normally healing rat fracture models , and PDGF expression in normal and impaired human fracture healing 30 ., Furthermore , these results correlate with those of Brownlow et al . 6 who noted in a rabbit atrophic non-union model that by POW 8 , there was little or no expression of TGF-β , FGFb , PDGF or BMP 2/4 ., Cho et al . 21 , Pepper et al . 22 and Lienau et al 23 classified growth factors in functional families , similar to those used in the mathematical model ., The reported evolution over time of these growth factor families corresponds well to the model predictions for the healing group ( Fig . 4B ) ., For the vascular growth factors , after the initial decrease , upregulation is predicted during the endochondral ossification process taking place in the healing group , where VEGF is being expressed by ( hypertrophic ) chondrocytes , corresponding well to previously reported experimental observations 31 ., The experimentally observed decrease in osteogenic growth factors after week 3 is not present in the model as the osteoblasts remain active ( i . e . keep producing growth factors ) ., Addition of a supplementary variable representing the matured osteoblasts ( osteocytes ) could resolve this discrepancy between experiments and simulations ., For the non-union group , the osteogenic growth factor levels are predicted to rise between POW3 and POW8 , corresponding to experimental in vivo observations ., Furthermore , the simulations show occurrence of the highest levels of chondrogenic and vascular growth factors in the first week , as experimentally observed ., In the mathematical model chondrocytes are the only cell type capable of chondrogenic growth factor production and the main responsible for the production of vascular growth factor ( upon hypertrophy ) ., As chondrocytes are predicted to be only marginally present in the callus of the non-union group , the predicted chondrogenic and vascular growth factor concentrations further on in the healing process are very low for this group in comparison to the growth factor release at fracture induction ., Furthermore , for the non-union group , initial growth factor concentrations are strongly driven by boundary conditions that were applied during the first days after fracture to mimic local and systemic reactions occurring outside of the modelling domain 32 , 33 ., Comparison of experimental and simulation results showed a number of discrepancies that could be attributed to a number of model simplifications and suggestions were made above to overcome those ., Additional model simplifications were made in the framework of this study such as the use of a fixed callus size ( rather than a size that varies over time ) and the continuum representation of blood vessel formation , which will be dealt with in future versions of the mathematical model ., For a thorough discussion on the model limitations we refer the reader to 15 ., In this study , mechanical loading was not modelled explicitly as the initial mechanical situation was the same for both the healing and the non-union group ., During the healing process , the development of stiffer tissues such as bone might alter the local mechanical conditions in the healing group thereby possibly influencing the healing process ., However , under normal mechanical conditions ( i . e . appropriate external stabilisation that allows for normal healing to occur ) the bioregulatory model used in this study 15 behaves the same as an extended mechanobioregulatory variant 16 that incorporates mechanical loading explicitly ., The use of stem cells in the treatment of non-unions is gaining interest ( reviewed in 34 ) ., However , to date these stem cells were delivered in a ( ceramic ) scaffold or carrier structure requiring invasive surgery ., Non-invasive techniques such as direct injection of a cell-buffer mixture or the use of injectable carriers could substantially reduce the additional trauma for the patient and were investigated by an in silico – in vivo approach in this study ., As blood vessel formation is delayed in the non-union group , the most suitable time point for intervention of either growth factor treatment or cell transplant , seems to be three weeks post-osteotomy , when the blood supply to the interfragmentary gap has started to recover ., Injection of MSCs directly into the callus area elicited to a good healing response in silico ., Experimental results confirmed that transplantation of MSCs into the interfragmentary gap at POW 3 prevented the onset of an atrophic non-union ., There was significantly more bone present in the treatment group ( cell transplant ) than in the control group ( carrier solution ) ., Union by bridging callus had occurred in three of the four treatment animals ., Yet , all of the animals treated by MSC transplantation displayed an asymmetric healing with endochondral ossification in progress in part of the intercortical callus ( Fig . 5D ) ., As the zone of endochondral ossification was not always observed on the same aspect of the tibia ( anterior vs posterior ) , mechanical loading was ruled out as the major cause ., The in silico experiments carried out in the framework of this study identified another potential cause , namely injection of the cell transplant excentrally in the callus ., Upon excentral injection at POW 3 , cartilage undergoing endochondral ossification was predicted to be present at the intracortical gap opposite the injection site at POW 16 ( Fig . 5Bi ) ., The predicted amount of soft tissues present in the callus in that case agreed well with the experimentally measured amount ( Fig . 5Bii ) ., The use of an injectable carrier to deliver the cells close to the fracture healing site did not generate a good healing response in silico ., The limitation of the MSCs migration speed due to the fibrous extracellular matrix that has formed during the first three weeks is not problematic in case of injection directly into the callus area ., However , this does become an issue when the cells are administered in a carrier close to ( but not within ) the fracture site ., MSCs entering the callus area start differentiating under the influence of the growth factors that are present ., As the cells migrate slowly into the callus area , they become differentiated before they reach the central area of the callus ., Differentiated cells deposit bone matrix to replace the fibrous matrix which further decreases the migration speed of the cells that enter the callus area while the osteogenic growth factors that are expressed enhance the differentiation , resulting in a layer of bone close to the cell source while fibrous tissue persist in the major part of the callus area ., In this study we have shown that a mathematical model , initially developed for normal fracture healing , can be used as a clinical tool to investigate aetiology and treatment of atrophic non-unions ., Despite a number of deviations mainly due to simplifications in the model , the mathematical model is able to capture essential aspects of the atrophic non-union as observed experimentally in vivo ., Interestingly , the correspondence between simulations and experiments was obtained without changing the previously established parameter values , which clearly adds to the models potential ., Moreover , the mo

Introduction, Materials and Methods, Results, Discussion

Recently developed atrophic non-union models are a good representation of the clinical situation in which many non-unions develop ., Based on previous experimental studies with these atrophic non-union models , it was hypothesized that in order to obtain successful fracture healing , blood vessels , growth factors , and ( proliferative ) precursor cells all need to be present in the callus at the same time ., This study uses a combined in vivo-in silico approach to investigate these different aspects ( vasculature , growth factors , cell proliferation ) ., The mathematical model , initially developed for the study of normal fracture healing , is able to capture essential aspects of the in vivo atrophic non-union model despite a number of deviations that are mainly due to simplifications in the in silico model ., The mathematical model is subsequently used to test possible treatment strategies for atrophic non-unions ( i . e . cell transplant at post-osteotomy , week 3 ) ., Preliminary in vivo experiments corroborate the numerical predictions ., Finally , the mathematical model is applied to explain experimental observations and identify potentially crucial steps in the treatments and can thereby be used to optimize experimental and clinical studies in this area ., This study demonstrates the potential of the combined in silico-in vivo approach and its clinical implications for the early treatment of patients with problematic fractures .

In light of the ageing population , the occurrence of bone fractures is expected to rise substantially in the near future ., In 5 to 10% of these cases , the healing process does not succeed in repairing the bone , leading to the formation of delayed unions or even non-unions ., In this study we used a combination of an animal model mimicking a clinical non-union situation and a mathematical model developed for normal fracture healing to investigate both the causes of non-union formation and potential therapeutic strategies that can be applied to restart the healing process ., After showing that the mathematical model is able to simulate key aspects of the non-union formation , we have used it to investigate several treatment strategies ., One of these strategies , the treatment of a non-union involving a transplantation of cells from the bone marrow to the fracture site , was also tested in a pilot animal experiment ., Both the simulations and the experiments showed the formation of a bony union between the fractured bone ends ., In addition , we used the mathematical model to explain some unexpected experimental observations ., This study demonstrates the added value of using a combination of mathematical modelling and experimental research as well the potential of using cell transplantation for the treatment of non-unions .

physiology/muscle and connective tissue, rheumatology/bone and mineral metabolism, pathology/histopathology, rheumatology/orthopedics and sports medicine

journal.pgen.1005928

S/HIC: Robust Identification of Soft and Hard Sweeps Using Machine Learning

The availability of population genomic data has empowered efforts to uncover the selective , demographic , and stochastic forces driving patterns of genetic variation within species ., Chief among these are attempts to uncover the genetic basis of recent adaptation 1 ., Indeed , recent advances in genotyping and sequencing technologies have been accompanied by a proliferation of statistical methods for identifying recent positive selection see 2 for recent review ., Most methods for identifying positive selection search for the population genetic signature of a “selective sweep” 3 , wherein the rapid fixation of a new beneficial allele leaves a valley of diversity around the selected site 4–6 , about which every individual in the population exhibits the same haplotype ( i . e . the genetic background on which the beneficial mutation occurred ) ., At greater genetic distances , polymorphism recovers as recombination frees linked neutral variants from the homogenizing force of the sweep 4 ., This process also produces an excess of low- and high-frequency derived alleles 7 , 8 , and increased allelic association , or linkage disequilibrium ( LD ) , on either side of the sweep 9 , but not across the two flanks of the sweep 10 , 11 ., Selective fixation of de novo beneficial mutations such as described by Maynard Smith and Haigh 5 are often referred to as “hard sweeps . ”, More recently , population geneticists have begun to consider the impact of positive selection on previously standing genetic variants 12 , 13 ., Under this model of adaptation , an allele initially evolves under drift for some time , until a change in the selective environment causes it to confer a fitness advantage and sweep to fixation ., In contrast to the hard sweep model , the selected allele is present in multiple copies prior to the sweep ., Thus , because of mutation and recombination events occurring near the selected site during the drift phase , the region containing this site may exhibit multiple haplotypes upon fixation 14 ., The resulting reduction in diversity is therefore less pronounced than under the hard sweep model 12 , 15 ., For this reason sweeps from standing genetic variation are often referred to as “soft sweeps . ”, Soft sweeps will not skew the allele frequencies of linked neutral polymorphisms toward low and high frequencies to the same extent as hard sweeps 16 , and may even present an excess of intermediate frequencies 17 ., This mode of selection will also have a different impact on linkage disequilibrium: LD will be highest at the target of selection rather than in flanking regions 18 ., In very large populations , selection on mutations that are immediately beneficial may also produce patterns of soft sweeps rather than hard sweeps , as the adaptive allele may be introduced multiple times via recurrent mutation before the sweep completes 14 , 19 ., While this model of a soft sweep is similar to that of selection on standing variation in that it will produce additional haplotypes carrying the selected allele , there are differences in the patterns of polymorphism produced by these two types of soft sweeps 18 , 20 ., Here , we examine only the model of selection on a single standing variant ., Adaptation could proceed primarily through selection on standing variation if the selective environment shifts frequently relative to the time scale of molecular evolution , and if there is enough standing variation segregating in the population on which selection may act following such a shift 12 , 21 ., However , it is important to note that selection on standing variation may produce a hard sweep of only one haplotype containing the adaptive mutation if this allele is present at low enough frequency prior to sweep 16 , 22 ., In other words , the observation of hard sweeps may be consistent with selection on standing variation as well as selection on de novo mutations ., For these and other reasons , there is some controversy over whether adaptation will result in soft sweeps in nature 22 ., This could be resolved by methods that can accurately discriminate between hard and soft sweeps ., To this end , some recently devised methods for detecting population genetic signatures of positive selection consider both types of sweeps 23–25 ., Unfortunately , it may often be difficult to distinguish soft sweeps from regions flanking hard sweeps due to the “soft shoulder” effect 18 ., Here we present a method that is able to accurately distinguish between hard sweeps , soft sweeps on a single standing variant , regions linked to sweeps ( or the “shoulders” of sweeps ) , and regions evolving neutrally ., This method incorporates spatial patterns of a variety of population genetic summary statistics across a large genomic window in order to infer the mode of evolution governing a focal region at the center of this window ., We combine many statistics used to test for selection using an Extremely Randomized Trees classifier 26 , a powerful supervised machine learning classification technique ., We refer to this method as Soft/Hard Inference through Classification ( S/HIC , pronounced “shick” ) ., By incorporating multiple signals in this manner S/HIC achieves inferential power exceeding that of any individual test ., Furthermore , by using spatial patterns of these statistics within a broad genomic region , S/HIC is able to distinguish selective sweeps not only from neutrality , but also from linked selection with much greater accuracy than other methods ., Thus , S/HIC has the potential to identify more precise candidate regions around recent selective sweeps , thereby narrowing down searches for the target locus of selection ., Further , S/HIC’s reliance on large-scale spatial patterns makes it more robust to non-equilibrium demography than previous methods , even if the demographic model is misspecified during training ., This is vitally important , as the true demographic history of a population sample may be unknown ., Finally , we demonstrate the utility of our approach by applying it to chromosome 18 in the CEU sample from the 1000 Genomes dataset 27 , recovering most of the sweeps identified previously in this population through other methods; we also highlight a compelling novel candidate sweep in this population ., We sought to devise a method that could not only accurately distinguish among hard sweeps , soft sweeps , and neutral evolution , but also among these modes of evolution and regions linked to hard and soft sweeps , respectively 18 ., Such a method would not only be robust to the soft shoulder effect , but would also be able to more precisely delineate the region containing the target of selection by correctly classifying unselected but closely linked regions ., In order to accomplish this , we sought to exploit the impact of positive selection on spatial patterns of several aspects of variation surrounding a sweep ., Not only will a hard sweep create a valley of diversity centered around a sweep , but it will also create a skew toward high frequency derived alleles flanking the sweep and intermediate frequencies at further distances 7 , 8 , reduced haplotypic diversity at the sweep site 24 , and increased LD along the two flanks of the sweep but not between them 10 ., For soft sweeps , these expected patterns may differ considerably 14 , 16 , 18 , but also depart from the neutral expectation ., While some of these patterns of variation have been used individually for sweep detection e . g . 10 , 28 , we reasoned that by combining spatial patterns of multiple facets of variation we would be able to do so more accurately ., To this end , we designed a machine learning classifier that leverages spatial patterns of a variety of population genetic summary statistics in order to infer whether a large genomic window recently experienced a selective sweep at its center ., We accomplished this by partitioning this large window into adjacent subwindows , measuring the values of each summary statistic in each subwindow , and normalizing by dividing the value for a given subwindow by the sum of values for this statistic across all subwindows within the same window to be classified ., Thus , for a given summary statistic x , we used the following vector:, x1∑ixix2∑ixi…xn∑ixi, where the larger window has been divided into n subwindows , and xi is the value of the summary statistic x in the ith subwindow ., Thus , this vector captures differences in the relative values of a statistic across space within a large genomic window , but does not include the actual values of the statistic ., In other words , this vector captures only the shape of the curve of the statistic x across the large window that we wish to classify ., Our goal is to then infer a genomic region’s mode of evolution based on whether the shapes of the curves of various statistics surrounding this region more closely resemble those observed around hard sweeps , soft sweeps , neutral regions , or loci linked to hard or soft sweeps ., In addition to allowing for discrimination between sweeps and linked regions , this strategy was motivated by the need for accurate sweep detection in the face of a potentially unknown nonequilibrium demographic history , which may grossly affect values of these statistics but may skew their expected spatial patterns to a much lesser extent ., In total , we constructed these vectors for each of π 29 ,, θ^w, 30 ,, θ^H, 8 , the number of distinct haplotypes , average haplotype homozygosity , H12 and H2/H1 24 , 31 , ZnS 9 , and the maximum value of ω 10 ., Thus , we represent each large genomic window by the following vector , to which we refer as the feature vector:, π1∑iπiπ2∑iπi…πn∑iπiθ^w1∑iθ^wiθ^w2∑iθ^wi…θ^wn∑iθ^wiθ^H1∑iθ^Hiθ^H2∑iθ^Hi…θ^Hn∑iθ^Hi………ω1∑iωiω2∑iωi…ωn∑iωi, We sought to discriminate between hard sweeps , regions linked to hard sweeps , soft sweeps , regions linked to soft sweeps , and neutrally evolving regions on the basis of the values of the vectors defined above ., While Berg and Coop 20 recently derived approximations for the site frequency spectrum ( SFS ) for a soft sweep under equilibrium population size , and π , the joint probability distribution of the values all of the above statistics at varying distances from a sweep is unknown ., Moreover expectations for the SFS surrounding sweeps ( both hard and soft ) under nonequilibrium demography remain analytically intractable ., Thus rather than taking a likelihood approach , we opted to use a supervised machine learning framework , wherein a classifier is trained from simulations of regions known to belong to one of these five classes ., We trained an Extra-Trees classifier ( aka extremely randomized forest; 26 ) from coalescent simulations ( described below ) in order to classify large genomic windows as experiencing a hard sweep in the central subwindow , a soft sweep in the central subwindow , being closely linked to a hard sweep , being closely linked to a soft sweep , or evolving neutrally according to the values of its feature vector ( Fig 1 ) ., Briefly , the Extra-Trees classifier is an ensemble classification technique that harnesses a large number classifiers referred to as decision trees ., A decision tree is a simple classification tool that uses the values of multiple features for a given data instance , and creates a branching tree structure where each node in the tree is assigned a threshold value for a given feature ., If a given data point’s ( or instance’s ) value of the feature at this node is below the threshold , this instance takes the left branch , and otherwise it takes the right ., At the next lowest level of the tree , the value of another feature is examined ., When the data instance reaches the bottom of the tree , it is assigned a class inference based on which leaf it has landed 32 ., Typically , a decision tree is built according to an algorithm designed to optimize its accuracy 32 ., The Extra-Trees classifier , on the other hand , builds a specified number of semi-randomly generated decision trees ., Classification is then performed by simply taking the class receiving the most “votes” from these trees 26 , building on the strategy of random forests 33 ., While individual decision trees may be highly inaccurate , the practice of aggregating predictions from many semi-randomly generated decision trees has been proved to be quite powerful 34 ., In the following sections we describe our methodology for training , testing , and applying our Extra-Trees classifier for identifying positive selection ., We simulated data for training and testing of our classifier using our coalescent simulator , discoal_multipop ( https://github . com/kern-lab/discoal_multipop ) ., As discussed in the Results , we simulated training sets with different demographic histories ( S1 Table ) , and , for positively selected training examples , different ranges of selection coefficients ( α = 2Ns , where s is the selective advantage and N is the population size ) ., For each combination of demographic history and range of selection coefficients , we simulated large chromosomal windows that we later subdivided into 11 adjacent and equally sized subwindows ., We then simulated training examples with a hard selective sweep whose selection coefficient was uniformly drawn from the specified range , U ( αlow , αhigh ) ., We generated 11 , 000 sweeps: 1000 where the sweep occurred in the center of the leftmost of the 11 subwindows , 1000 where the sweep occurred in the second subwindow , and so on ., We repeated this same process for soft sweeps at each location; these simulations had an additional parameter , the derived allele frequency , f , at which the mutation switches from evolving under drift to sweeping to fixation , which we drew from U ( 0 . 05 , 0 . 2 ) , U ( 2/2N , 0 . 05 ) , or U ( 2/2N , 0 . 2 ) as described in the Results ., For our equilibrium demography scenario , we drew the fixation time of the selective sweep from U ( 0 , 0 . 2 ) ×N generations ago , while for non-equilibrium demography the sweeps completed more recently ( see below ) ., We also simulated 1000 neutrally evolving regions ., Unless otherwise noted , for each simulation the sample size was set to 100 chromosomes ., For each combination of demographic scenario and selection coefficient , we combined our simulated data into 5 equally-sized training sets ( Fig 1 ) : a set of 1000 hard sweeps where the sweep occurs in the middle of the central subwindow ( i . e . all simulated hard sweeps ) ; a set of 1000 soft sweeps ( all simulated soft sweeps ) ; a set of 1000 windows where the central subwindow is linked to a hard sweep that occurred in one of the other 10 windows ( i . e . 1000 simulations drawn randomly from the set of 10000 simulations with a hard sweep occurring in a non-central window ) ; a set of 1000 windows where the central subwindow is linked to a soft sweep ( 1000 simulations drawn from the set of 10000 simulations with a flanking soft sweep ) ; and a set of 1000 neutrally evolving windows unlinked to a sweep ., We then generated a replicate set of these simulations for use as an independent test set ., We used the python scikit-learn package ( http://scikit-learn . org/ ) to train our Extra-Trees classifier and to perform classifications ., Given a training set , we trained our classifier by performing a grid search of multiple values of each of the following parameters: max_features ( the maximum number of features that could be considered at each branching step of building the decision trees , which was set to 1 , 3 ,, n, , or n , where n is the total number of features ) ; max_depth ( the maximum depth a decision tree can reach; set to 3 , 10 , or no limit ) , min_samples_split ( the minimum number of training instances that must follow each branch when adding a new split to the tree in order for the split to be retained; set to 1 , 3 , or 10 ) ; min_samples_leaf ., ( the minimum number of training instances that must be present at each leaf in the decision tree in order for the split to be retained; set to 1 , 3 , or 10 ) ; bootstrap ( a binary parameter that governs whether or not a different bootstrap sample of training instances is selected prior to the creation of each decision tree in the classifier ) ; criterion ( the criterion used to assess the quality of a proposed split in the tree , which is set to either Gini impurity 35 or to information gain , i . e . the change in entropy 32 ) ., The number of decision trees included in the forest was always set to 100 ., After performing a grid-search with 10-fold cross validation in order to identify the optimal combination of these parameters , we used this set of parameters to train the final classifier ., We used the scikit-learn package to assess the importance of each feature in our Extra-Trees classifiers ., This is done by measuring the mean decrease in Gini impurity , multiplied by the average fraction of training samples that reach that feature across all decision trees in the classifier ., The mean decrease in impurity for each feature is then divided by the sum across all features to give a relative importance score , which we show in S2 Table ., We also show values of Extra-Trees classifier parameters resulting from grid searchers in S3 Table ., We compared the performance of our classifier to that of various other methods ., First , we examined two population genetic summary statistics: Tajima’s D 36 and Kim and Nielsen’s ωmax 10 ( which we refer to as ω for simplicity ) , calculating their values in each subwindow within each large simulated chromosome that we generated for testing ( see above ) ., We also used Nielsen et al . ’s composite-likelihood ratio test , referred to as CLR or SweepFinder 28 , which searches for the spatial skew in allele frequencies expected surrounding a hard selective sweep ., When testing SweepFinder’s ability to discriminate between modes of evolution within larger regions , we computed the composite-likelihood ratio between the sweep and neutral models at 200 sites across each of the 11 subwindows of our large simulated test regions , taking the maximum CLR value ., The only training necessary for SweepFinder was to specify the neutral site frequency spectrum ., Next , we used scikit-learn to implement Ronen et al . ’s 37 SFselect , a support vector machine classifier that discriminates between selection and neutrality on the basis of a region’s binned and weighted SFS ., In our implementation we collapsed the SFS into 10 bins as suggested by Ronen et al . , and also added soft sweeps as a third class ( in addition to hard sweeps and neutrality ) , using Knerr et al . ’s 38 method for extending a binary classifier to perform multi-class classification ., We trained this classifier from simulated data following the same demographic and selective scenarios used to train our own classifier , and with the same number of simulated training instances , but these simulations encapsulated much smaller regions ( equivalent to the size of one of our eleven subwindows ) ., To avoid confusion with the original SFselect , which only handles hard sweeps , we refer to this implementation as SFselect+ ., For further comparisons , we also trained a support vector machine using a vector of two statistics: the maximum values of the SweepFinder CLR statistic and ω ( a subset of the features in the Pavlidis et al . ’s SVM 39 ) ., We refer to this method as CLR+ω , and trained it in the same manner as SFselect+ , except for the different feature vector ., We also tested the performance of the evolBoosting 40 , an R package which uses an machine learning approach called boosting 41 to classify genomic windows as sweeps or neutral on the basis of several statistics , including Tajima’s D , Fay and Wu’s H 8 , integrated haplotype homozygosity ( iHH; 42 ) , and several others ., Boosting was also recently used by Pybus et al . 43 to accurately detect hard and partial sweeps and make coarse inferences about sweep ages ., Like S/HIC , this method uses a vector of the values of each of these statistics from several subwindows surrounding the region being classified ., However , unlike S/HIC , this method does not take the relative values of these statistics in each subwindow divided by the sum across all subwindows , instead just taking the value of the statistic measured in that subwindow ., As with SFselect , we extend this method to discriminate between hard sweeps , soft sweeps , and neutral windows ., This was done by first training a classifier to distinguish between sweeps ( hard and soft , balanced in number within the training set ) from neutral windows and secondarily training a classifier to distinguish between hard and soft sweeps ., Finally , we implemented a version of Garud et al . ’s 24 scan for hard and soft sweeps ., Garud et al . ’s method uses an Approximate Bayesian Computation-like approach to calculate Bayes Factors to determine whether a given region is more similar to a hard sweep or a soft sweep by performing coalescent simulations ., For this we performed simulations with the same parameters as we used to train SFselect+ , but generated 100 , 000 simulations of each scenario in order to ensure that there was enough data for rejection sampling ., We then used two statistics to summarize haplotypic diversity within these simulated data: H12 and H2/H1 31 ., All simulated regions whose vector H12 H2/H1 lies within a Euclidean distance of 0 . 1 away from the vector corresponding to the data instance to be classified are then counted 24 ., The ratio of simulated hard sweeps to simulated soft sweeps within this distance cutoff is then taken as the Bayes Factor ., Note that Garud et al . restricted their analysis of the D . melanogaster genome to only the strongest signals of positive selection , asking whether they more closely resembled hard or soft sweeps ., Therefore when testing the ability of Garud et al . ’s method to distinguish selective sweeps from both linked and neutrally evolving regions , we used large simulated windows and simply examined the value of H12 within the subwindow that exhibited the largest value in an effort to mimic their strategy of using H12 peaks 24 ., We summarized each method’s power using the receiver operating characteristic ( ROC ) curve , making these comparisons for the following binary classification problems: discriminating between hard sweeps and neutrality , between hard sweeps and soft sweeps , between selective sweeps ( hard or soft ) and neutrality , and between selective sweeps ( hard or soft ) and unselected regions ( including both neutrally evolving regions and regions linked to selective sweeps ) ., For each of these comparisons we constructed a balanced test set with a total of 1000 simulated regions in each class , so that the expected accuracy of a completely random classifier was 50% , and the expected area under the ROC curve ( AUC ) was 0 . 5 ., Whenever the task involved a class that was a composite of two or more modes of evolution , we ensured that the test set was comprised of equal parts of each subclass ., For example , in the selected ( hard or soft ) versus unselected ( neutral or linked selection ) test , the selected class consisted of 500 hard sweeps and 500 soft sweeps , while the unselected class consisted of 333 neutrally evolving regions , 333 regions linked to hard sweeps , and 333 regions linked to soft sweeps ( and one additional simulated region from one of these test sets randomly selected , so that the total size of the unselected test set was 1000 instances ) ., As with our training sets , we considered the true class of a simulated test region containing a hard ( soft ) sweep occurring in any but the central subwindow to be hard-linked ( soft-linked ) —even if the sweep occurred only one subwindow away from the center ., The ROC curve is generated by measuring performance at increasingly lenient thresholds for discriminating between the two classes ., We therefore required each method to output a real-valued measure proportional to its confidence that a particular data instance belongs the first of the two classes ., For S/HIC , we used the posterior classification probability from the Extra-Trees classifier obtained using scikit-learn’s predict_proba method ., For SFselect+ , we used the value of the SVM decision function ., For SweepFinder , we used the composite likelihood ratio ., For Garud et al . ’s method , we used the fraction of accepted simulations ( i . e . within a Euclidean distance of 0 . 1 from the test instance ) that were of the first class: for example , for hard vs . soft , this is the number of accepted simulations that were hard sweeps divided by the number of accepted simulations that were either hard sweeps or soft sweeps ., For Tajima’s D 36 and Kim and Nielsen’s ω 10 , we simply used the values of these statistics ., To examine the power and sensitivity of S/HIC under non-equilibrium demographic histories , we simulated training and test datasets from a few scenarios that might be relevant to researchers ., Firstly we examined the power of our method under two complex population size histories that are relevant to humans ., Secondly we examined the case of simple population bottlenecks , as might be common in populations that have recently colonized new locales , using two levels of bottleneck severity ., We simulated training and test datasets from Tennessen et al . ’s 44 European demographic model ( S1 Table ) ., This model parameterizes a population contraction associated with migration out of Africa , a second contraction followed by exponential population growth , and a more recent phase of even faster exponential growth ., Values of θ and ρ = 4Nr were drawn from prior distributions ( S1 Table ) , allowing for variation within the training data , whose means were selected from recent estimates of human mutation 45 and recombination rates 46 , respectively ., For simulations with selection , we drew values of α from U ( 5 . 0×103 , 5 . 0×105 ) , and drew the fixation time of the sweeping allele form U ( 0 , 51 , 000 ) years ago ( i . e . the sweep completed after the migration out of Africa ) ., We also generated simulations of Tennessen et al . ’s African demographic model , which consists of exponential population growth beginning ~5 , 100 years ago ( S1 Table ) ., We generated two sets of these simulations: one where α was drawn from U ( 5 . 0×104 , 5 . 0×105 ) , and one with α drawn from U ( 5 . 0×104 , 5 . 0×105 ) ., The sample size of these simulated data sets was set to 100 chromosomes ., These two sets were then combined into a single training set ., For these simulations , the sweep was constrained to complete some time during the exponential growth phase ( no later than 5 , 100 years ago ) ., Finally , we examined two models with a population size bottleneck ., The first was taken from Thornton and Andolfatto 47 , and models the demographic history of a European population sample of D . melanogaster ( S1 Table ) ., This model consists of a population size reduction 0 . 044×2N generations ago to 2 . 9% of the ancestral population size , and then 0 . 0084×2N generations ago the population recovers to its original size ., The second bottleneck model we used was identical except the population contraction was less severe ( reduction to 29% of the ancestral population size ) ., For sweep simulations under both of these bottleneck scenarios , we drew α from U ( 1 . 0×102 , 1 . 0×104 ) ., For all of our non-equilibrium demographic histories , when simulating soft sweeps on a previously standing variant , we drew the derived allele frequency at the onset of positive selection from U ( 2/2N , 0 . 2 ) ., For each demographic model in S1 Table , we show in S1 Fig the means and standard deviations of Tajima’s D across 11 windows at increasing distances from a selective sweep ( for one possible sweep scenario ) , as well as values from neutrally evolving windows for comparison ., For each demographic model in S1 Table , we show in S1 Fig the means and standard deviations of Tajima’s D across 11 windows at increasing distances from a selective sweep ( for one possible sweep scenario ) , as well as values from neutrally evolving windows for comparison ., We applied our method to chromosome 18 from the Phase I data release from the 1000 Genomes project 27 ., We restricted this analysis to the CEU population sample ( individuals with European ancestry , sampled from Utah ) , and trained S/HIC using data from the European demographic model described above ., After training this classifier , we prepared data from chromosome 18 in CEU for classification ., Prior to constructing feature vectors , we first performed extensive filtering for data quality ., First , we masked all sites flagged by the 1000 Genomes Project as being unfit for population genetic analyses due to having either limited or excessive read-depth or poor mapping quality ( according to the strictMask files for the Phase I data set which are available at http://www . 1000genomes . org/ ) ., In order to remove additional sites lying within repetitive sequence wherein genotyping may be hindered , we eliminated sites with 50 bp read mappability scores less than one 48 and also sites masked by RepeatMasker ( http://www . repeatmasker . org ) ., Finally , we attempted to infer the ancestral state at each remaining site , using the chimpanzee 49 and macaque 50 genomes as outgroups ., For each site , if the chimpanzee and macaque genomes agreed , we used this nucleotide as our inferred ancestral state ., If instead only the chimpanzee or the macaque genome had a nucleotide aligned to the site , we used this base as our inferred ancestral state ., For sites that were SNPs , we also required that the inferred ancestral state matched one of the two human alleles ., For all cases where these criteria were not met , we discarded the site ., After data filtering , we calculated summary statistics within adjacent 200 kb windows across the entire chromosome ., Importantly , we divided the values of each summary statistic by the number of sites in the window , ignoring sites filtered as discussed above ( i . e . π summarizes average nucleotide diversity per site rather than total diversity in the subwindow ) ., Windows with >50% of sites removed during the filtering processes were omitted from our analysis ., These two steps limited the effect of variation in the number of unfiltered sites from window to window our classification ., For the remaining windows , we used a sliding window approach with a 2 . 2 Mb window and a 200 kb step size to calculate the feature vector in the same manner as for our simulated data , and then applied S/HIC to this feature vector to infer whether the central subwindow of this 2 . 2 Mb region contained a hard sweep , a soft sweep , was linked to a hard sweep , linked to a soft sweep , or evolving neutrally ., Visualization of candidate regions was performed using the UCSC Genome Browser 51 ., We used hg19 coordinates for all of our analyses using human data ., Our classification tool is available at https://github . com/kern-lab/shIC , along with software for generating the feature vectors used in this paper ( either from simulated training data or from real data for classification ) ., The most basic task that a selection scan must be able to perform is to distinguish between hard sweeps and neutrally evolving regions , as the expected patterns of nucleotide diversity , haplotypic diversity , and linkage disequilibrium produced by these two modes of evolution differ dramatically 5 , 8 , 10 , 18 , 24 , 52 ., We therefore begin by comparing S/HIC’s power to discriminate between hard sweeps and neutrality to that of several previously published methods: these include SweepFinder aka CLR; 28 , SFselect 37 , Garud et al . ’s haplotype approach using the H12 and H2/H1 statistics 24 , Tajima’s D 36 , and Kim and Nielsen’s ω 10 , evolBoosting 40 , and a support vector machine implemented that uses CLR and ω statistics ( Methods ) ., We e

Introduction, Methods, Results, Discussion

Detecting the targets of adaptive natural selection from whole genome sequencing data is a central problem for population genetics ., However , to date most methods have shown sub-optimal performance under realistic demographic scenarios ., Moreover , over the past decade there has been a renewed interest in determining the importance of selection from standing variation in adaptation of natural populations , yet very few methods for inferring this model of adaptation at the genome scale have been introduced ., Here we introduce a new method , S/HIC , which uses supervised machine learning to precisely infer the location of both hard and soft selective sweeps ., We show that S/HIC has unrivaled accuracy for detecting sweeps under demographic histories that are relevant to human populations , and distinguishing sweeps from linked as well as neutrally evolving regions ., Moreover , we show that S/HIC is uniquely robust among its competitors to model misspecification ., Thus , even if the true demographic model of a population differs catastrophically from that specified by the user , S/HIC still retains impressive discriminatory power ., Finally , we apply S/HIC to the case of resequencing data from human chromosome 18 in a European population sample , and demonstrate that we can reliably recover selective sweeps that have been identified earlier using less specific and sensitive methods .

The genetic basis of recent adaptation can be uncovered from genomic patterns of variation , which are perturbed in predictable ways when a beneficial mutation “sweeps” through a population ., However , the detection of such “selective sweeps” is complicated by demographic events , such as population expansion , which can produce similar skews in genetic diversity ., Here , we present a method for detecting selective sweeps that is remarkably powerful and robust to potentially confounding demographic histories ., This method , called S/HIC , operates using a machine learning paradigm to combine many different features of population genetic variation , and examine their values across a large genomic region in order to infer whether a selective sweep has recently occurred near its center ., S/HIC is also able to accurately distinguish between selection acting on de novo beneficial mutations ( “hard sweeps” ) and selection on previously standing variants ( “soft sweeps” ) ., We demonstrate S/HIC’s power on a variety of simulated datasets as well as human population data wherein we recover several previously discovered targets of recent adaptation as well as a novel selective sweep .

cognitive science, engineering and technology, demography, population genetics, neuroscience, simulation and modeling, decision analysis, management engineering, artificial intelligence, population biology, research and analysis methods, computer and information sciences, decision trees, people and places, population metrics, population size, natural selection, evolutionary processes, genetics, biology and life sciences, genomics, evolutionary biology, genomics statistics, computational biology, machine learning

journal.pntd.0007270

Revisiting the taxonomy and evolution of pathogenicity of the genus Leptospira through the prism of genomics

Leptospirosis is an emerging zoonotic disease of worldwide distribution that affects more than 1 million people with 60 , 000 deaths per year 1 ., In addition , numerous animal hosts ( wild and domestic ) , such as livestock , can contract leptospirosis causing economical cost to subsistence and industrial farming 2 ., Exposure to soil or water contaminated with the urine of reservoir animals ( mostly rodents ) infected with pathogenic Leptospira is the most common way in which humans or animals contract leptospirosis 3 ., Importantly , the tropism of the infectious agent is not limited to a single host , but rather to multiple hosts that can be asymptomatic carriers or develop mild or severe diseases ., The life cycle of pathogenic Leptospira is therefore complex , including the natural environment , asymptomatic reservoir and susceptible hosts 4 ., Since its original description in 1907 by Stimson 5 , the genus Leptospira has been traditionally divided into two groups , saprophytes—Leptospira biflexa sensu lato—and pathogens—Leptospira interrogans sensu lato—based on their virulence ., More recently , phylogenetic analysis revealed that Leptospira can be divided in three lineages that correlate with the level of pathogenicity of the species: saprophytic , intermediate , and pathogenic 6 ., The intermediate species share a near common ancestor with pathogen species while exhibiting moderate pathogenicity in both humans and animals ., Both pathogenic and non-infectious environmental saprophytic Leptospira strains have been isolated from environmental sources as they are able to survive in moist soil and fresh water for several weeks 7 , 8 ., The ability of Leptospira to occupy various ecological niches is undoubtedly due to a diversity of mechanisms , such as signal transduction systems 9 , encoded by its large genome and that allow it to adapt and resist to stressful conditions 9 , 10 ., It has been suggested that pathogens might have evolved from an environmental ancestor by the acquisition of new functions through lateral gene transfers associated with the adaptation to new hosts 9 , 11 ., The discovery of novel Leptospira species , including species belonging to the pathogen and intermediate lineages , is critical for the development of robust detection and diagnostic tools that are desperately needed to treat infected hosts quicker and adequately ., Further characterization of populations of Leptospira in soil and water will also help inform prevention and control efforts aimed at reducing the risk of Leptospira infection from the environment ., It will also enable to better understand the ecology of Leptospira in the natural environment and its interactions with other microbial communities ., A deeper understanding of the biodiversity of strains that can lead to infections in both humans and animals is lacking ., For example , the role of intermediate species in both human and animal infections remains to be clearly established 10 ., Information concerning the genetic diversity of circulating Leptospira strains is also important to evaluate for the efficacy of current vaccines for the control of leptospirosis ., Accurate identification of infectious Leptospira is also of prime importance as antibiotic therapy is beneficial in the early stage of the disease ., Isolating Leptospira from soil using a novel combination of antimicrobial agents to prevent contamination 12 has recently uncovered many novel Leptospira species 13 , 14 ., The increasing availability of Next-Generation Sequencing ( NGS ) methods has also provided opportunities to identify novel Leptospira species ., All together , these recent advances resulted in an important expansion in leptospiral taxonomy with 35 named Leptospira species 14 , 15 ., In the present study , we have isolated new strains from diverse geographical origins and have undertaken a large genomic study in order to dust off the Leptospira genus to draw a better picture of its diversity and to propose new standards on its classification and nomenclature to replace the current one that is complex and obsolete 16 ., Thus , the classical method of DNA-DNA Hybridization ( DDH ) for species identification and the serological techniques for serovar identification of Leptospira strains will most probably not be used in any laboratory in the near future ., The taxonomic status of all species of the genus Leptospira , as well as 90 strains isolated from the natural environment across a wide geographic range , was evaluated by comparative genomics ., Our results reveal that the genus Leptospira now contains 64 named species , including species from a new subclade that is sister to the one that contains the traditional saprophytic species ., We propose a new systematic classification scheme of Leptospira species to replace the former one that heavily rely on assumption based on virulence level that is often uncharacterized ., The high resolution of the dataset used in this study allowed us to investigate the specificities of each clade and to demonstrate significant divergence in pathogenic strains ., We have also been able to point a dichotomy in these pathogenic species that is corroborated by different genomic characters ., This study will advance many aspects of the leptospirosis field including diagnostics , and basic knowledge including species diversity , evolution , ecology , and virulence ., Leptospira strains used in this study were isolated from water or soil samples from mainland France ( two sites ) , Algeria ( one site ) , Japan ( four sites ) , Malaysia ( four sites ) , Mayotte ( four sites ) and New Caledonia ( three sites ) as previously described 14 , 17 ., Leptospira strains were grown at 30°C in liquid Ellinghausen , McCullough , Johnson and Harris ( EMJH ) medium ., Phenotypic characterization of representative strains was performed by assessing their growth at 14°C , 30°C , and 37°C in liquid EMJH without shaking ., Growth in EMJH liquid medium supplemented with 225 μg/ml of the purine analogue 8-azaguanine at 30°C was also tested ., Representative strains were plated on 1% agar solid EMJH media and incubated at 30°C until individual subsurface colonies were visible ., Strains used in this study are available at the National Reference Center for Leptospirosis , Institut Pasteur , Paris , France ., Type strains of new Leptospira species were also deposited in the DSMZ-German Collection of Microorganisms ( www . dsmz . de ) and the National Collaborating Centre for Reference and Research on Leptospirosis , Amsterdam , The Netherlands ( http://leptospira . amc . nl/leptospira-library/ ) , except for species Leptospira kobayashii , Leptospira ryugenii , Leptospira ellinghausenii , and Leptospira johnsonii which were deposited in the CIP-Collection of Institut Pasteur ( www . pasteur . fr/fr/crbip ) and Japan Collection of Microorganisms ( http://jcm . brc . riken . jp/en/ ) ., Collection of the strains was conducted according to the Declaration of Helsinki ., A written informed consent from patients was not required as the study was conducted as part of routine surveillance of the national reference center and no additional clinical specimens were collected for the purpose of the study ., Cultures originating from human samples were anonymized ., Approval for bacterial isolation from soil and water was not required as the study was conducted as part of investigations of leptospirosis outbreaks ., For New Caledonia , approval for bacterial isolation from the natural environment was obtained from the South Province ( reference 1689–2017 ) and North Province ( reference 60912-2002-2017 ) ., Protocols for animal experiments conformed to the guidelines of the Animal Care and Use Committees of the Institut Pasteur ( Comité d’éthique d’expérimentation animale CETEA # 2016–0019 ) , agreed by the French Ministry of Agriculture ., All animal procedures carried out in our study were performed in accordance with the European Union legislation for the protection of animals used for scientific purposes ( Directive 2010/63/EU ) ., In this study , the DNA of a total of 90 Leptospira strains were sequenced ( S1 Table ) , including the type strain L . idonii Eri-1T 13 , whose genome sequence was not available ., Genomic DNA was prepared by centrifugation of exponential-phase cultures and extraction with MagNA Pure 96 Instrument ( Roche ) ., Next-generation sequencing was performed by the Mutualized Platform for Microbiology ( P2M ) at Institut Pasteur , using the Nextera XT DNA Library Preparation kit ( Illumina ) , the NextSeq 500 sequencing systems ( Illumina ) , and the CLC Genomics Workbench 9 software ( Qiagen ) for de novo assemblies ., The draft genomes with 50x minimum coverage were used for subsequent analysis and they were submitted to GenBank; accession numbers are available in S1 Table ., The genomic DNA of L . kobayashii E30T , L . ryugenii YH101T , L . ellinghausenii E18T , and L . johnsonii E8T were sequenced at the Sequencing facility at the University of Hokkaido ( Japan ) ( Mazusawa et al . submitted ) ., A delineation of the species for the genome sequences was performed by Average Nucleotide Identity ( ANI ) using pyani version 0 . 2 . 7 ( https://github . com/widdowquinn/pyani ) ., Subsequently , one genome per species was chosen and added to reference genomes available in GenBank , in order to compose a dataset of 64 genomes of Leptospira ., The genome sequences of Turneriella parva DSM 2152 ( GenBank Assembly # GCA_000266885 . 1 ) and Leptonema illini DSM 21528 ( GenBank Assembly # GCA_000243335 . 1 ) were added as outgroup for phylogenetic analysis ., All 66 genomic sequences were annotated with Prokka version 1 . 12 18 ., The orthology between the coding sequences has been inferred with GET_HOMOLOGUES version 20092018 using the COG and OMCL algorithms 19 ., Sequences of 1371 orthologous genes that are in single copy and in the softcore ( present in at least 95% of genomes ) were codon aligned using MAFFT version 7 . 397 20 through TranslatorX version 1 . 1 21 ., The resulting alignments were filtered using BMGE version 1 . 12 22 and concatenated in a partitioned supermatrix using AMAS 23 ., The best-fit model was determined for each of the partitions using IQ-TREE version 1 . 6 . 7 24 ., A maximum likelihood phylogenetic analysis with 10 , 000 ultrafast bootstraps was subsequently performed with the same tool 25 ., The gene sequences predicted to be in the core genome ( present in all genomes ) by GET_HOMOLOGUES were aligned as previously described ., A phylogenetic tree was made with each of the 553 resulting alignments using IQ-TREE ( the best-fit model was found for each of the alignments ) ., The Robinson-Foulds distance was calculated for each of the trees compared to the softcore based one , also using IQ-TREE ., The 16S rRNA sequences of the 66 genomes ( including the outgroups ) , those of strains detected in the environment of the Peruvian Amazon 26 and insectivorous bats from eastern China 27 were aligned and positions of low confidence level masked using SSU-ALIGN version 0 . 1 . 1 ( http://eddylab . org/software/ssu-align ) ., The best-fit model was determined and a maximum likelihood phylogenetic analysis with 10 , 000 ultrafast bootstraps was performed with the same tool with IQ-TREE version 1 . 6 . 7 ., The Amino Acid Identity ( AAI ) and the Percentage Of Conserved Proteins ( POCP ) values were determined using GET_HOMOLOGUES version 20092018 ., The core and pan genome of Leptospira was also evaluated using the same tool ., The genomic characteristics were determined using a combination of QUAST version 5 . 0 . 0 28 , Artemis version 17 . 0 . 1 29 and the DFAST web server ( for the number of pseudogenes ) ., The genes coding for lipoproteins have been annotated with SpLip version 1 30 ., Finally , the CDSs were classified into functional categories using eggnog-mapper version 1 . 0 . 3 31 and the PFAM motifs found by InterProScan version 5 . 31–70 . 0 32 ., All statistical analyses were performed with PRISM 6 ., In order to estimate the level of significance between the different clades an Agostino and Pearson omnibus normality test was carried out to check if the data followed a normal distribution ., In case where the data were normally distributed , one-way ANOVA with Tukeys multiple test comparisons were performed ., In the opposite case , a Kruskal-Wallis test with a Dunns multiple test comparison were performed ., In order to compare the level of significance between the two groups of species composing the S1 subclade ( see the result section ) , normality was also verified by an Agostino & Pearson omnibus normality test ., A t-test was then performed if the data were normally distributed or a Mann Whitney test where appropriate ., Groups of 4 golden Syrian hamsters ( 4-week-old males; Janvier , Le Genest , France ) were infected via intraperitoneal injection of 108 L . ilyithenensis and L . ognonensis type strains or 106 L . interrogans serovar Manilae strain L495 ., Animals were monitored daily for clinical signs of leptospirosis ( i . e . prostration , jaundice ) and survival ., Surviving animals were euthanized after a 14 day post-challenge follow-up period , and the kidneys and liver from each animal were harvested for culturing in EMJH medium ., A collection of environmental isolates from Asia ( Japan and Malaysia ) , Africa ( Algeria and the island of Mayotte , a French overseas department in the Indian Ocean ) , Europe ( France ) and Oceania ( New Caledonia ) were included in this study ., Leptospira isolates were retrieved from water and soil samples from different sites from 2008 to 2017 ., Except for the Japan isolates 17 , environmental isolates reported in this study were not described previously ., The DNA of a total of 90 isolates was sequenced using Illumina technology ( S1 Table ) ., The 90 Leptospira strains had an average genome size of 4 , 128 , 000 ± 221 , 345 bp ., The largest genome was 4 , 993 , 538 bp , belonging to Leptospira putramalaysiae strain SSW20 ., The smallest genome was the genome of Leptospira fletcheri strain SSW15T , 3 , 733 , 663 bp in size ., The GC content of the genomes in this study ranged from 37 . 06 to 47 . 70 ., The average genome assembly contained 49 ± 50 contigs ( S1 Table ) ., The complete genome sequences of the 90 strains described in this study were compared to the previously published genome sequences from the already known Leptospira species and strain GWTS#1 , that was wrongly assigned to the species Leptospira alstonii 33 , 34 ( S1 Table ) ., As a note , we sequenced the genome of L . idonii strain Eri-1T because a genome sequence was not available for this species 13 ., Also , the genome of the recently described species Leptospira macculloughii 15 was excluded from further analysis in our list of reference genomes as the genome of L . macculloughii was the result of a mixed culture of L . meyeri and L . levetti ., The results obtained from pairwise comparisons of the 124 genome sequences are summarized as a matrix in the S2 Table ., Using a ANI cutoff of 95% generally used as the metrics to delineate bacterial species 35 , we established the existence of 64 different species of Leptospira ., These species include 34 previously described species , 4 new species from Japan ( Masuzawa et al . submitted ) , 25 newly isolated species ( Leptospira kemamanensis sp . nov . , Leptospira andrefontaineae sp . nov . , Leptospira bandrabouensis sp . nov . , Leptospira bouyouniensis sp . nov . , Leptospira congkakensis sp . nov . , Leptospira dzianensis sp . nov . , Leptospira dzoumogneensis sp . nov . , Leptospira fletcheri sp . nov . , Leptospira fluminis sp . nov . , Leptospira gomenensis sp . nov . , Leptospira ilyithenensis sp . nov . , Leptospira jelokensis sp . nov . , Leptospira kanakyensis sp . nov . , Leptospira langatensis sp . nov . , Leptospira montravelensis sp . nov . , Leptospira mtsangambouensis sp . nov . , Leptospira noumeaensis sp . nov . , Leptospira ognonensis sp . nov . , Leptospira perdikensis sp . nov . , Leptospira . putramalaysiae sp . nov . , Leptospira sarikeiensis sp . nov . , Leptospira selangorensis sp . nov . , Leptospira semungkisensis sp . nov . , Leptospira koniamboensis sp . nov . , Leptospira bourretii sp . nov . ) and one new species ( Leptospira tipperaryensis sp . nov . ) which was previously wrongly assigned to L . alstonii ( strain GWTS#1T ) based on the 16S rRNA and secY genes 33 , 34 ., Only one representative strain of each of the 64 Leptospira species was retained for further analysis ., The names and the descriptions of origins of these new species are indicated below ., Given our curated database , we assessed the taxonomic assignation of genome sequences already available in GenBank with our corresponding reference strains ( S1 File ) ., In doing so , we could easily detect misclassifications of strains ( such as for some L . santarosai , L . weilii and L . interrogans assigned strains , see S1 File ) , which validate the use of our database and the methodology ., The phylogenetic position of each species was robustly evaluated by performing a molecular phylogeny based on 1371 orthologous gene sequences coupled to a matrix of ANI values for each of the 64 Leptospira species ( Fig 1 ) ., The figure confirms 64 well-delineated species of Leptospira ., The ANI values are consistent with their phylogenetic relationships ., The interspecies ANI values ranged from ~69% to ~94% ., Four of the new 30 species , isolated from the natural environment in Malaysia , Mayotte , and New Caledonia and in small mammals in Ireland , were classified within the lineage of pathogens ., Ten novel species , isolated from Malaysia , Mayotte , Japan and New Caledonia , were identified as part of the intermediates ., Twelve novel species , isolated from Malaysia , Mayotte , Japan , and New Caledonia , were assigned to the saprophytes ., Finally , four novel species , isolated from Japan , Algeria , and France , were positioned in a clade sister to the one formed by saprophytes together with L . idonii ., Using this large amount of new species , we could refine the different clades and identified two major clades and four subclades in the Leptospira genus ., The two major clades are: “Saprophytes” containing species isolated in the natural environment and not responsible for infections and “Pathogens” containing all the species responsible for infections in humans and/or animals , plus environmental species for which the virulence status has not been proven ., The two clades are further subdivided in two subclades each ., We propose a new nomenclature in order to limit the assumption of virulence character that remain to be characterized ( Fig 1 ) : clades P and S and subclade P1 ( formerly described as the pathogen group ) , P2 ( formerly described as the intermediate group ) , S1 ( formerly described as the saprophyte group ) and S2 ( the new subclade described here that includes L . idonii ) ., We also compared the amino acid identity ( AAI ) values from the translated sequences of the different species ., This analysis is complementary to that of ANIs in the sense that amino acids evolve less rapidly than nucleotides ( degeneracy of the genetic code ) , thus making it possible to visualize larger and more ancestral groups 36 ., As expected , we find the same clades and subclades as with ANI values ( S1 Fig ) ., However , the signal is stronger and the clades better defined ., Similarly , the percentage of conserved proteins ( POCP ) values were also calculated across the 64 genome sequences to be compared ( S2 Fig ) ., POCP was determined using all the proteins of the genomes to infer the genetic and phenotypic relatedness between a pair of species 37 ., With this analysis , we can clearly see the two clades S and P ( again confirming the genetic relatedness between the ‘former’ saprophytes clade S1 and the new subclade S2 ) ., The 30 novel Leptospira species grow well in liquid EMJH at 30°C ., Under dark-field microscopy , strains of these novel species are motile and exhibit the characteristic hook- and spiral-shaped ends that are due to the rotation of the endoflagella ., They all exhibit a morphology which is consistent with the genus Leptospira , i . e . thin , long and helix-shaped cells ., A more detailed phenotypic analyze of the strains composing the new S2 subclade was performed ., L . kobayashii , L . ilyithenensis , L . idonii , and L . ognonensis were tested for their growth phenotypes under different conditions ., Isolates were also selected to provide one representative from each of the three other subclades ( P1: L . interrogans strain L495; P2: L . licerasiae strain VAR010T; S1: L . biflexa strain Patoc 1 ) ., The optimum temperature for growth is 30°C for all Leptospira species ., The doubling time in liquid EMJH at 30°C is between 15 and 23 hours , except for L . kobayashii with a doubling time of 8, h . Species of the subclade S2 grow well at 14°C but not at 37°C showing growth characteristics normally observed for species from subclade S1 ( formerly called saprophytes ) and not observed for species from the P clade that can also grow at 37°C ., Similarly , species of the subclade S2 can grow in EMJH supplemented with the purine analogue 8-azaguanine which was usually used as a differential agent for the separation of pathogenic ( P1 ) and saprophytic ( S1 ) Leptospira ( S4 Table ) ., To evaluate the virulence of species of this new subclade , two representative isolates ( L . ognonensis and L . ilyithenensis ) were injected at high dose ( 108 bacteria ) in hamsters ., Animals infected with these novel species did not exhibit any clinical sign of leptospirosis , and bacteria were not recovered from kidneys or livers of infected animals using homogenate’s culture ., In contrast , challenge with the virulent L . interrogans strain L495 caused death in infected hamsters ., These results indicate the inability of these novel species to establish acute infection or renal colonization in this animal model ., Cell morphology was similar to those of members of the genus Leptospira ., Cells were helix-shaped with a length of 9 to 14 μm , a diameter of ~0 . 2 μm and a wavelength ranging from 0 . 6 to 0 . 9 μm ( S3 Fig and S4 Table ) ., Description of Leptospira dzianensis sp ., nov ., : dzi . an . ensis ., N . L . fem ., adj . dzianensis of Dziani , a lake in Mayotte ., The type strain is M12AT , isolated from a water sample in Dziani , Mayotte ., Genome Accession Number is RQHR00000000 . The genomic G+C content of the type strain is 45 . 5% ., Belonging to the subclade P1 ., Description of Leptospira tipperaryensis sp ., nov ., : tip . pe . ra . ry . ensis ., N . L . fem ., adj . tipperaryensis pertaining to Tipperary , a county in Ireland ., The type strain is GWTS#1T , isolated from Crocidura russula in Tipperary , Ireland ., Previously assigned to L . alstonii 33 , 34 , this strain belongs to the subclade P1 ., Genome Accession Number is GCA_001729245 . 1 ., The genomic G+C content of the type strain is 42 . 4% ., Description of Leptospira gomenensis sp ., nov ., : go . men . ensis ., N . L . fem ., adj . gomenensis of Kaala-Gomen , a village in New Caledonia ., The type strain is KG8-B22T , isolated from a soil sample in Kaala-Gomen , North Province of New Caledonia ., Genome Accession Number is RQFA00000000 ., The genomic G+C content of the type strain is 46 . 1% ., Belonging to the subclade P1 ., Description of Leptospira putramalaysiae sp ., nov ., : put . ra . ma . laysi . ae ., N . L . gen . n ., putramalaysiae of Putra Malaysia , university hosting the laboratory who isolated the strain ., The type strain is SSW20T , isolated from a water sample in Sungai Congkak , Malaysia ., Genome Accession Number is RQEQ00000000 ., The genomic G+C content of the type strain is 42 . 5% ., Belonging to the subclade P1 ., Description of Leptospira andrefontaineae sp ., nov ., : an . dre . fon . taine . ae ., N . L . gen . n . andrefontaineae of André‐Fontaine , named in honor of Geneviève André‐Fontaine , a french veterinarian , who made significant contribution to the study of animal leptospirosis ., The type strain is PZF11-2T , isolated from a water sample in Nouméa , New Caledonia ., Genome Accession Number is RQEY00000000 ., The genomic G+C content of the type strain is 39 . 9% ., Belonging to the subclade P2 ., Description of Leptospira dzoumogneensis sp ., nov ., : dzou . mog . ne ., ensis ., N . L . fem ., adj . dzoumognensis of Dzoumogné , a village in Mayotte ., The type strain is M11AT , isolated from a water sample in Dzoumogné , Mayotte ., Genome Accession Number is RQHS00000000 ., The genomic G+C content of the type strain is 41 . 0% ., Belonging to the subclade P2 ., Description of Leptospira koniamboensis sp ., nov ., : N . L . fem ., adj . koniamboensis of Koniambo , mountain in New Caledonia ., The type strain is TK1-4T , isolated from a water sample in Koné , North Province of New Caledonia ., Genome Accession Number is RQFY00000000 ., The genomic G+C content of the type strain is 39 . 0% ., Belonging to the subclade P2 ., Description of Leptospira sarikeiensis sp ., nov ., : sa . ri . kei . ensis ., N . L . fem ., adj . sarikeienis of the district of Sarikei ., The type strain is LIMR175T , isolated from a water sample in Sarawak , Malaysia ., Genome Accession Number is RQGF00000000 ., The genomic G+C content of the type strain is 40 . 3% ., Belonging to the subclade P2 ., Description of Leptospira johnsonii sp ., nov ., : john . soni . i ., N . L . gen . n ., johnsonii of Johnson , named in honor of Russel C . Johnson , an American microbiologist who developed EMJH medium that is commonly used for Leptospira culture ., The type strain is E8T , isolated from a soil sample in Ibaraki , Japan 17 ., Genome Accession Number is BFAY00000000 ., The genomic G+C content of the type strain is 41 . 3% ., Belonging to the subclade P2 ., Description of Leptospira fluminis sp ., nov ., : flumi . nis ., L . gen . n ., fluminis of a river ., The type strain is SCS5T , isolated from a soil sample in Sungai Congkak , Malaysia ., Genome Accession Number is RQEV00000000 . The genomic G+C content of the type strain is 47 . 7% ., Belonging to the subclade P2 ., Description of Leptospira fletcheri sp ., nov ., : fletche . ri ., N . L . gen . n ., fletcheri of Fletcher , named in honor of William Fletcher who reported the first case of leptospirosis in Malaysia in 1927 ., The type strain is SSW15T , isolated from a water sample in Sungai Congkak , Malaysia ., Genome Accession Number is RQET00000000 ., The genomic G+C content of the type strain is 47 . 3% ., Belonging to the subclade P2 ., Description of Leptospira semungkisensis sp ., nov ., : se . mung . kis . ensis ., N . L . fem ., adj . semungkisensis of Semungkis , a river in the Hulu Langat district of Selangor state , Malaysia ., The type strain is SSS9T , isolated from a soil sample in Sungai Congkak , Malaysia ., Genome Accession Number is RQEP00000000 ., The genomic G+C content of the type strain is 42 . 8% ., Belonging to the subclade P2 ., Description of Leptospira langatensis sp ., nov ., : lan . gat . ensis ., N . L . fem ., adj . langatensis of the district of Langat , Malaysia ., The type strain is SSW18T , isolated from a water sample in Sungai Congkak , Malaysia ., Genome Accession Number is RQER00000000 ., The genomic G+C content of the type strain is 44 . 8% ., Belonging to the subclade P2 ., Description of Leptospira selangorensis sp ., nov ., : se . lan . gor . ensis ., N . L . fem ., adj . selangorensis of the state of Selangor , Malaysia ., The type strain is SSW17T , isolated from a water sample in Sungai Congkak , Malaysia ., Genome Accession Number is RQES00000000 ., The genomic G+C content of the type strain is 40 . 0% ., Belonging to the subclade P2 ., Description of Leptospira ognonensis sp ., nov ., : og . non . ensis ., N . L . fem ., adj . ognonensis of Ognon , river in France ., The type strain is 201702476T , isolated from a water sample in the region Bourgogne-Franche-Comté , France ., Genome Accession Number is RQHS00000000 ., The genomic G+C content of the type strain is 39 . 7% ., Belonging to the subclade S2 ., Description of Leptospira ilyithenensis sp ., nov ., : il . yi . then . ensis ., N . L . fem ., adj . ilyithenensis of Ilyithen , a village located in Algeria ., The type strain is 201400974T , isolated from a water sample in Ilyithen , a village located in the Djurdjura mountains , Algeria ., Genome Accession Number is RQHV00000000 ., The genomic G+C content of the type strain is 40 . 5% ., Belonging to the subclade S2 ., Description of Leptospira kobayashii sp ., nov ., : ko . ba . yashi . i ., N . L . gen . n ., kobayashii of Kobayashi , named in honor of Yuzuru Kobayashi , a Japanese physician and microbiologist who introduced monoclonal antibodies for the classification of Leptospira ., The type strain is E30T , isolated from a soil sample in Gifu , Japan 17 ., Genome Accession Number is BFBA00000000 ., The genomic G+C content of the type strain is 40 . 7% ., Belonging to the subclade S2 ., Description of Leptospira ryugenii sp ., nov ., : ru . geni . i ., N . L . gen . n ., ryugenii of Ryugen , nicknamed in honor of Yasutake Yanagihara , a Japanese microbiologist , University of Shizuoka , who contributed to the chemotaxonomy and study of Leptospira ., The type strain is YH101T , isolated from a water sample in Shizuoka , Japan 17 ., Genome Accession Number is BFBB00000000 ., The genomic G+C content of the type strain is 39 . 9% ., Belonging to the subclade S2 ., Description of Leptospira bandrabouensis sp ., nov ., : ban . dra . bou . a ., ensis ., N . L . fem ., adj . bandrabouaensis of Bandraboua , a commune in Mayotte ., The type strain is M10AT , isolated from a water sample in Bandraboua , Mayotte ., Genome Accession Number is RQHT00000000 ., The genomic G+C content of the type strain is 37 . 9% ., Belonging to the subclade S1 ., Description of Leptospira noumeaensis sp ., nov ., : nou . me . a ., ensis ., N . L . fem ., adj . noumeaensis of Nouméa , the capital city of New Caledonia ., The type strain is PZF14-4T , isolated from a water sample in Nouméa , South Province of New Caledonia ., Genome Accession Number is RQFK00000000 ., The genomic G+C content of the type strain is 38 . 3% ., Belonging to the subclade S1 ., Description of Leptospira jelokensis sp ., nov ., : je . lok . ensis ., N . L . fem ., adj ., Jelokensis of Jelok , a housing area in Kajang from where the strain was isolated ., The type strain is L5S1T , isolated from a soil sample in Sungai Jelok , Malaysia ., Genome Accession Number is RQGR00000000 ., The genomic G+C content of the type strain is 38 . 9% ., Belonging to the subclade S1 ., Description of Leptospira bourretii sp ., nov ., : N . L . gen . n ., bourretii of Bourret , named in honor of Henri Désiré Gaston Bourret ( 1875–1917 ) , a medical doctor and microbiologist who developed medical microbiology in New Caledonia ., The type strain is PZF7-6T , isolated from a soil sample in Nouméa , South Province of New Caledonia ., Genome Accession Number is RQFM00000000 ., The genomic G+C content of the type strain is 38 . 2% ., Belonging to the subclade S1 ., Description of Leptospira kanakyensis sp ., nov ., : ka . na . ky . ensis ., N . L . fem ., adj . kanakyensis of Kanaky , the name of New Caledonia for Kanak people ., The type strain is TK5-11T , isolated from a soil sample in Koné , North Province of New Caledonia ., Genome Accession Number is RQFG000000 ., The genomic G+C content of the type strain is 38 . 5% ., Belonging to the subclade S1 Description of Leptospira kemamanensis sp ., nov ., : ke . ma . man . ensis ., N . L . fem ., adj . kemamanensis of the distr

Introduction, Methods, Results, Discussion

The causative agents of leptospirosis are responsible for an emerging zoonotic disease worldwide ., One of the major routes of transmission for leptospirosis is the natural environment contaminated with the urine of a wide range of reservoir animals ., Soils and surface waters also host a high diversity of non-pathogenic Leptospira and species for which the virulence status is not clearly established ., The genus Leptospira is currently divided into 35 species classified into three phylogenetic clusters , which supposedly correlate with the virulence of the bacteria ., In this study , a total of 90 Leptospira strains isolated from different environments worldwide including Japan , Malaysia , New Caledonia , Algeria , mainland France , and the island of Mayotte in the Indian Ocean were sequenced ., A comparison of average nucleotide identity ( ANI ) values of genomes of the 90 isolates and representative genomes of known species revealed 30 new Leptospira species ., These data also supported the existence of two clades and 4 subclades ., To avoid classification that strongly implies assumption on the virulence status of the lineages , we called them P1 , P2 , S1 , S2 ., One of these subclades has not yet been described and is composed of Leptospira idonii and 4 novel species that are phylogenetically related to the saprophytes ., We then investigated genome diversity and evolutionary relationships among members of the genus Leptospira by studying the pangenome and core gene sets ., Our data enable the identification of genome features , genes and domains that are important for each subclade , thereby laying the foundation for refining the classification of this complex bacterial genus ., We also shed light on atypical genomic features of a group of species that includes the species often associated with human infection , suggesting a specific and ongoing evolution of this group of species that will require more attention ., In conclusion , we have uncovered a massive species diversity and revealed a novel subclade in environmental samples collected worldwide and we have redefined the classification of species in the genus ., The implication of several new potentially infectious Leptospira species for human and animal health remains to be determined but our data also provide new insights into the emergence of virulence in the pathogenic species .

Leptospirosis which is an emerging zoonotic disease worldwide , is transmitted to humans through contact with soils or surface waters contaminated with the urine of reservoir animals ., Species of Leptospira , which include infectious and non-infectious strains , are ubiquitous in the environment ., In this study we have sequenced the genomes of strains of Leptospira isolated from several environmental sources worldwide ., Comparison of these genomes with other members of the Leptospira genus revealed the existence of 30 novel Leptospira species ., A comparative genomic analysis of species of the genus allowed us to identify genes or genome features that are specific of infectious species , providing insights into virulence evolution in these atypical bacteria but also allow refinement of the Leptospira classification .

taxonomy, medicine and health sciences, leptospira, pathology and laboratory medicine, pathogens, tropical diseases, microbiology, bacterial diseases, phylogenetics, data management, phylogenetic analysis, genome analysis, neglected tropical diseases, bacteria, bacterial pathogens, infectious diseases, computer and information sciences, zoonoses, medical microbiology, microbial pathogens, comparative genomics, evolutionary systematics, leptospirosis, genetics, biology and life sciences, leptospira interrogans, genomics, evolutionary biology, computational biology, genomic medicine, organisms

journal.pcbi.1000736

Dynamics and Control of Diseases in Networks with Community Structure

Mitigating or preventing the spread of infectious diseases is the ultimate goal of infectious disease epidemiology , and understanding the dynamics of epidemics is an important tool to achieve this goal ., A rich body of research 1 , 2 , 3 has provided major insights into the processes that drive epidemics , and has been instrumental in developing strategies for control and eradication ., The structure of contact networks is crucial in explaining epidemiological patterns seen in the spread of directly transmissible diseases such as HIV/AIDS 1 , 4 , 5 , SARS 6 , 7 , influenza 8 , 9 , 10 , 11 etc ., For example , the basic reproductive number R0 , a quantity central to developing intervention measures or immunization programs , depends crucially on the variance of the distribution of contacts 1 , 12 , 13 , known as the network degree distribution ., Contact networks with fat-tailed degree distributions , for example , where a few individuals have an extraordinarily large number of contacts , result in a higher R0 than one would expect from contact networks with a uniform degree distribution , and the existence of highly connected individuals makes them an ideal target for control measures 7 , 14 ., While degree distributions have been studied extensively to understand their effect on epidemic dynamics , the community structure of networks has generally been ignored ., Despite the demonstration that social networks show significant community structure 15 , 16 , 17 , 18 , and that social processes such as homophily and transitivity result in highly clustered and modular networks 19 , the effect of such microstructures on epidemic dynamics has only recently started to be investigated ., Most initial work has focused on the effect of small cycles , predominantly in the context of clustering coefficients ( i . e . the fraction of closed triplets in a contact network ) 20 , 21 , 22 , 23 , 24 ., In this article , we aim to understand how community structure affects epidemic dynamics and control of infectious disease ., Community structure exists when connections between members of a group of nodes are more dense than connections between members of different groups of nodes 15 ., The terminology is relatively new in network analysis and recent algorithm development has greatly expanded our ability to detect sub-structuring in networks ., While there has been a recent explosion in interest and methodological development , the concept is an old one in the study of social networks where it is typically referred to as a “cohesive subgroups , ” groups of vertices in a graph that share connections with each other at a higher rate than with vertices outside the group 18 ., Empirical data on social structure suggests that community structuring is extensive in epidemiological contacts 25 , 26 , 27 relevant for infectious diseases transmitted by the respiratory or close-contact route ( e . g . influenza-like illnesses ) , and in social groups more generally 16 , 17 , 28 , 29 , 30 ., Similarly , the results of epidemic models of directly transmitted infections such as influenza are most consistent with the existence of such structure 8 , 9 , 11 , 31 , 32 , 33 ., Using both simulated and empirical social networks , we show how community structure affects the spread of diseases in networks , and specifically that these effects cannot be accounted for by the degree distribution alone ., The main goal of this study is to demonstrate how community structure affects epidemic dynamics , and what strategies are best applied to control epidemics in networks with community structure ., We generate networks computationally with community structure by creating small subnetworks of locally dense communities , which are then randomly connected to one another ., A particular feature of such networks is that the variance of their degree distribution is relatively low , and thus the spread of a disease is only marginally affected by it 34 ., Running standard susceptible-infected-resistant ( SIR ) epidemic simulations ( see Methods ) on these networks , we find that the average epidemic size , epidemic duration and the peak prevalence of the epidemic are strongly affected by a change in community structure connectivity that is independent of the overall degree distribution of the full network ( Figure 1 ) ., Note that the value range of Q shown in Figure 1 is in agreement with the value range of Q found in the empirical networks used further below , and that lower values of Q do not affect the results qualitatively ( see Suppl . Mat . Figure S1 ) ., Epidemics in populations with community structure show a distinct dynamical pattern depending on the extent of community structure ., In networks with strong community structure , an infected individual is more likely to infect members of the same community than members outside of the community ., Thus , in a network with strong community structure , local outbreaks may die out before spreading to other communities , or they may spread through various communities in an almost serial fashion , and large epidemics in populations with strong community structure may therefore last for a long time ., Correspondingly , the incidence rate can be very low , and the number of generations of infection transmission can be very high , compared to the explosive epidemics in populations with less community structure ( Figures 2a and 2b ) ., On average , epidemics in networks with strong community structure exhibit greater variance in final size ( Figures 2c and 2d ) , a greater number of small , local outbreaks that do not develop into a full epidemic , and a higher variance in the duration of an epidemic ., In order to halt or mitigate an epidemic , targeted immunization interventions or social distancing interventions aim to change the structure of the network of susceptible individuals in such a way as to make it harder for a pathogen to spread 35 ., In practice , the number of people to be removed from the susceptible class is often constrained for a number of reasons ( e . g . , due to limited vaccine supply or ethical concerns of social distancing measures ) ., From a network perspective , targeted immunization methods translate into indentifying which nodes should be removed from a network , a problem that has caught considerable attention ( see for example 36 and references therein ) ., Targeting highly connected individuals for immunization has been shown to be an effective strategy for epidemic control 7 , 14 ., However , in networks with strong community structure , this strategy may not be the most effective: some individuals connect to multiple communities ( so-called community bridges 37 ) and may thus be more important in spreading the disease than individuals with fewer inter-community connections , but this importance is not necessarily reflected in the degree ., Identification of community bridges can be achieved using the betweenness centrality measure 38 , defined as the fraction of shortest paths a node falls on ., While degree and betweenness centrality are often strongly positively correlated , the correlation between degree and betweenness centrality becomes weaker as community structure becomes stronger ( Figure 3 ) ., Thus , in networks with community structure , focusing on the degree alone carries the risk of missing some of the community bridges that are not highly connected ., Indeed , at a low vaccination coverage , an immunization strategy based on betweenness centrality results in fewer infected cases than an immunization strategy based on degree as the magnitude of community structure increases ( Figure 4a ) ., This observation is critical because the potential vaccination coverage for an emerging infection will typically be very low ., A third measure , random walk centrality , identifies target nodes by a random walk , counting how often a node is traversed by a random walk between two other nodes 39 ., The random walk centrality measure considers not only the shortest paths between pairs of nodes , but all paths between pairs of nodes , while still giving shorter paths more weight ., While infections are most likely to spread along the shortest paths between any two nodes , the cumulative contribution of other paths can still be important 40: immunization strategies based on random walk centrality result in the lowest number of infected cases at low vaccination coverage ( Figure 4b and 4c ) ., To test the efficiency of targeted immunization strategies on real networks , we used interaction data of individuals at five different universities in the US taken from a social network website 41 , and obtained the contact network relevant for directly transmissible diseases ( see Methods ) ., We find again that the overall most successful targeted immunization strategy is the one that identifies the targets based on random walk centrality ., Limited immunization based on random walk centrality significantly outperforms immunization based on degree especially when vaccination coverage is low ( Figure 5a ) ., In practice , identifying immunization targets may be impossible using such algorithms , because the structure of the contact network relevant for the spread of a directly transmissible disease is generally not known ., Thus , algorithms that are agnostic about the full network structure are necessary to identify target individuals ., The only algorithm we are aware of that is completely agnostic about the network structure network structure identifies target nodes by picking a random contact of a randomly chosen individual 42 ., Once such an acquaintance has been picked n times , it is immunized ., The acquaintance method has been shown to be able to identify some of the highly connected individuals , and thus approximates an immunization strategy that targets highly connected individuals ., We propose an alternative algorithm ( the so-called community bridge finder ( CBF ) algorithm , described in detail in the Methods ) that aims to identify community bridges connecting two groups of clustered nodes ., Briefly , starting from a random node , the algorithm follows a random path on the contact network , until it arrives at a node that does not connect back to more than one of the previously visited nodes on the random walk ., The basic goal of the CBF algorithm is to find nodes that connect to multiple communities - it does so based on the notion that the first node that does not connect back to previously visited nodes of the current random walk is likely to be part of a different community ., On all empirical and computationally generated networks tested , this algorithm performed mostly better , often equally well , and rarely worse than the alternative algorithm ., It is important to note a crucial difference between algorithms such as CBF ( henceforth called stochastic algorithms ) and algorithms such as those that calculate , for example , the betweenness centrality of nodes ( henceforth called deterministic algorithms ) ., A deterministic algorithm always needs the complete information about each node ( i . e . either the number or the identity of all connected nodes for each node in the network ) ., A comparison between algorithms is therefore of limited use if they are not of the same type as they have to work with different inputs ., Clearly , a deterministic algorithm with information on the full network structure as input should outperform a stochastic algorithm that is agnostic about the full network structure ., Thus , we will restrict our comparison of CBF to the acquaintance method since this is the only stochastic algorithm we are aware of the takes as input the same limited amount of local information ., In the computationally generated networks , CBF outperformed the acquaintance method in large areas of the parameter space ( Figure 4d ) ., It may seem unintuitive at first that the acquaintance method outperforms CBF at very high values of modularity , but one should keep in mind that epidemic sizes are very small in those extremely modular networks ( see Figure 1a ) because local outbreaks only rarely jump the community borders ., If outbreaks are mostly restricted to single communities , then CBF is not the optimal strategy because immunizing community bridges is useless; the acquaintance method may at least find some well connected nodes in each community and will thus perform slightly better in this extreme parameter space ., In empirical networks , CBF did particularly well on the network with the strongest community structure ( Oklahoma ) , especially in comparison to the similarly effective acquaintance method with n\u200a=\u200a2 ., ( Figure 5c ) ., As immunization strategies should be deployed as fast as possible , the speed at which a certain fraction of the network can be immunized is an additional important aspect ., We measured the speed of the algorithm as the number of nodes that the algorithm had to visit in order to achieve a certain vaccination coverage , and find that the CBF algorithm is faster than the similarly effective acquaintance method with n\u200a=\u200a2 at vaccination coverages <30% ( see Figure 6 ) ., A great number of infectious diseases of humans spread directly from one person to another person , and early work on the spread of such diseases has been based on the assumption that every infected individual is equally likely to transmit the disease to any susceptible individual in a population ., One of the most important consequences of incorporating network structure into epidemic models was the demonstration that heterogeneity in the number of contacts ( degree ) can strongly affect how R0 is calculated 12 , 13 , 34 ., Thus , the same disease can exhibit markedly different epidemic patterns simply due to differences in the degree distribution ., Our results extend this finding and show that even in networks with the same degree distribution , fundamentally different epidemic dynamics are expected to be observed due to different levels of community structure ., This finding is important for various reasons: first , community structure has been shown to be a crucial feature of social networks 15 , 16 , 17 , 19 , and its effect on disease spread is thus relevant to infectious disease dynamics ., Furthermore , it corroborates earlier suggestions that community structure affects the spread of disease , and is the first to clearly isolate this effect from effects due to variance in the degree distribution 43 ., Second , and consequently , data on the degree distribution of contact networks will not be sufficient to predict epidemic dynamics ., Third , the design of control strategies benefits from taking community structure into account ., An important caveat to mention is that community structure in the sense used throughout this paper ( i . e . measured as modularity Q ) does not take into account explicitly the extent to which communities overlap ., Such overlap is likely to play an important role in infectious disease dynamics , because people are members of multiple , potentially overlapping communities ( households , schools , workplaces etc . ) ., A strong overlap would likely be reflected in lower overall values for Q; however , the exact effect of community overlap on infectious disease dynamics remains to be investigated ., Identifying important nodes to affect diffusion on networks is a key question in network theory that pertains to a wide range of fields and is not limited to infectious disease dynamics only ., There are however two major issues associated with this problem:, ( i ) the structure of networks is often not known , and, ( ii ) many networks are too large to compute , for example , centrality measures efficiently ., Stochastic algorithms like the proposed CBF algorithm or the acquaintance method address both problems at once ., To what extent targeted immunization strategies can be implemented in a infectious diseases/public health setting based on practical and ethical considerations remains an open question ., This is true not only for the strategy based on the CBF algorithm , but for most strategies that are based on network properties ., As mentioned above , the contact networks relevant for the spread of infectious diseases are generally not known ., Stochastic algorithms such as the CBF or the acquaintance method are at least in principle applicable when data on network structure is lacking ., Community structure in host networks is not limited to human networks: Animal populations are often divided into subpopulations , connected by limited migration only 44 , 45 ., Targeted immunization of individuals connecting subpopulations has been shown to be an effective low-coverage immunization strategy for the conservation of endangered species 46 ., Under the assumption of homogenous mixing , the elimination of a disease requires an immunization coverage of at least 1-1/R0 1 but such coverage is often difficult or even impossible to achieve due to limited vaccine supply , logistical challenges or ethical concerns ., In the case of wildlife animals , high vaccination coverage is also problematic as vaccination interventions can be associated with substantial risks ., Little is known about the contact network structure in humans , let alone in wildlife , and progress should therefore be made on the development of immunization strategies that can deal with the absence of such data ., Stochastic algorithms such as the acquaintance method and the CBF method are first important steps in addressing the problem , but the large difference in efficacy between stochastic and deterministic algorithms demonstrates that there is still a long way to go ., To investigate the spread of an infectious disease on a contact network , we use the following methodology: Individuals in a population are represented as nodes in a network , and the edges between the nodes represent the contacts along which an infection can spread ., Contact networks are abstracted by undirected , unweighted graphs ( i . e . all contacts are reciprocal , and all contacts transmit an infection with the same probability ) ., Edges always link between two distinct nodes ( i . e . no self loops ) , and there must be maximally one edge between any single pair of nodes ( i . e no parallel edges ) ., Each node can be in one of three possible states: ( S ) usceptible , ( I ) nfected , or ( R ) esistant/immune ( as in standard SIR models ) ., Initially , all nodes are susceptible ., Simulations with immunization strategies implement those strategies before the first infection occurs ., Targeted nodes are chosen according to a given immunization algorithm ( see below ) until a desired immunization coverage of the population is achieved , and then their state is set to resistant ., After this initial set-up , a random susceptible node is chosen as patient zero , and its state is set to infected ., Then , during a number of time steps , the initial infection can spread through the network , and the simulation is halted once there are no further infected nodes ., At each time step ( the unit of time we use is one day , i . e . a time step is one day ) , an infected node can get infected with probability 1−exp ( −βi ) , where β is the transmission rate from an infected to a susceptible node , and i is the number of infected neighboring nodes ., At each time step , infected nodes recover at rate γ , i . e . the probability of recovery of an infected node per time step is γ ( unless noted otherwise , we use γ\u200a=\u200a0 . 2 ) ., If recovery occurs , the state of the recovered node is toggled from infected to resistant ., Unless mentioned otherwise , the transmission rate β is chosen such that R0∼ ( β/γ ) * d≈3 where d is the mean network degree , i . e the average number of contacts per node ., For the networks used here , this approximation is in line with the result from static network theory 47 , R0∼T ( <k2>/<k>−1 ) , where <k> and <k2> are the mean degree and mean square degree , respectively , and where T is the average probability of disease transmission from a node to a neighboring node , i . e . T≈β/γ ., Note that the variation in the degree is too small to be of relevance here ( see further below and Figure 1d ) ., The reason we chose γ\u200a=\u200a0 . 2 ( i . e . an average length of infectious period of 5 days ) and R0≈3 in most simulations ( unless mentioned otherwise ) is that these parameter values reflect , very roughly , some of the most widespread infectious diseases to which our study is relevant ( i . e . flu-like infectious diseases that are transmitted directly from person to person by the respiratory or close-contact route 8 , 9 , 48 , 49 , 50 ) ., After a simulation , we record the total number of cases infected ( the epidemic size ) , the maximum frequency of infection at any point during the simulation ( the peak prevalence ) , and the number of days that have passed between the first infected case and the simulation stop ( the duration of the epidemic ) ., In order to understand the effect of community structure , we generated networks with 2000 nodes from scratch with varying degrees of community structure ., The strength of community structure is generally measured as network modularity Q , which is defined aswhere eij is the fraction of all edges in the network that link nodes in community i to nodes in community j , and15 ., Thus , ai represents the fraction of edges in the network that connect to nodes in community i ., If edges were to fall between nodes without any regard for communities , we would have eij\u200a=\u200aai aj , and thus Q\u200a=\u200a0 ., There are numerous methods to calculate the value of Q for a given network , and the development of more accurate and efficient methods is still a very active research field ., In particular , one has to be careful when comparing values of Q because some measures are normalized while others or not 51 ., We have used the spin glass method introduced by Reichhardt and Bornholdt 52 to measure Q throughout this manuscript ., To generate networks with community structure , we initialize a network by creating 50 small-world communities ( as found in various social networks , see e . g . ref . 53 ) of 40 nodes using the Watts-Strogatz algorithm 54 such that each node has exactly 8 edges connecting to nodes of the same community ., We then add 2000 edges randomly between randomly chosen nodes , making sure that the edges fall between communities only ., Thus , we create a graph with 2000 nodes and 10000 ( i . e . ( 2000+50 * 40 * ( 8/2 ) ) ) undirected edges where one out of five edges falls between communities ., The average degree of the network is 10 , which is in line with recent reports on social contact patterns 55 ., Starting from this initial network where Q≈0 . 76 , we create networks with increasing community structure by rewiring between-community edges so that they become within-community edges ., More precisely , at each rewiring step , we, ( i ) randomly choose a between-community edge ,, ( ii ) randomly choose one of the two communities that the edge connects ,, ( iii ) pick a random node of the chosen community , and, ( iv ) rewire the edge by detaching it from the node of the community that was not chosen in step, ( ii ) , and attaching it to the new node in the community that was chosen in step, ( iii ) ., At all times , edges must always fall between two distinct nodes , and there can only be one edge between any two pair of nodes ., Weve also tested if all networks thus created consist of only a single connected component ( they do ) ., The quantity ( CV ) 2 is the square of the coefficient of variation in degree ( i . e . the squre of the ratio of the standard deviation of degree to the mean degree , where degree is defined as the number of edges incident to a node ) ., ( CV ) 2 is important for the spread of infectious diseases since it is known thatwhere ρ0 is the value of R0 under the assumption of a homogenous network ( i . e . no variance in the degree distribution ) 1 , 56 ., We used the network data collected on the social network website Facebook ( www . facebook . com ) by Traud et al . 41 ., The data contains the friendship network at five US universities , where nodes represent individuals ( i . e . members of the university ) , and edges represent friendship links between two individuals ., Additionally , the data includes covariate information ( if available ) about the individuals , such as the gender of the individual , the dormitory residence , major ( field of specialization ) etc ., While such friendship network data are interesting for various reasons , they do not necessarily reflect the contact network relevant for the spread of infectious diseases ., Clearly , a friendship connection between two individuals on a social network website does not necessarily mean that there is also a connection between the two individuals in the contact network relevant to the spread of infectious diseases ., Thus , in order to obtain contact network data that are relevant for the spread of infectious diseases transmitted directly from person to person by the respiratory or close-contact route , we make the following assumptions: Individuals who have a friendship relation in the network , and who either, ( a ) have the same dormitory residence , or, ( b ) who major in the same field and the same class year , are likely to be in close enough physical contact on a regular basis as to be able to transmit an infection to each other ., Thus , using the raw friendship data and the available information on dormitory residence , major , and class year , we extract the subgraph which reflects our assumptions ., Having extracted the subgraph , we remove all nodes who are not part of the largest connected component ( i . e . small subgraphs that are not part of the larger network ) ., The networks thus reduce to the following contact networks: We note that the modularity Q of these networks is within the range of modularities measured in the computationally generated networks ( see for example Figure 1 ) , with the exception of one network ( Georgetown ) ., Clearly , these networks will contain contacts that are not relevant for the spread of diseases ( false positives ) - at the same time , they will also miss some relevant contacts ( false negatives ) ., However , given the accuracy and amount of data , these networks are well suited to study the spread of infectious diseases on human contact networks , in particular for diseases transmitted directly from person to person by the respiratory or close-contact route ., Degree distributions of these networks are shown in Suppl ., Mat ., Figure S2 ., The algorithms used to identify nodes can be divided into two classes: deterministic and stochastic algorithms ., Deterministic algorithms require the complete information about each node ( i . e . either the number or the identity of all connected nodes for each node in the network ) , and they rank nodes by processing that information by a procedure specific to that algorithm ., Target nodes are then chosen by their rank ( from high to low ) ., Thus , for a given network structure , deterministic algorithms always give the same result , i . e . they identify the same target nodes ( except for random choices when two nodes have exactly the same rank ) ., Stochastic algorithms , on the other hand , do not require such detailed structural information - they identify target nodes by collecting information locally from randomly chosen nodes in the network ., These algorithms represent the type of investigation-related information in actual epidemics ., We will now describe a number of deterministic and stochastic algorithms as we have used them in the main text .

Introduction, Results, Discussion, Methods

The dynamics of infectious diseases spread via direct person-to-person transmission ( such as influenza , smallpox , HIV/AIDS , etc . ) depends on the underlying host contact network ., Human contact networks exhibit strong community structure ., Understanding how such community structure affects epidemics may provide insights for preventing the spread of disease between communities by changing the structure of the contact network through pharmaceutical or non-pharmaceutical interventions ., We use empirical and simulated networks to investigate the spread of disease in networks with community structure ., We find that community structure has a major impact on disease dynamics , and we show that in networks with strong community structure , immunization interventions targeted at individuals bridging communities are more effective than those simply targeting highly connected individuals ., Because the structure of relevant contact networks is generally not known , and vaccine supply is often limited , there is great need for efficient vaccination algorithms that do not require full knowledge of the network ., We developed an algorithm that acts only on locally available network information and is able to quickly identify targets for successful immunization intervention ., The algorithm generally outperforms existing algorithms when vaccine supply is limited , particularly in networks with strong community structure ., Understanding the spread of infectious diseases and designing optimal control strategies is a major goal of public health ., Social networks show marked patterns of community structure , and our results , based on empirical and simulated data , demonstrate that community structure strongly affects disease dynamics ., These results have implications for the design of control strategies .

Understanding the spread of infectious diseases in populations is key to controlling them ., Computational simulations of epidemics provide a valuable tool for the study of the dynamics of epidemics ., In such simulations , populations are represented by networks , where hosts and their interactions among each other are represented by nodes and edges ., In the past few years , it has become clear that many human social networks have a very remarkable property: they all exhibit strong community structure ., A network with strong community structure consists of smaller sub-networks ( the communities ) that have many connections within them , but only few between them ., Here we use both data from social networking websites and computer generated networks to study the effect of community structure on epidemic spread ., We find that community structure not only affects the dynamics of epidemics in networks , but that it also has implications for how networks can be protected from large-scale epidemics .

infectious diseases/epidemiology and control of infectious diseases

journal.pgen.1000090

Combined Analysis of Murine and Human Microarrays and ChIP Analysis Reveals Genes Associated with the Ability of MYC To Maintain Tumorigenesis

Overexpression of MYC is one of the most frequent events in human tumorigenesis 1 ., MYC overexpression is thought to induce tumorigenesis by causing inappropriate gene expression resulting in autonomous cellular growth , proliferation , and the inhibition of cellular differentiation 2 , 3 ., Many laboratories have conditionally overexpressed c-MYC ( MYC ) utilizing conditional transgenic model systems 4–9 ., In these models , the suppression of MYC led to permanent loss of tumorigenesis through proliferative arrest , differentiation and/or apoptosis 4–6 , 10 ., In some circumstances , even the brief suppression of MYC overexpression permanently prevents its ability to sustain tumorigenesis 6 ., These and other observations have suggested the possibility that oncogenes such as MYC exhibit the phenomena of oncogene addiction 11 ., However , the molecular basis of oncogene addiction is not clear ., Recently , we have suggested that cellular senescence , which involves chromatin modifications and heterochromatin formation 12 , 13 , may be an important mechanism for sustained tumor regression upon MYC inactivation 14 ., MYC is thought to play a role in the regulation of up to 15% of genes in the fly , mouse or human 3 , 15 , 16 ., Thus , it seems likely that changes in gene expression programs , rather than individual genes , account for the phenotypic consequences of MYC inactivation ., Consistent with this notion , MYC has recently been shown to globally influence chromatin structure through histone modifications 17–19 ., Similarly , N-MYC was shown to globally regulate acetylation and methylation of histone molecules 20 ., We have reported that MYC inactivation in tumors induces specific global changes in histone modification 14 ., Although many MYC target genes have been identified in various cells or tissue contexts ( summarized in http://www . myc-cancer-gene . org ) , it is hard to discern which of the many of MYC targets are associated with the ability of MYC to initiate and/or sustain tumorigenesis ., Many previous studies have examined changes in gene expression associated with the induction of MYC expression in cells 3 , 15 , 21–26 ., Other groups have performed comparative analysis of gene expression profiles between murine constitutive MYC-induced tumors and human tumors in liver and prostate cancers 27 , 28 ., Both of these analyses identified similarities in gene expression between MYC-induced tumor models and human tumors ., Although revealing , these studies would not necessarily identify gene products that are responsible for the ability of MYC to induce tumorigenesis ., We speculated that by analyzing gene expression profiles in tumors generated from conditional transgenic models would allow us to identify gene expression signature specifically associated with the ability of MYC to initiate and maintain tumorigenesis ., We performed microarrays on mRNA samples from a time-course experiment with MYC inactivated and then reactivated in osteosarcoma ., The expression data was then examined using the StepMiner algorithm 29 to generate a list of genes associated with MYC-induced tumorigenesis in osteosarcomas ( Figure 1 ) ., The StepMiner algorithm analyzes microarray time courses by identifying genes that undergo abrupt transitions in expression level , and the time at which the transitions occur ., Importantly , by ChIP we were able to demonstrate that permanent changes in gene expression were frequently associated with measurable alterations in the ability of MYC to bind to the promoter regions of these genes in osteosarcomas ., Furthermore , gene expression profiles were compared between osteosarcomas and the previously published MYC conditional pancreatic tumor 25 to generate a common gene signature associated with MYC-induced tumorigenesis in mice ( Figure 1 ) ., Finally , Boolean analysis was used to further examine the correlation between levels of expression of this identified subset of genes among the published dataset of 7 , 171 human microarrays in U133A format ., From this analysis , we were able to deduce a list of genes strongly correlated with the ability of MYC to maintain tumorigenesis ., The MYC induced osteosarcoma derived cell line , 1325 , was grown in vitro 6 and treated with 20ng/ml of doxycycline in complete DMEM medium for various length of time to inactivate MYC expression ., To reactivate MYC expression , doxycyline was removed by rinsing the bone tumor cells with an excess amount of PBS ., mRNA was collected from bone tumor cells treated with doxycycline for 0 , 4 , 8 , 12 , 18 , 24 , 36 , and 48 hours , and after removal of doxycycline for 4 , 8 , 12 , 18 , 24 , 36 , and 48 hours ., MYC levels were greatly reduced as early as 4 hours after doxycycline treatment ( Figure 2A and Figure S1 ) ., We confirmed that the expression of MYC could be reactivated to a level similar to that of MYC-on tumors by thoroughly washing the cells with PBS ( Figure 2A and Figure S1 ) ., cDNA microarray analysis was performed on the RNA samples prepared from tumors in which MYC was inactivated and reactivated for different lengths of time ., StepMiner analysis ( Figure 2B , 2C ) 29 was applied to this time-course microarray experiment to identify changes in gene expression at discrete time points before and after MYC inactivation and reactivation ., StepMiner fits step functions to the data points using an adaptive regression scheme and identifies time points at which a gene is significantly induced or repressed ., Examples of one-step expression pattern are illustrated in Figure 2C ., Recently , we have shown that MYC inactivation generally induces cellular senescence in several tumor models 14 ., Therefore , we specifically examined if the expression of senescence associated genes changed upon MYC inactivation in osteosarcomas ., Indeed , we did find that senescence associated genes such as p15INK4b , p21CIP , PCNA , MCM3 , CYCLIN A 12 , 30 , 31 were up-regulated or down-regulated upon MYC inactivation ( Figure S2 and Table S1 ) ., Thus , our results support the notion that MYC inactivation is inducing changes in gene expression that is associated with cellular senescence ., Generally , analysis of gene expression changes after StepMiner analysis revealed four discrete patterns of changes in gene expression upon MYC inactivation and reactivation: Permanently Repressed ( PR ) , Permanently Induced ( PI ) , Reversibly Repressed ( RR ) and Reversibly Induced ( RI ) ., For this analysis , we set p<0 . 01 as a cutoff for statistically significant changes in gene expression ., We identified 1016 unique probes in the PR group , 1777 unique probes in the PI group , 1148 unique probes in the RI group , and 1167 unique probes in the RR group ( Figure 3 and Tables S2 for lists of genes ) ., Based upon our previously published observation that even brief inactivation of MYC can result in the sustained loss of the neoplastic properties of MYC-induced osteosarcomas 6 , we speculated that genes which are potentially important for sustained tumorigenesis would be permanently repressed or induced ( e . g . the PR group or the PI group ) upon MYC inactivation ., To identify associated functional activities associated with the PR and PI groups of genes , we applied Gene Ontology analysis ( GO Term analysis ) to the list of genes generated above ., Biological functions that were identified for each step upon MYC inactivation are listed ( Table S3 , S4 , and S5 ) ., Associated functions identified include gene products known to regulate metabolism , biosynthesis of nucleotides and proteins and genes involved in the regulation or function of ribonucleoprotein complexes ., Notably , MYC has been shown to regulate expression of ribosomal structure proteins and ribosomal RNAs 18 , 32 ., Hence , it is striking that the mRNA expression of 61 ribosomal structural proteins out of 82 ribosomal structural protein genes was decreased upon MYC inactivation and further decreased upon MYC reactivation in bone tumor ( see Figure 4A and Table S3 for results of GO term analysis ) ., To validate that these genes expression did change , we performed quantitative real-time PCR of 11 ribosomal structural proteins in osteosarcomas ( Figure 4B ) ., Moreover , we found that the same ribosomal structural proteins also changed upon MYC inactivation in our conditional model of lymphomas 4 ( Figure 4B ) ., We then examined if the decreased expression of ribosomal structural proteins associated with changes in rate of protein synthesis ., We found that the protein synthesis rates were decreased in both bone tumor and lymphomas upon MYC inactivation ( Figure 4C ) ., Furthermore , the protein synthesis rate remained lowed upon MYC reactivation in bone tumor ( Figure 4C ) ., MYC has been shown to regulate the gene expression of a multitude of genes 3 , 15 , 21–26 , 33–35 ., To examine if these genes changed in gene expression upon MYC inactivation and reactivation , we used two approaches ., First , we retrieved the mouse homologs of MYC target genes listed in www . myc-cancer-gene . org , a collection of most of the published MYC target genes in different organisms and tissues ( total of 1697 MYC targets ) 16 ., In osteosarcomas , 71 of the published MYC targets are permanently induced and 52 of the published MYC targets are permanently repressed upon MYC inactivation and reactivation ( p<0 . 01 ) ( Figure 5 , see PR and PI ) ., Second , we examined direct MYC target genes identified as defined by several recent publications 33–35 ., Interestingly , only 7–11% of these identified direct MYC target genes exhibited sustained changes upon MYC inactivation in osteosarcoma ( Figure 6 and Table S6 ) ., An important recent report suggests that MYC binding to promoters is regulated by the chromatin structure at these gene loci 36 ., Recently , we have shown that MYC inactivation is associated with global changes in chromatin structures 14 ., Thus , it seemed that a possible explanation for the permanent changes in gene expression that we observed ( Figure 3 ) is that the ability of MYC to bind to specific gene products is perturbed by changes in chromatin structure ., To address this possibility directly , we used ChIP to examine MYC binding to E-box sequences of target genes in MYC activated and MYC reactivated conditions for osteosarcomas ., We specifically examine three groups of genes: the ribosomal structural proteins ( Figure 4 ) , the PR group ( Figure 6 , 35 ) and the RR group genes that were identified previously as direct MYC targets before ( Figure 6 , 35 ) ., A total of 168 E-box regions were examined by ChIP ., As a control , we performed ChIP for osteosarcoma in the MYC OFF condition ( Table S7 ) ., Binding of MYC to E-box regions is shown as the percentage of DNA brought down by ChIP for the MYC ON versus the MYC reactivated conditions ( Figure 7 ) ., Note , that upon MYC reactivation the majority of ribosomal structural genes exhibited decreased MYC binding to E-boxes relative to the MYC ON condition ( 31 out of 41 data points fall below the line of X\u200a=\u200aY , p-value\u200a=\u200a4 . 34×10−4 ) ., Similarly , the majority of the genes with the PR pattern of gene expression exhibited a significant decrease of MYC binding to E-boxes relative to the MYC ON condition ( 42 out of 60 data points fall below the line of X\u200a=\u200aY , p-value\u200a=\u200a0 . 0016 ) when MYC was reactivated ( Figure 7 and Table S7 ) ., In contrast , the group of genes that exhibited the RR pattern of gene expression exhibited no particular increase or decrease in MYC binding to E-boxes compared with the MYC ON condition ( 33 out of 67 data points fall below the line of X\u200a=\u200aY , p-value\u200a=\u200a0 . 4 ) ., Our results support the possibility that the permanent changes in gene expression upon MYC inactivation can be explained in many cases because of a change in the ability of MYC to bind to specific promoter loci ., To determine if the gene signature we identified would also be seen in another tumor model system , we compared our microarray data from MYC-induced osteosarcoma with a previously reported microarray data set from a MYC-induced pancreatic tumor model to identify a common expression signature for MYC-induced tumorigenesis 25 ., In the published report , MYC-ERTAM was expressed specifically in β-cell pancreatic tissues with MYC-on for 2 , 4 , 8 , 24 hours , and 21 days ( referred as tumorigenesis arrays in the published paper ) , and MYC off in pancreatic tumors for 2 , 4 , and 6 days ( referred as tumor regression arrays in the published paper ) ., MYC activation induced pancreatic tumors and MYC inactivation resulted in tumor regression through apoptosis 7 ., cDNA from these samples was applied to oligo arrays from Affymetrix 25 ., As previously suggested in the paper , we assumed that genes were induced ( repressed ) upon MYC activation and repressed ( induced ) upon MYC inactivation were potentially important for MYC induced tumorigenesis ., We first used the StepMiner algorithm was applied to the raw data generated from these published experiments to obtain lists of genes that increase ( or decrease ) in expression upon tumorigenesis and decrease ( or increase ) in expression upon tumor regression ( Figure 8 and Table S8 ) ., After StepMiner analysis , 196 and 65 unique probes were identified as induced and repressed genes respectively , which are associated with MYC-induced tumorigenesis ., The osteosarcoma data set was filtered via the induced gene list or the repressed gene list generated from the pancreatic tumors ., Then , we applied StepMiner analysis to identify genes that are permanently repressed or permanently induced with a p-value<0 . 01 ., By comparing microarray data from two independent MYC conditional tumor models , we found a common gene signature with 42 genes associated with MYC-induced tumorigenesis ( Figure 9 ) ., Among the list of genes , there are 34 unique genes positively correlating with MYC-induced tumorigenesis and 8 unique genes negatively correlating with MYC-induced tumorigenesis in mice ( Figure 9 ) ., MYC overexpression has been implicated in the pathogenesis of many types of human cancer , in particular , hematopoietic tumors 1 ., To see if the gene signature we defined in murine tumor models was predictive of genes whose expression was strongly correlated with MYC between MYC and human homologs in human lymphomas , we retrieved all publicly available human microarrays ( n\u200a=\u200a7 , 171 ) in Affymatrix U133A platform ., Then , we classified the expression level of each gene on each array as “low” or “high” relative to a threshold using Boolean analysis ( 29 and Sahoo et al . RECOMB 2007 in press , see Figure 10A ) ., We found that MYC expression is “high” in human lymphomas ( 204 out of 221 lymphoma cases ignoring the “intermediate” values , see Figure 10B ) ., Figure 10B shows the gene expression scatter plot of MYC and RPS2 , which are both highly expressed in lymphoma arrays ( total of 273 lymphoma microarrays are highlighted with red color ) ., We then examined to see if the expression of MYC-associated genes identified above ( Figure 4 and 9 ) are “high” or “low” in more than 95% of the lymphoma microarrays ., The Boolean analysis identified that the expression of both small and large ribosomal structural proteins is high in human lymphomas ( Figures S3 and S4 ) as was observed in murine osteosarcomas and lymphomas ( Figure 4 ) ., We further investigated if the expression of human homologs of the common gene signature from the murine microarray data is “high” or “low” in human lymphomas ., 63 unique probes from the induced list ( Figure 9 ) and 9 probes from the repressed list ( Figure 9 ) were found in the U133A format ( see Table S9 ) ., We found 14 out of 63 probes correlated with the human arrays ., Genes whose expression was “high” in more than 95% of human lymphomas , whose gene names include: BZW2 , H2AFY , SFRS3 , NAP1L1 , NOLA2 , UBE2D2 and CCNG1 ( p\u200a=\u200a4 . 07×10−5 , Figure 11 ) ., From the repressed list of genes 4 out of 9 probes had low expression in more than 95% of the human lymphomas , whose gene names include LIFR , FABP3 and EDG1/HEXIM1 ( p\u200a=\u200a0 . 03 , Figure 11 ) ., We have listed al the genes identified and their associated functions ( listed in the Swiss-Prot data base ) ( Figure 11 ) ., Many of these genes have functions that could account for MYC activity ., Notably , CCNG1 , LIFR and EDG1/HEXIM1 are involved in cell cycle or signaling pathways ., H2AFY and NAP1l1 are involved in modulating chromatin structures ., SFRS3 and NOLA2 are involved in mRNA and rRNA processing ., BZW2 , UBE2D2 and FABP3 are involved in metabolism such as protein or fatty acid synthesis ., Finally , we validated our results obtained by microarray analysis through quantitative real-time PCR ( Figure S5 ) ., Moreover , we found that these identified genes exhibited similar patterns of changes in gene expression upon MYC inactivation in our model of MYC-induced lymphoma ( Figure S6 ) ., Therefore , we have identified a subset of MYC regulated gene products that are highly correlated with the ability of MYC to maintain tumorigenesis ., MYC target genes have been implicated in a multitude of biological functions 16 ., Many additional potential MYC targets have been identified through microarray analysis 3 , 15 , 21–26 , 37 ., However , it has not been easy to discern which if any of these genes are involved in the ability of MYC to initiate or maintain tumorigenesis ., We have combined microarray analysis of two conditional transgenic model systems and a human comparative Boolean analysis to determine which of these identified genes most strongly correlated with MYC expression from total of 273 datasets of human lymphoma microarrays in U133A format ., We also utilized ChIP to demonstrate that a large number of the genes that were permanently suppressed upon MYC inactivation exhibited changes in the ability of MYC to bind to their promoter loci ., Thus , we identified a gene signature strongly correlated with the ability of MYC to maintain tumorigenesis ., Our results have possible implications for why MYC induces tumorigenesis in specific cellular contexts ., To identify this gene signature , we utilized our conditional transgenic model system of MYC-induced osteosarcoma in which we have previously shown that upon MYC inactivation tumors permanently lost the ability of MYC to induce tumorigenesis 6 ., Thereby , we defined an initial gene signature consisting of 2 , 793 unique probe sets of genes that included genes whose expression was permanently changed ( Figure 3 ) ., This gene signature includes gene products that have been already implicated as MYC targets ( Figure 5 ) ., Most notably , ribosomal structural proteins were strongly correlated with MYC-induced tumorigenesis in murine osteosarcomas , lymphomas ( Shachaf CM et . al . submitted ) and in human lymphomas ., These results suggest that the ability of MYC to induce ribosomal gene products is important to its ability to initiate and maintain tumorigenesis ., Our results are consistent with a multitude of evidence suggesting that MYC can regulate ribosomal gene expression 15 ., In Drosophila , the biological connection of MYC and ribosomal structural proteins can also be seen in the small cell-size phenotypes of both MYC mutants and ribosomal structural protein genes mutants 38–40 ., MYC globally regulates protein synthesis through regulating expression of ribosomal RNAs , tRNAs , RNA helicases , and translation elongation factors 18 , 41 ., Notably , it had been shown that rate of protein synthesis was increased 3-fold in MYC-overexpressing fibroblasts compared to MYC knockout fibroblasts 42 ., We confirmed that the inactivation of MYC in tumor cells resulted in a reduction of both ribosomal protein gene expression and rate of protein synthesis in murine tumor models ( Figure 4 ) ., Ribosomal genes could play important function in influencing protein translation and thus in this manner influence the ability of MYC to function as an oncogene ., In this regard , it is notable that a recent study in Zebra fish identified some ribosomal protein genes as tumor-suppressors 43 ., Nevertheless , it is not clear how ribosomal structural protein genes function as tumor-suppressors during tumorigenesis ., Interestingly , changes in the gene expression of ribosomal structural proteins , although observed in both our model of MYC induced osteosarcoma and lymphoma , were not seen in a model of pancreatic islet cell tumors ( Figure 4 , 8 , and 25 ) ., Thus , it is possible that ribosomal protein genes expression play a role MYC-induced tumorigenesis only in specific types of cancer ., We are reassured of the likely importance of ribosomal gene products in MYC associated tumorigenesis for we were able to confirm that MYC and ribosomal structural proteins are highly correlated in human lymphomas ( Figure S4 and S5 ) ., It remains to be directly determined if these ribosomal genes are playing a role in MYC induced tumorigenesis ., Genes that we identified as most strongly correlated with MYC-induced tumorigenesis ( Figure 9 ) in mice are involved in diverse biological processes such as transcription regulation , RNA processing , proliferation , fatty acid transport and cell signaling ( Figure 11 ) ., Furthermore , some of the genes identified have been previously implicated in tumors or oncogenic signaling pathways ., BLMH has been previously shown to be a MYC target 44 ., UBE2d2 has been implicated as a target of the WNT signaling pathway in a microarray experiment 45 ., NAP1l1 has been shown to be a tumor marker for colon cancer 46 ., TRIP13 expression was highly elevated in tumor tissues 47 ., Altered regulation of CCNG1 has been observed in breast cancer 48 ., High expression of NOLA2 has been seen in squamous cell lung cancer 49 ., Interestingly , anti-tumor effects have been observed for genes with expression reversely correlated with MYC ., FABP3 has been proposed as tumor suppressor in breast cancer 50 ., EDG1 has been shown to be an inhibitor for breast cancer growth 51 ., Our data now suggest that BZW2 , H2AFY and SFRS3 , which function in translation initiation 52 , chromatin structure 53 , and mRNA splicing 54 , respectively , may also be involved in tumorigenesis ., We were able to utilize our MYC conditional tumor models as tools to uncover genes that are strongly correlated with tumor maintenance ., However , we recognize that it is very unlikely that any of the individual genes we identified are sufficient alone to explain the ability of MYC to initiate or maintain tumorigenesis ., Rather it is highly likely that it is a constellation of gene expression changes that are responsible for the ability of MYC to maintain tumorigenesis ., We can now offer a possible explanation for why the brief inactivation of MYC can result in the permanent loss of the ability of MYC to sustain tumorigenesis 6 ., MYC inactivation appears to result in permanent changes in the ability of MYC to function as a transcription factor ( Figure 12 ) ., Recently , we have shown that MYC inactivation induced chromatin modifications associated with cellular senescence 14 ., The particular structural state of chromatin has been shown to influence the ability of MYC to bind to specific promoter loci 36 ., Indeed , our results illustrate that upon MYC inactivation there were permanent changes in the ability of MYC to bind to the promoters of specific gene loci ( Figure 7 ) ., It remains to be determined the mechanism of these changes in chromatin structure ., One possibility is that MYC itself is contributing to changes in chromatin structure through global changes in chromatin modifications , which seems an attractive possibility based upon the work from many laboratories 14 , 20 , 55 ., Regardless of the mechanism , our results point to the fact that the genes that MYC can regulate are different in different cellular contexts and that this appears to have a direct bearing on when MYC overexpression results in a neoplastic phenotype ., We note that we could not explain all of the permanent changes in gene expression based upon differences in MYC binding to promoter loci ., Thus , it is likely there are additional mechanisms by which MYCs ability to regulate gene expression has been altered ., One of the biggest challenges in understanding how MYC contributes to tumorigenesis has been to address the conundrum that MYC has both direct and indirect influence on the expression of so many different genes and these genes are involved in a multitude of biologic functions ., Many of these genes may not be relevant to how MYC overexpression contributes to tumorigenesis ., Here we have illustrated by using a defined transgenic mouse model that exhibits conditional tumorigenesis such that upon MYC inactivation tumor cells permanently loses a neoplastic phenotype that we can define a specific gene list that is specifically correlated with MYCs ability to maintain tumorigenesis ., To perform this analysis we combined two novel methods of gene expression analysis , the StepMiner and the Boolean analysis , as a powerful strategy to perform an unbiased comparative analysis of microarray data from conditional MYC-induced tumor models and all the available published human data with Affymetrix U133A format ., Our strategy may be generally useful for the identification of gene signatures associated with the ability of specific oncogenes to initiate and sustain tumorigenesis and the identification of potential new therapeutic targets for the treatment of cancer ., Osteosarcoma-derived cell line 1325 6 were cultured with DMEM medium supplemented with 10% FBS , 1% Pen/Strep , L-Glutamine , and non-essential amino acids ( Invitrogen ) ., Lymphomas were cultured with RPMI medium supplemented with 10% FBS , 1% Pen/Strep , L-Glutamine and 3 . 96×10−4% of 2-mercaptoethanol ( Sigma ) ., 20ng/ml of doxycycline was added to the medium for inactivating MYC expression ., Seven times , each time with 20 mls of PBS , was applied to cells to completely remove doxycycline in the medium ., For rate of protein synthesis , lymphoma-derived cell line 6780 14 or bone tumor cell line 1325 grown in complete medium with or without doxycycline were rinsed with PBS and then replenished with DMEM ( with or without doxycyline ) without methionine and cysteine ( Invitrogen ) , containing 10% dialyzed fetal calf serum ( Invitrogen ) , 1% Pen/Strep , L-Glutamine ., One hour later , cells were labeled with 30 μCi of EXPRE35S35S ( PerkinElmer ) per plate for 60 minutes and then washed with PBS ., Cells were lysed and TCA precipitation was applied to determine the incorporation of radiolabeled amino acids ., Aliquots of cell lysate were used for protein determination by DC Protein Assay ( Bio-Rad ) ., The protein synthesis rate was calculated as TCA-precipitable counts per minute divided by micrograms of protein in the same sample ., cDNA were synthesized by Superscript II ( invitrogen ) followed by manufactures protocol ., Real-time PCR for human c-MYC ( probes and primers from Applied Biosystems ) and mouse GAPDH 56 were performed in ABI PRIZM analyzer ., Sequences for primers for quantitative real-time are listed in Table S8 ., Mouse cDNA microarrays were produced at Stanford Functional Genomic Facility ., cDNA labeling and hybridization were followed as previously described 57 ., Briefly , mRNA from bone tumor cells were extracted by Trizol ( Invitrogen ) based on the protocol provided by the manufacturer ., 30 μg of total RNA from bone tumor and reference RNA generated by pooling RNA from various mouse tissues were used for each microarray experiment ., cDNA from bone tumor cells was labeled with Cy5-dUTP and reference cDNA was labeled with by Cy3-dUTP ( Amershan ) after reverse-transcription ., Labeled cDNAs were concentrated by Microcon YM-30 ( Millipore ) before hybridizing with microarrays for 16 hours at 65°C ., After hybridization , microarrays were washed and spin dry before scanned on the GenePix 40000B Array Scanner ( Axon ) ., Raw array images were analyzed using the GenePix 5 software ( Axon ) ., Microarray data was then submitted to the Stanford Microarray Database ( SMD ) for normalization ., Data after normalization was then applied with the StepMiner algorithm to identify changes in gene expression ., The StepMiner fits step functions to time-course microarray data and provides a statistical measure of the goodness of fit 29 ., The steps are placed between time points at the sharpest change between low expression and high expression levels , which gives insight into the timing of the gene expression-switching event ., Mathematically , steps are placed at a position that minimizes the sum of square error and an F-statistic with appropriate degrees of freedom is used to produce a p-value for the goodness of fit ., The StepMiner automatically characterizes the genes in to five different groups: Up , Down , Up-Down , Down-Up and Other 29 ., The genes are primarily sorted in ascending order according to the timing of their change and secondarily sorted in ascending order according to their p-values ., ChIP was performed based on the protocol provided in the kit with some modifications ( ChIP assay kit by Upstate Biotech ) ., Briefly , bone tumor cells were grown on the condition described above with ( MYC OFF and MYC reactivated conditions ) or without ( MYC ON condition ) doxycycline ( 20ng/ml ) ., 48 hours treated with doxycycline , cells were either harvested ( as MYC OFF condition ) or extensively washed with PBS ( see above ) to remove doxycycline in the medium ., 48 hours after washing , cells were harvested ( as MYC reactivated ) ., Formaldehyde ( Fisher ) was added to the medium to a final concentration of 1% for cross-linking at 37°C for 10 minutes ., Cross-linking was stopped by adding glycine to a final concentration of 0 . 125M ., Cells were washed with cold PBS containing protease inhibitors ( 1mM PMSF , 1 μg/ml aprotinin and 1 μg/ml pepstatin A ) and pelleted by centrifugation ., Cell pellets were then lysed in SDS lysis buffer ( 1% SDS , 10mM EDTA , 50mM Tris , pH 8 . 1 , with proteases inhibitors mentioned above ) ., Cells were sonicated with a Branson 250 sonicator at a power setting of 3 for 3 times with 10 sec for each sonication and the cells were cooled down with ice for 1 min between each sonication ., This condition of sonication yielded genomic DNA fragments with a size about 100–600 base pairs ., Samples were then immunoprecipitated with c-MYC antibody ( 2 μg of N262 from Santa Cruz Biotech ) followed the protocol provided by the kit ( Upstate Biotech ) ., DNA samples from the ChIP experiments were applied for quantification by Real-time PCR ( ABI PRISM 7900 HT ) with SYBR green ., Promoter sequences ( −2000 to +2000 relative to the transcription start sites ) of murine MYC targets were retrieved from UCSC genome browser and primers flanking the E-box were designed by Primer3 ( http://frodo . wi . mit . edu/ ) ( Table S10 ) ., Data from 7 , 171 publicly available raw Affymetrix U133A human microarrays were collected from the Gene Expression Omnibus ( GEO ) 58 and normalized together using the RMA algorithm 59 , 60 ., Thresholds were assigned for each probe set by first sorting the expression values for that probe set on all arrays in ascending order , and then fitting a step function to the data using the StepMiner ., This approach places the threshold cutoff at the largest jump from low values to high values ., In the case where the gene expression levels are evenly distributed from low to high , the threshold cutoff tends to be near the mean expression level ., If the assigned cutoff for a gene is t , expression levels above t + 0 . 5 are classified as “high , ” expression levels below t−0 . 5 are classified as “low , ” and values between t −0 . 5 and t+0 . 5 are classified as “interm

Introduction, Results, Discussion, Materials and Methods

The MYC oncogene has been implicated in the regulation of up to thousands of genes involved in many cellular programs including proliferation , growth , differentiation , self-renewal , and apoptosis ., MYC is thought to induce cancer through an exaggerated effect on these physiologic programs ., Which of these genes are responsible for the ability of MYC to initiate and/or maintain tumorigenesis is not clear ., Previously , we have shown that upon brief MYC inactivation , some tumors undergo sustained regression ., Here we demonstrate that upon MYC inactivation there are global permanent changes in gene expression detected by microarray analysis ., By applying StepMiner analysis , we identified genes whose expression most strongly correlated with the ability of MYC to induce a neoplastic state ., Notably , genes were identified that exhibited permanent changes in mRNA expression upon MYC inactivation ., Importantly , permanent changes in gene expression could be shown by chromatin immunoprecipitation ( ChIP ) to be associated with permanent changes in the ability of MYC to bind to the promoter regions ., Our list of candidate genes associated with tumor maintenance was further refined by comparing our analysis with other published results to generate a gene signature associated with MYC-induced tumorigenesis in mice ., To validate the role of gene signatures associated with MYC in human tumorigenesis , we examined the expression of human homologs in 273 published human lymphoma microarray datasets in Affymetrix U133A format ., One large functional group of these genes included the ribosomal structural proteins ., In addition , we identified a group of genes involved in a diverse array of cellular functions including: BZW2 , H2AFY , SFRS3 , NAP1L1 , NOLA2 , UBE2D2 , CCNG1 , LIFR , FABP3 , and EDG1 ., Hence , through our analysis of gene expression in murine tumor models and human lymphomas , we have identified a novel gene signature correlated with the ability of MYC to maintain tumorigenesis .

The targeted inactivation of oncogenes may be a specific and effective treatment of cancer ., However , how oncogene inactivation leads to tumor regression is not clear ., Previously , we have shown that even the brief inactivation of the MYC oncogene can result in the sustained regression of at least some tumors ., To understand the mechanism , we have utilized several novel genomic analyses to define a set of genes that strongly correlate with the ability of the MYC oncogene to maintain tumorigenesis ., First , we generated a novel data set from microarray analyses of murine tumors that we analyzed by StepMiner to identify discrete step changes in gene expression after the inactivation or the reactivation of the MYC oncogene ., Second , we utilized Boolean Network Analysis to further define the subset of genes highly correlated with MYC in human tumorigenesis ., Third , we utilized ChIP analysis to demonstrate that in many cases the permanent changes of gene expression we uncovered were associated with changes in the ability of MYC to occupy the promoter locus ., Our general strategy could be similarly utilized in other experimental model systems to understand how specific oncogenes contribute to the maintenance of tumorigenesis .

genetics and genomics/disease models, genetics and genomics/cancer genetics

journal.pgen.1004398

Positive Feedback of NDT80 Expression Ensures Irreversible Meiotic Commitment in Budding Yeast

During gametogenesis , cells integrate external signals with internal cell-cycle control mechanisms to initiate and sustain meiosis , and eventually to differentiate into gametes ., Although the external signals that initiate the switch into meiosis in various organisms are quite diverse , many of the features of meiosis are universal in the production of haploid meiotic products from a diploid progenitor cell 1 , 2 ., After cells enter into meiosis , maintenance of meiosis is important to ensure proper gametogenesis ., In humans , an inability to properly maintain meiosis can result in developmental abnormalities or possibly oncogenesis in the germ line 1 ., In many organisms , cells that have initiated meiosis pass through an irreversible transition near the end of prophase I . These cells irreversibly commit to undergoing the meiotic divisions ., Cells in prophase I have undergone pre-meiotic DNA replication and have initiated meiosis-specific events such as double strand break ( DSB ) formation , pairing of homologous chromosomes , synaptonemal complex ( SC ) assembly , and the initiation of recombination , but they have not yet entered into the meiotic divisions 2 ., Indeed , human , mouse , and frog oocytes arrest at the end of prophase I and enter into the meiotic divisions only if stimulated by hormones 3 , 4 ., The hormones induce resumption of meiosis and the oocyte becomes committed to finishing meiosis I ., In the D . melanogaster ovarian cyst , 16 cells enter into meiosis; however , at prophase I , only 1 cell , chosen as the oocyte , continues meiosis 5–8 ., The other 15 cells will exit meiosis and enter an endocycle ., In S . cerevisiae , cells commit to meiosis as they exit prophase I and enter the meiotic divisions 9 ., In a process termed return-to-growth ( RTG ) , budding yeast cells in prophase I exit meiosis and return to mitosis if the meiosis-inducing signal is withdrawn and the mitosis-inducing signal is provided 10–14 ., Once they have passed the commitment point , cells are committed to meiosis even without the continued presence of the meiosis-inducing signal ., An understanding of the regulatory mechanisms that drive cells through meiotic commitment points will provide insight into mechanisms that constrain cells to a developmental path ., The ability of budding yeast cells to make the developmental switch from meiosis back to mitosis confers upon them the advantage to alter their developmental program in response to fluctuating environmental conditions 9 ., Nutrient limitation induces the process of sporulation , in which cells enter meiosis and then package the meiotic products into spores ., The spores can survive adverse conditions and then germinate when nutrients become available ., If nutrient-rich conditions return prior to cells reaching the meiotic commitment point , cells exit meiosis and return to mitosis ., In budding yeast , the temporal coordination of cell-cycle events in meiosis is tightly controlled and intertwined with transcriptional cascades 15 , 16 ., Cells initiate meiosis when starved of nitrogen and glucose but provided acetate as an energy source ., The starvation signal stimulates the expression of the Ime1 transcription factor , which induces a class of early genes required for the entry into meiosis , pre-meiotic DNA replication , and prophase I . At pachytene of prophase I , the paired chromosomes have formed extensive synaptonemal complex and have initiated crossing-over ., To exit pachytene of prophase I , the Ndt80 transcription factor is induced and turns on ∼150 middle meiosis genes by binding to a midsporulation element ( MSE ) in the promoter of the genes 17–20 ., Once the Ndt80-dependent genes are expressed , the SC disassembles , a meiotic spindle forms , and cells segregate homologous centromeres in meiosis I and sister centromeres in meiosis II 16 , 21 ., As cells are undergoing meiosis II , a wave of late genes is expressed , many of which encode proteins required for the packaging of the four meiotic products into spores 15 ., Here , we investigated the irreversibility of meiotic commitment ., We hypothesized that the NDT80 transcriptional regulatory network was essential for generating irreversibility due to its requirement in meiotic commitment , its sensitivity to nutritional changes , and its activation through positive feedback to give a high-level burst of NDT80 expression 16 , 17 , 21 , 22 ., The Ime1 transcription factor induces NDT80 transcription by binding an upstream regulatory sequence within the promoter 17 ., NDT80 is expressed later than most Ime1-dependent genes because the NDT80 promoter is regulated by a repressor complex comprised of Sum1 , Rfm1 , and a histone deacetylase Hst1 23 , 24 ., The loss of repression occurs after the phosphorylation of Sum1 by multiple kinases at the end of pachytene 25–27 ., A high-level of NDT80 expression is induced by an autoregulatory positive feedback loop in which Ndt80 binds to MSEs in its own promoter and enhances its own transcription 16 , 17 ., Since positive feedback loops are often found in irreversible cell-cycle transitions 28 , the NDT80 transcriptional network becomes a strong candidate for generating the irreversibility of meiotic commitment ., Using single-cell analysis , we analyzed the role of the NDT80 transcriptional network in regulating the irreversibility of meiotic commitment ., We found that cells commit to meiosis in prometaphase I , after the induction of the Ndt80-dependent genes ., And , that high-level induction of NDT80 was needed for the irreversibility of meiotic commitment ., By making strains that allowed us to manipulate both the timing and level of NDT80 expression , we showed that decreasing the levels of NDT80 could uncouple the entrance into meiosis and meiotic commitment ., We found cells that were inappropriately uncommitted to meiosis; these cells underwent meiosis I and then returned to mitosis instead of finishing meiosis II , becoming multi-nucleate polyploid cells ., Further reducing the levels of NDT80 by making NDT80 promoter mutations to disrupt positive feedback resulted in a complete loss of meiotic commitment ., With complete medium addition , all of the cells returned to mitosis from stages beyond metaphase I . Our work suggests that a threshold level of Ndt80 is needed for the irreversibility of meiotic commitment and that positive feedback in the NDT80 transcriptional regulatory network ensures that threshold level ., Past studies performed on populations of cells and one study on individual cells showed that meiotic commitment occurs after pachytene , but before the first meiotic division 9 , 29 ., The meiotic commitment point was thought to occur at the end of prophase I 10 , 16 , and predicted to be associated with the initiation of the separation of spindle pole bodies ( SPBs ) , the yeast equivalent of the centrosome 9 , 30 ., We wanted to define the stage of meiotic commitment more precisely and needed to establish markers to differentiate prophase I exit , prometaphase I , metaphase I , and anaphase I . As cells exit prophase I , the transcription factor Ndt80 induces the transcription of the middle meiosis genes , including the M-phase cyclins , which are needed for spindle assembly and the meiotic divisions 15 , 21 , 31 , 32 ., As the bipolar spindle assembles , the cells are transitioning from prometaphase I to metaphase I ., In anaphase I , the spindle elongates ., Therefore , we decided to use the changes in spindle length to determine the meiotic stages ., With time-lapse microscopy , we captured images every 10 minutes as cells underwent meiosis ., We followed the expression of three proteins tagged with fluorescent markers: Spc42-mCherry , Zip1-GFP , and GFP-Tub1 ( Figure 1A ) ., Spc42 , a component of the spindle pole body ( SPB ) , the yeast equivalent of the centrosome , was fused to mCherry , allowing us to monitor the separation of the SPBs , and therefore the length of the spindle 33 ., Zip1 , a component of the SC , marks pachytene 34 , 35 ., The disassembly of the SC and concomitant loss of Zip1-GFP localization represents the end of prophase I . GFP-Tub1 , which encodes α-tubulin fused to GFP , allowed us to monitor different meiotic stages based on spindle morphology 36 , 37 ., Although Zip1 and Tub1 are both fused to GFP , the locations of the proteins are morphologically distinct 37 , 38 ( Figure 1A ) ., To ensure that we could indeed differentiate between Zip1-GFP and GFP-Tub1 , we measured the time from SC disassembly to SPB separation in cells with Zip1-GFP and Spc42-mCherry and compared this time to cells with all three marked proteins ., In both strains , we see that the SPBs separate approximately 3 minutes after SC loss ( n\u200a=\u200a100 cells per genotype , SI Figure S1A , B ) , suggesting that we can differentiate between Zip1-GFP and GFP-Tub1 ., To more precisely define prometaphase I and metaphase I , we measured the distance between SPBs to determine spindle lengths as cells progressed from prophase I to anaphase I ., The analysis of the spindle length showed that after the SC disassembles , the spindle undergoes a period of elongation for 31±1 mins ( average time ± S . E . , n\u200a=\u200a75 cells ) ( Table 1 ) ., Once the spindle reaches a length of 3 . 5±0 . 05 µm , the spindle maintains that length for 28±1 mins ., The spindle will increase its length to 4 . 4±0 . 07 µm and then elongates further and chromosomes segregate in anaphase I ., In Figure 1B , we plotted the spindle length at timepoints between the end of prophase I and the beginning of anaphase I for 5 of the 75 cells that we analyzed ., We defined prometaphase I as the time after SC disassembly in which the SPBs are separating and the spindle is elongating from 0 . 1–3 . 4 µm ., We defined the start of metaphase I as the time in which the spindle reaches the stable length of 3 . 5 µm ., To determine whether the spindle elongation from to 3 . 5 µm to 4 . 4 µm marked the transition into anaphase I , we used a strain with Spc42-mCherry and GFP-tagged securin ( Pds1 in budding yeast ) ., Since Pds1 is degraded at the metaphase I to anaphase I transition , we monitored the spindle length at the timepoint just prior to Pds1-GFP degradation to define the spindle length at the end of metaphase I ( Figure 1C ) ., We found that the spindle was on average 3 . 5±0 . 05 µm at the last timepoint the cells were in metaphase I ( n\u200a=\u200a70 cells , average spindle length ± S . E . ) ., In the first timepoint after Pds1 degradation , the cells enter anaphase I and the spindle lengths ranged from 3 . 8 to 9 . 1 µm ., Therefore , we defined metaphase I as the time in which the spindle has reached the length of 3 . 5 µm and anaphase I at the time in which the spindle elongates beyond 3 . 5 µm ( Figure 1D ) ., We next used these defined meiotic stages to pinpoint meiotic commitment ., To monitor meiotic commitment , we used a microfluidics assay coupled to time-lapse microscopy to monitor individual cells 38 ., W303 cells were placed in microfluidic chambers and introduced to sporulation medium to induce meiosis ., After 12 hours , cells were at a variety of meiotic stages , and synthetic complete medium was flowed into the chambers ., Cells were scored based on their spindle length when exposed to complete medium and their cell-cycle outcomes: returned to mitosis , finished meiosis , or arrested ., We found that as the spindle length increased , the percent of cells committed to meiosis also increased ( n\u200a=\u200a300 , Figure 2A , Sup . Table S2 ) ., If complete medium was added to cells with a spindle length between 1 . 0–1 . 99 µm , the cells returned to mitosis ., If complete medium was added to cells with a spindle length from 3 . 0–5 . 5 µm , the cells finished meiosis ., In cells with spindle lengths between 2 . 0–2 . 99 µm , some cells returned to mitosis and some cells finished meiosis ., Since spindle assembly requires the activity of the M-phase cyclins whose transcription is dependent on Ndt80 , our findings suggest that cells commit to meiosis after the expression of the Ndt80-dependent genes ., We next assigned the cells a meiotic stage based on their spindle length ( Figure 2B ) , and found that 100% of cells in prophase I returned to mitosis upon complete medium addition , as previously reported ( n\u200a=\u200a100 , Figure 2B , C , Supp . Table S3 ) 9 ., Of cells in prometaphase I upon complete medium addition , 64% returned to mitosis ( Video S1 , Figure 2B , D ) , 34% finished meiosis ( Video S2 , Figure 2B , E ) , and 2% arrested , in meiosis I ( n\u200a=\u200a100 ) ., Finally , 99% of cells finished meiosis upon complete medium addition in metaphase I ( n\u200a=\u200a100 , Figure 2B , F ) ., These data show that the meiotic commitment point lies in prometaphase I ( Figure 2G ) ., Prior experiments did not have the sensitivity to resolve these variations in cell-cycle outcomes upon nutrient addition in prometaphase I 9 ., Prometaphase I occurs after the synaptonemal complex is disassembled and after spindle formation initiates ., This suggests that prior to commitment , Ndt80 has started transcribing the Ndt80-dependent genes such as the M phase cyclins , which are needed for spindle formation ., To verify that the Ndt80-dependent genes were indeed transcribed and translated prior to commitment , we monitored the timing of the accumulation of the encoded protein of the Ndt80-dependent gene CDC5 , which encodes polo kinase ., Cdc5 is necessary and sufficient for SC disassembly at the end of prophase I 39 ., To determine whether Cdc5 was present prior to commitment in prometaphase I , we tagged Cdc5 with Superfolder-GFP ( Cdc5-sfGFP ) , a fast folding variant of GFP 40 , 41 in a strain with Spc42-mCherry ., As expected , cells in prophase I that have not yet expressed NDT80 do not have Cdc5-sfGFP present ( Figure 3A ) ., In contrast , 100% of cells in prometaphase I , with a spindle length between 0 . 1–3 . 4 µm , have Cdc5-GFP present ( n\u200a=\u200a44 cells ) ( Figure 3B-C ) ., These results demonstrate that the Ndt80-dependent gene CDC5 is expressed and the protein is present in the cell prior to meiotic commitment ., Although cells commit to meiosis after the initiation of the expression of the Ndt80-dependent genes , we considered that Ndt80 could have a role in the irreversibility of meiotic commitment due to its tightly regulated transcriptional activation ., We asked if modifying NDT80 expression could lead to an alteration in the establishment of meiotic commitment ., We placed NDT80 under the control of the GAL1 , 10 promoter ( PGAL1 , 10-NDT80 ) and the Gal4 activator was fused to the estradiol receptor so that the addition of estradiol activated NDT80 expression via Gal4 ( GAL4-ER ) 32 , 42 ., The PGAL1 , 10-NDT80/PGAL1 , 10-NDT80 cells undergo meiosis and form spores with the addition of estradiol , as previously described 32 ., Because the NDT80 promoter is absent in the PGAL1 , 10-NDT80 cells , NDT80 transcription is no longer subject to transcriptional positive feedback , but instead can be regulated by an inducible promoter to limit the duration of transcription ., PGAL1 , 10-NDT80/PGAL1 , 10-NDT80 cells in microfluidics chambers were exposed to sporulation medium to induce meiosis until they arrested at pachytene ( due to the lack of Ndt80 ) ., Estradiol was added to induce NDT80 , and , at 10-minute timepoints after the induction of NDT80 , we introduced complete medium into separate chambers and recorded cell-cycle outcomes of each chamber ( Figure 4A ) ., The addition of complete medium , which contains glucose but not estradiol , represses the transcription of the GAL1 , 10 promoter ., In Figure 4B , we have plotted the percent of cells for each cell-cycle outcome ( y-axis ) when complete medium was added at a timepoint after the cells disassembled the SC and exited prophase I ( x-axis ) ( n\u200a=\u200a100 cells counted for each timepoint ) ( Supp . Table S4 ) ., When complete medium was added ten minutes after SC loss , 99% of the cells returned to mitosis ., However , if complete medium was added 30 minutes after SC loss , only 5% of cells returned to mitosis; 60% finished meiosis; 3% arrested; and , the remaining 32% of cells took another path , described below ., By 50 minutes after SC loss , enough Ndt80 had accumulated such that 100% of the cells were committed to meiosis and finished meiosis after the addition of complete medium ., Importantly , we found an additional cell-cycle outcome in the inducible NDT80 strain: those cells that completed meiosis I , assembled two spindles , then returned to mitosis as demonstrated by bud formation ( Video S3 , Figure 4B , C ) ., After budding , both nuclei divided , resulting in multi-nucleate cells ., We defined these cells as inappropriately uncommitted –they underwent meiosis I but were not committed to finishing meiosis II and instead returned to mitosis ., This is a novel behavior representing the absence of commitment , and different from the cells that are not yet committed to meiosis I and returned to mitosis from prophase I ., When complete medium was added 20 minutes and 30 minutes after SC loss , 36% and 32% of cells were inappropriately uncommitted , respectively ., Also , at the 20- and 30-minute timepoints , the percentage of cells that returned to mitosis decreased and those that finished meiosis increased when compared to the 10-minute timepoint ., These data suggest that , with an inducible NDT80 , either the duration of NDT80 transcription or the level of Ndt80 protein after SC breakdown is important for meiotic commitment ., With NDT80 expressed for a short time period ( 0–10 mins ) after SC disassembly , the cells returned to mitosis ., With NDT80 expressed for a longer time period ( 50 minutes or more ) , the cells committed to meiosis ., We propose that at timepoints in between ( 20–40 minutes ) , Ndt80 may be at intermediate levels:, i ) some cells had Ndt80 below a threshold level and returned to mitosis;, ii ) some cells had Ndt80 above a threshold level and committed to meiosis , and;, iii ) some cells had enough Ndt80 to finish meiosis I , but not enough Ndt80 to commit to meiosis II and returned to mitosis after meiosis I ., We considered that insufficient levels of Ndt80 protein in the PGAL1 , 10-NDT80 strain could cause the inappropriately uncommitted cell phenotype ., It has previously been shown that expression from the GAL1 , 10 promoter in meiosis does not lead to a large overproduction of the expressed protein 32 ., Furthermore , the addition of complete medium with glucose should repress NDT80 transcription from the GAL1 , 10 promoter ., Therefore , we asked whether Ndt80 protein levels in the PGAL1 , 10-NDT80 cells decrease with the addition of complete medium ., Cells were induced to enter meiosis ., When cells arrested at pachytene , estradiol was added to induce the expression of Ndt80 ., After 110 minutes of Ndt80 induction , complete medium was added and aliquots were taken every 30 minutes for 2 . 5 hours ., The 110-minute timepoint was chosen due to the large fraction of inappropriately uncommitted cells at this timepoint ( after 110-minutes of estradiol addition , most of the cells had disassembled their SC 30 minutes earlier , and therefore this timepoint corresponds to the 30-minute timepoint shown in Figure 4B ) ., Analysis of Ndt80 levels by western blot showed that the Ndt80 protein levels decreased by 86% within 30-minutes of complete medium addition and remained at that very low level ( Figure 5A , upper panel ) ., In a control experiment , cells that continued in meiosis with estradiol in the sporulation medium maintained the high level of Ndt80 protein ( Figure 5A , lower panel ) ., We next determined if there was a difference between Ndt80 protein levels in the PGAL1 , 10-NDT80 strain with the addition of complete medium compared to a strain with NDT80 under control of its endogenous promoter ., Since the cells with PNDT80-NDT80 are not as synchronous as the PGAL1 , 10-NDT80 cells , we added a cdc20 meiotic null to both strains so that the cells would arrest in metaphase I due to a loss of Cdc20 and remain in meiosis ., Cells were induced to enter meiosis ., When the PGAL1 , 10-NDT80 cells arrested at pachytene , estradiol was added to induce the expression of NDT80 ., After 110 minutes of NDT80 induction , complete medium was added and aliquots were taken at 0 , 15 , and 30 minutes to isolate protein ., For the PNDT80-NDT80 cells , complete medium was added after 12 hours since the cells were in prometaphase I- metaphase I at this timepoint ., The protein was isolated at 0 , 15 , and 30-minute timepoints after complete medium addition ., Analysis of Ndt80 by western blot showed that in the PGAL1 , 10-NDT80 cells , the Ndt80 protein decreased by 92% within 15-minutes of complete medium addition ( Figure 5B ) ., In contrast , in the PNDT80-NDT80 cells , there was only a 27% decrease of Ndt80 protein 15 minutes and a 55% decrease 30 minutes after the introduction of complete medium ., These results suggest that insufficient levels of Ndt80 in PGAL1 , 10-NDT80 cells after complete medium addition is likely to result in the inappropriately uncommitted cells ., Since the levels of Ndt80 decreased with the addition of complete medium in the inducible NDT80 strain , we predicted that lower levels of Ndt80 could result in an altered commitment point ., To test our prediction , we further decreased Ndt80 levels by making a PGAL1 , 10-NDT80/ndt80Δ strain ., In this strain , one copy of NDT80 is under the GAL1 , 10 promoter and the other copy of NDT80 is deleted ., These cells will progress through meiosis and form spores after the induction of NDT80 with estradiol ., Using our microfluidics assay , we monitored cell cycle outcomes after complete medium addition ., In Figure 5C , we compare the cell cycle outcomes of PGAL1 , 10-NDT80/PGAL1 , 10-NDT80 and PGAL1 , 10-NDT80/ndt80Δ ., Our results are presented as a graph of the percent of cells with each cell cycle outcome ( y-axis ) at the different meiotic stages in which complete medium was added ( x-axis ) ( Figure 5C , Supp . Table S5 ) ., Our data show an important difference in the percent of committed cells between the different strain backgrounds ., In the cells with two copies of PGAL1 , 10-NDT80 , there are inappropriately uncommitted cells upon nutrient addition in prometaphase I , metaphase I , and anaphase I ( n>200 cells counted at each stage ) ., However , most cells were committed to meiosis upon nutrient addition in metaphase I and anaphase I ., In contrast , in cells with only one copy of PGAL1 , 10-NDT80 , most cells did not finish meiosis upon nutrient addition ., Of the PGAL1 , 10-NDT80/ndt80Δ cells in prometaphase I upon nutrient addition , 98% returned to mitosis ( n\u200a=\u200a178 cells ) ., Of the cells in metaphase I upon nutrient addition 51% returned to mitosis , 8% finished meiosis , 14% were inappropriately uncommitted , and the remainder arrested in either meiosis I or meiosis II ( n\u200a=\u200a157 cells ) ., Of the cells in anaphase I upon nutrient addition , 84% were inappropriately uncommitted , 8% arrested in meiosis II , and only 8% finished meiosis ( n\u200a=\u200a100 cells ) ., These results show that by decreasing the levels of Ndt80 , fewer cells commit to finishing meiosis ., A comparison of the two strains , PGAL1 , 10-NDT80/PGAL1 , 10-NDT80 and PGAL1 , 10-NDT80/ndt80Δ , suggests that decreasing the levels of NDT80 resulted in fewer cells committed to meiosis ., Therefore , we predict that there is a threshold level of Ndt80 needed to drive cells through meiotic commitment ., In a normal meiosis , the high-level burst of NDT80 expression at the end of pachytene is driven by an auto-regulatory feedback loop in which Ndt80 enhances its own transcription by binding to MSE sequences found in its own promoter 17 , 43 ., We hypothesize that positive feedback in NDT80 induction may ensure that cells reach that switch threshold level of NDT80 required for meiotic commitment ., To test this hypothesis , we eliminated positive feedback by deleting the two 9 bp MSEs in the NDT80 promoter ( PNDT80-MSE1Δ , 2Δ-NDT80 ) ., In this strain , NDT80 can be activated by the Ime1 transcription factor , which binds to an upstream regulatory sequence , but cannot be activated through an Ndt80-dependent autoregulatory loop ., We verified by western blot that the PNDT80-MSE1Δ , 2Δ-NDT80 cells had a substantially lower level of Ndt80 protein ( Figure 5D ) ., We next tested whether the meiotic commitment point was altered in the PNDT80-MSE1Δ , 2Δ-NDT80 cells ., We performed our microfluidics assay and found that the cells were not committed to meiosis ., In Figure 5E , we plot the percent of cells with each cell cycle outcome ( y-axis ) at the different meiotic stages in which complete medium was added ( x-axis ) ( n\u200a=\u200a100 cells counted for each cell-cycle stage ) ( Supp . Table S6 ) ., Of the cells in prometaphase I at the time of complete medium addition , 99% returned to mitosis ., PNDT80-MSE1Δ , 2Δ-NDT80 cells in metaphase I at the time of nutrient addition gave several cell-cycle outcomes:, i ) 16% were inappropriately uncommitted and underwent meiosis I , then budded and underwent mitosis , dividing both nuclei;, ii ) 5% arrested in meiosis I; and, iii ) 79% budded and returned to mitosis from metaphase I ( Video S4 , Figure 5E , 6A ) ( n\u200a=\u200a100 cells counted ) ., The PNDT80-MSE1Δ , 2Δ-NDT80 cells in anaphase I at the time of complete medium addition were also not committed to meiosis; they finished meiosis I with complete medium addition , but 100% of the cells budded and underwent mitosis instead of meiosis II ( Figure 6B , n\u200a=\u200a100 cells counted ) ., Cells beyond anaphase I were also inappropriately uncommitted to meiosis; 98% of cells in prophase II at the time of complete medium addition budded and underwent mitosis ( n\u200a=\u200a100 cells counted ) ., These results show that meiotic commitment requires positive feedback in NDT80 expression and support our hypothesis that a threshold level of Ndt80 is needed to drive meiotic commitment ., We next asked whether the PNDT80-MSE1Δ , 2Δ-NDT80 cells could go through meiosis ., We used time-lapse microscopy to monitor the cells undergoing meiosis and found that 78% of the cells that entered meiosis underwent meiosis I but then arrested at metaphase II ( Figure 6C , n\u200a=\u200a100 ) ., Only 5% of the cells finished both meiotic divisions , 10% arrested at pachytene , and 7% arrested in metaphase I . This suggests that the levels of Ndt80 in the PNDT80-MSE1Δ , 2Δ-NDT80 strain are sufficient to allow most cells to enter into meiosis I and II , but not sufficient to finish meiosis II ., Since the PNDT80-MSE1Δ , 2Δ-NDT80 cells do not commit to meiosis , a higher-level induction of NDT80 is required for meiotic commitment ., To determine if modestly lower levels of NDT80 result in an alteration of the meiotic commitment point , we decreased Ndt80 levels by making an NDT80/ndt80Δ heterozygote ., In this strain , one copy of NDT80 was under the control of its native promoter and the other copy was deleted ., Using our microfluidics assay , we found that of the NDT80/ndt80Δ cells that were in prometaphase I at complete medium addition , 92% returned to mitosis and 7% finished meiosis ( n\u200a=\u200a100 cells counted , Figure 7 , Supp . Table S3 ) ., Of the cells in metaphase I at complete medium addition , 2% returned to mitosis , 84% finished meiosis , 11% arrested in metaphase II , and 3% were inappropriately uncommitted ( n\u200a=\u200a100 cells counted ) ( Figure 7 ) ., The results show that in NDT80/ndt80Δ cells , fewer cells commit to meiosis in prometaphase I when compared to NDT80/NDT80 , suggesting that the commitment point shifted to the end of prometaphase I/beginning of metaphase I ., We conclude that reducing the NDT80 levels alters the timing of the meiotic commitment point ., Once NDT80/ndt80Δ cells exit prophase I , the timing of meiosis was not delayed in comparison to wildtype cells ., In wildtype cells , the duration of prometaphase I and metaphase I was 31±1 mins and 28±1 mins , respectively ( n\u200a=\u200a75 cells counted , average time in minutes ± S . E . ) ., In NDT80/ndt80Δ cells , the duration of prometaphase I and metaphase I was 28±1 mins and 21±1 mins , respectively ( n\u200a=\u200a80 cells , average time in minutes ± S . E . ) ., The timing of the other stages of meiosis were also not delayed ( Table 1 ) ., These results suggest that although two copies of NDT80 are needed for the proper timing of meiotic commitment , one copy of NDT80 is sufficient for the normal duration of each of the meiotic stages ., The importance of the irreversibility of meiotic commitment can be demonstrated through phenotypic analysis of the inappropriately uncommitted cell; the failure to commit to meiosis beyond prometaphase I results in the formation of multi-nucleate cells ., The PNDT80-GAL1 , 10NDT80/PNDT80-GAL1 , 10NDT80 inappropriately uncommitted cells underwent a first meiotic division , creating two nuclei in the mother cell ., After the first meiotic division , the uncommitted cells budded and underwent mitosis instead of finishing meiosis II ( Figure 4C ) ., After the mitotic division , depending on how the nuclei divided , the mother cell had either 2 or 3 nuclei and the daughter had either 2 nuclei or 1 nucleus ., There were three nuclear segregation phenotypes ( Figure 8A-C ) :, i ) 36% segregated both nuclei across the bud neck , resulting in two nuclei each in the mother and daughter cells ( Figure 8A ) ;, ii ) 32% segregated one nucleus in the mother cell and one nucleus across the bud neck , resulting in three nuclei in the mother and one in the daughter cell ( Figure 8B ) , and;, iii ) 32% segregated one nucleus in the mother cell and one nucleus in the daughter cell , resulting in two nuclei each in the mother and daughter cells ( Figure 8C ) ( n\u200a=\u200a100 ) ., The multi-nucleate mother and daughter cells continued to divide ( Video S3 ) and could increase genome copy number in the subsequent divisions due to the nuclear segregation phenotype described for Figure 8B ., Our results show that the loss of meiotic commitment can have the drastic consequence of the loss of genome integrity ., To determine the ploidy of the inappropriately uncommitted cells , we asked if DNA re-replication occurred after meiosis I but prior to the mitotic division ., We monitored the nuclear localization of a component of the replicative helicase , Mcm7 , tagged with GFP ( Mcm7-GFP ) ., The replicative helicase enters the nucleus and loads onto origins prior to DNA replication 44 ., We have previously found that Mcm7-GFP does not enter into the nucleus when cells return to mitosis from prophase I , and these cells also do no re-replicate their DNA 38 , 45 ., To determine if the inappropriately uncommitted cells license their origins , we monitored the localization of Mcm7-GFP from the strain with PNDT80-GAL1 , 10NDT80/ndt80Δ ., Using time-lapse microscopy , we found that in the inappropriately uncommitted cells , Mcm7-GFP entered the nuclei after meiosis I but prior to the mitotic division in 92% of the cells ( n\u200a=\u200a27 cells ) ., Figure 8D shows Mcm7-GFP localized in the two nuclei of an inappropriately uncommitted cell ., These results suggest that in the inappropriately uncommitted cells , the levels of CDK decrease after meiosis I such that Mcm7-GFP enters the nuclei and licenses origins ., The data support the conclusion that the cell exits meiosis after the first division and then begins the mitotic cell cycle ., To confirm that DNA replication did indeed occur , we monitored a marked chromosome after segregation ., We labeled one copy of chromosome III with a lactose operator array ( LacO ) near the centromere in a strain expressing GFP fused to the lactose repressor ( LacI-GFP ) 46 ., We scored the inappropriately uncommitted cells that w

Introduction, Results, Discussion, Materials and Methods

In budding yeast , meiotic commitment is the irreversible continuation of the developmental path of meiosis ., After reaching meiotic commitment , cells finish meiosis and gametogenesis , even in the absence of the meiosis-inducing signal ., In contrast , if the meiosis-inducing signal is removed and the mitosis-inducing signal is provided prior to reaching meiotic commitment , cells exit meiosis and return to mitosis ., Previous work has shown that cells commit to meiosis after prophase I but before entering the meiotic divisions ., Since the Ndt80 transcription factor induces expression of middle meiosis genes necessary for the meiotic divisions , we examined the role of the NDT80 transcriptional network in meiotic commitment ., Using a microfluidic approach to analyze single cells , we found that cells commit to meiosis in prometaphase I , after the induction of the Ndt80-dependent genes ., Our results showed that high-level expression of NDT80 is important for the timing and irreversibility of meiotic commitment ., A modest reduction in NDT80 levels delayed meiotic commitment based on meiotic stages , although the timing of each meiotic stage was similar to that of wildtype cells ., A further reduction of NDT80 resulted in the surprising finding of inappropriately uncommitted cells: withdrawal of the meiosis-inducing signal and addition of the mitosis-inducing signal to cells at stages beyond metaphase I caused return to mitosis , leading to multi-nucleate cells ., Since Ndt80 enhances its own transcription through positive feedback , we tested whether positive feedback ensured the irreversibility of meiotic commitment ., Ablating positive feedback in NDT80 expression resulted in a complete loss of meiotic commitment ., These findings suggest that irreversibility of meiotic commitment is a consequence of the NDT80 transcriptional positive feedback loop , which provides the high-level of Ndt80 required for the developmental switch of meiotic commitment ., These results also illustrate the importance of irreversible meiotic commitment for maintaining genome integrity by preventing formation of multi-nucleate cells .

There are two main types of cell division cycles , mitosis and meiosis ., During mitosis , DNA is replicated and then chromosomes segregate , producing two daughter cells with the same ploidy as the progenitor cell ., During meiosis , DNA is replicated and then chromosomes undergo two rounds of segregation , producing four gametes with half the ploidy of the progenitor cell ., As the cell enters into the meiotic divisions , it irreversibly commits to finishing meiosis and cannot return to mitosis ., The molecular mechanisms that define meiotic commitment are not well understood ., In this study , we asked whether the regulatory network involved in the transcription of NDT80 has a role in meiotic commitment ., Ndt80 is a transcription factor that induces the genes needed for the meiotic divisions ., We found that a high-level of Ndt80 activity is required for meiotic commitment; in wildtype cells , this is achieved through a transcriptional positive feedback loop– a regulatory mechanism in which the Ndt80 protein increases the transcription of its own gene ., In the absence of positive feedback , cells escape meiosis inappropriately , resulting in an aberrant cell cycle that causes an increase in genome copy number ., This study shows the important role of positive feedback in meiotic commitment and in the maintenance of genome integrity .

biology and life sciences

journal.pgen.1007355

Targeted next generation sequencing identifies functionally deleterious germline mutations in novel genes in early-onset/familial prostate cancer

Prostate cancer ( PrCa ) is the most frequent non-cutaneous cancer diagnosed in men worldwide and the third leading cause of male cancer deaths in Europe 1 ., Despite efforts in early detection and screening strategies 2 , PrCa is estimated to be responsible for the death of 27 , 540 men in the United States in 2015 1 ., Contrarily to other cancer types , very little is known about the genetic contribution to the 10–20% of PrCa cases with evidence of familial clustering 3 ., In fact , besides age and race , family history is the only other well-established risk factor for PrCa 4 ., While familial PrCa is defined by an aggregation of PrCa in families , hereditary prostate cancer ( HPC ) is characterized by a pattern of Mendelian inheritance associated with rare mutations in susceptibility genes 3 , 5 ., First-degree relatives of a PrCa patient have a two-fold increased risk of developing the disease compared to the general population ., The risk is even higher when the number of affected relatives increases and the age at diagnosis decreases 3 , 5 , 6 ., The existence of a genetic component behind PrCa development is strengthened by the four-fold higher concordance rate of PrCa among monozygotic twins compared to dizygotic twins 7 , 8 ., Linkage analysis and genome-wide association studies have pinpointed some loci associated with PrCa predisposition , but the majority has not been consistently reproduced 9 ., In 2004 , a combined genome-wide linkage analysis of 426 families from four HPC studies identified a locus at 17q21-22 strongly associated with PrCa 10 ., Despite previous reports linking mutations in BRCA1 ( at 17q22 ) with PrCa predisposition 11 , 12 , Ewing et al . later identified a rare but recurrent mutation ( G84E ) in the HOXB13 gene ( at 17q21 ) in up to 3% of the patients with both early-onset and family history of the disease , using a next-generation sequencing ( NGS ) approach covering the 202 genes present in the defined region of interest ( ROI ) at the 17q21-22 locus 13 ., An increased risk of PrCa for the HOXB13 G84E mutation carriers has been confirmed by several groups 14 , 15 and other HOXB13 variants associated with PrCa have been found in other populations 16 , 17 ., Besides HOXB13 , BRCA2 mutation carriers are also at increased risk of developing PrCa 18–20 ., Overall , BRCA2 mutations seem to explain about 2% of early-onset PrCa cases 19 , a frequency that can be slightly higher for BRCA2 mutations with a founder effect in specific populations 21 , 22 ., Additionally , a higher risk for PrCa in Lynch syndrome families has been proposed 23 , 24 , with some studies reporting a five- to ten-fold increased risk of PrCa development for carriers of MSH2 mutations compared to non-carriers 25 , 26 ., However , recent studies of our group found germline mutations in HOXB13 , BRCA2 and MSH2 in only 1 . 5% of early-onset and/or familial PrCa cases 17 , 27 ., Mutations in a few additional genes or specific variants , namely in CHEK2 28–31 , NBN 32 , 33 , ATM 34 , 35 , and BRIP1 36 , have been reported to increase the risk of PrCa , although some in a population-specific context ., Despite these reports , the large majority of prostate carcinomas showing Mendelian inheritance still have no explanation concerning highly penetrant susceptibility variants ., In this work , we aimed to evaluate the proportion of cases with early-onset and/or familial/hereditary PrCa that can be attributed to mutations in 94 genes associated with inherited cancer predisposition , using our validated targeted next generation sequencing ( NGS ) pipeline 37 ., This approach allowed to identify functionally deleterious/“potentially pathogenic” mutations in nine genes , revealing six genes ( CEP57 , FANCD2 , FANCI , RAD51C , RECQL4 and TP53 ) not previously associated with PrCa predisposition ., Overall , a candidate disease-causing mutation in a cancer predisposing gene was identified in 18 patients ( 14 . 9% ) , with ATM and CHEK2 representing 61 . 1% of the cases ., Of the genes previously reported to increase the risk for PrCa development ( after excluding cases with known mutations in HOXB13 , BRCA2 and MSH2 in this series ) , we found a nonsense mutation in ATM and a splicing mutation in CHEK2 ( Table 1 ) ., The ATM mutation c . 652C>T , which leads to a premature stop codon at codon 218 , was found in a patient ( HPC177 ) with five brothers diagnosed with PrCa ( Fig 1A ) , including twin brothers diagnosed before the age of 61 years , thus fulfilling the A1 and A2 criteria ( see Material and Methods section for criteria description ) ., The family is living abroad , which renders segregation analysis difficult to perform ., The CHEK2 mutation c . 593-1G>T , predicted to affect the splice site by three of the four queried in silico predictors ( S1 Table ) and reported as “likely pathogenic” in ClinVar , was found in a patient ( HPC395 ) with a family history of three breast cancer ( BrCa ) cases , two of them diagnosed at early age ( Fig 1B ) , thus fulfilling the B3 criterion ., One of the nices with BrCa is carrier of the CHEK2 mutation c . 593-1G>T ., To strengthen the causality between these mutations and cancer development , we used KASP genotyping in 710 healthy controls and searched for the variant among 504 samples from non-prostate cancer cases analyzed with the same NGS panel and pipeline in the Department of Genetics of IPO Porto ., Among the 504 cancer cases , the same ATM stop mutation was found in a patient diagnosed with bilateral BrCa at early age ( previously described 38 ) and the same CHEK2 splicing mutation was found in an early-onset breast and colon cancer patient ., No carriers were found either among our 710 healthy controls or in ExAC , for both mutations ., Considering that most of the genes so far associated with an increased risk for PrCa development have previously been described to predispose to breast/ovarian cancer and/or Lynch syndrome , we looked for missense variants in the 18 genes associated with these diseases , namely ATM , BLM , BRCA1 , BRCA2 , BRIP1 , CDH1 , CHEK2 , MLH1 , MSH2 , MSH6 , NBN , PALB2 , PMS2 , PTEN , RAD51C , RAD51D , STK11 , and TP53 ., Missense variants predicted to be pathogenic by at least 12 of the 15 in silico pathogenicity predictors ( including at least three conservation tools ) were considered “potentially pathogenic” ., Of the 42 missense variants found ( S2 Table 45 ) , ten variants fulfill these criteria ( Table 2 ) and include four variants in ATM , two in CHEK2 , and one in each of the genes BRIP1 , MSH2 , MSH6 and TP53 ., The CHEK2 missense mutation c . 349A>G , found in two patients ( HPC188 and HPC289 ) , was the only mutation classified as “pathogenic/likely pathogenic” in ClinVar ., Patient HPC188 has family history of PrCa , with three first- and second-degree relatives diagnosed at or before the age of 65 years ( one early-onset ) , thus fulfilling the A1 and A2 criteria ( S1A Fig ) ., Patient HPC289 is an early-onset PrCa case , fulfilling the B1 and B3 criteria for having a heavy family history of cancer , with several cases diagnosed at early age ( S1B Fig ) ., Both MSH2 and MSH6 variants are reported in public databases and classified as variants of unknown significance ( VUS ) in ClinVar ., As mutations in Lynch syndrome predisposing genes are usually associated with loss of protein expression in the tumor , we performed immunohistochemistry for the MSH2 and MSH6 proteins in the prostate tumors of the patients HPC371 and HPC332 , respectively , and no loss of expression was found , rendering the MSH2 and MSH6 mutations as probably not associated with PrCa development in these patients ., Apart from the CHEK2 missense mutation c . 349A>G , the remaining missense variants here identified were either not described in the literature or classified as VUS ., To increase our understanding on the pathogenic potential of these variants , we screened our 710 healthy controls and the 504 non-prostate cancer samples as described above ., For the ATM mutations c . 995A>G and c . 8560C>T and for the TP53 mutation c . 839G>A , no additional carriers were found ., The BRIP1 mutation c . 847T>C and the CHEK2 mutation c . 695G>T were found in one of the 504 cancer cases ( each ) and the ATM mutations c . 1595G>A and c . 5750G>A were found in two cases ( each ) of the 710 healthy controls and in six and three cases , respectively , of the 504 cancer cases ., Of all these variants , only the ATM mutation c . 8560C>T was found significantly increased in our PrCa patients comparing with our healthy controls ( P = 0 . 024; S3 Table ) , with the CHEK2 mutation c . 349A>G reaching borderline significance ( P = 0 . 057 ) ., With the exception of the TP53 mutation c . 839G>A , not found in any of the 504 non-prostate cancer patients , all missense mutations have no significant frequency differences in cancer patients fulfilling criteria for other hereditary cancer syndromes ( P>0 . 05; S3 Table ) ., When comparing the frequencies obtained in our PrCa patients with those of the Non-Finnish Europeans ( NFE ) described in ExAC , highly significant associations are obtained for all the ATM and CHEK2 missense mutations ( S3 Table ) ., Following the guidelines from the American College of Medical Genetics and Genomics and the Association for Molecular Pathology ( ACMG-AMP ) for variant interpretation and classification 46 using InterVar 47 , all the missense variants here identified in ATM , CHEK2 , BRIP1 and TP53 are classified as VUS ., Adding the PS3 criterion ( “well-established in vitro or in vivo functional studies supportive of a damaging effect on the gene or gene product” ) to the classification of the CHEK2 variant c . 349A>G 48 , 49 and the PS4 criterion ( “the prevalence of the variant in affected individuals is significantly increased compared with the prevalence in controls” ) to the classification of the ATM variant c . 8560C>T , supports the “likely pathogenic” nature of these variants in PrCa development ., Of the 121 cases enrolled in this study , 45 have criteria to be classified as familial/hereditary PrCa ( A group ) and 86 are cases of early-onset PrCa and/or PrCa associated with clustering of other cancers in the family ( group B ) , with ten cases fulfilling both A and B criteria ( S4 Table ) ., Regarding age at diagnosis , 64 cases ( 52 . 9% ) were diagnosed with PrCa at or before the age of 55 years , thus being considered early-onset PrCa cases ., Considering the number of cases with prostate carcinomas in the 121 families , 91 cases ( 75 . 2% ) have family history of two or more relatives with PrCa , with 27 cases ( 22 . 3% ) having three and 33 cases ( 27 . 3% ) having at least four ., When comparing clinicopathological characteristics of the patients harboring the deleterious/”potentially pathogenic” mutations ( n = 18; excluding the cases with the MSH2 and MSH6 VUS , described above ) with the “negative” group ( n = 103 ) , no statistically significant associations were observed , either considering all cases or considering the subgroups of cases with familial/hereditary PrCa or early-onset PrCa ( S5 Table ) ., Consistent with the increasing chance of incidental findings of the NGS approaches , we found a c . 3846_3860del in-frame deletion in MSH6 that falls into this classification ., This mutation was found in the patient carrying also the truncating mutation in RAD51C ( HPC186 ) and is classified as pathogenic in two of the Lynch syndrome families diagnosed at IPO Porto ., However , we find unlikely its association with the PrCa in this patient , as no loss of MSH6 expression was observed in the tumor ( contrarily to what we observed in the colon carcinomas of our Lynch syndrome families ) ., Analyses in the available relatives showed segregation of the variant in the niece with colon cancer ( S2 Fig ) ., In this work we used a NGS approach targeting the full coding-sequence of 94 genes associated with cancer predisposition to identify germline mutations in a selected series of 121 PrCa patients with early-onset disease and/or criteria for familial/hereditary PrCa , alone or associated with other cancers ., This strategy is justifiable by the fact that , with the exception of HOXB13 , all the genes so far associated with PrCa hereditary predisposition were previously associated with an increased risk for BrCa , OvCa or other cancers , including those causing the phenotypically heterogeneous diseases hereditary breast/ovarian cancer ( HBOC ) and Lynch syndrome 20 , 24 , 27 , 50 ., Using our previously established NGS analysis pipeline 37 , and after excluding the few cases in our series with germline mutations in HOXB13 or in genes associated with HBOC or Lynch syndrome 17 , 27 , we found monoallelic truncating/deleterious mutations in seven genes , of which only ATM and CHEK2 have been previously implicated in PrCa development 28–31 , 34 , 35 , 43 ., Deleterious mutations in ATM and CHEK2 thus represent 0 . 8% ( each ) of the cases enrolled in this study , with the nonsense ATM mutation representing 2 . 2% of the cases fulfilling criteria for familial/hereditary PrCa ( A group ) , which resembles the frequency of mutations found in BRCA2 in earlier studies 27 , 35 ., Curiously , both mutations occur in families with several BrCa cases and were both found in one case ( each ) of the 504 non-prostate cancer cases analyzed with the same NGS approach in the Department of Genetics for fulfilling criteria for HBOC ., We found monoallelic functionally deleterious mutations in three genes of the FA family , namely RAD51C ( FANCO ) , FANCD2 and FANCI genes , the latter two not previously associated with cancer risk ., RAD51C , a RAD51 paralog involved in the homologous recombination ( HR ) repair pathway 51 , was first described as a susceptibility gene for BrCa and OvCa , showing complete segregation in six families 44 ., Nowadays , RAD51C deleterious mutations are established as a risk factor for OvCa only , with a prevalence of about 0 . 8% in familial OvCa and 0 . 4–1 . 1% in OvCa cases unselected for family history 42 ., The family history of the patient harboring the c . 890_899del mutation in RAD51C has no confirmed ovarian cancer diagnosis , but includes a relative with gynecological cancer deceased at young age ., FANCD2 and FANCI are involved in the initial steps of the FA pathway , leading to the activation of downstream repair factors , such as FANCD1 ( BRCA2 ) , FANCJ ( BRIP1 ) , FANCN ( PALB2 ) and FANCO ( RAD51C ) , to mediate HR 52 ., Considering that mutations in all these four FA members have been associated with risk for BrCa and/or OvCa 41 , 43 , 44 , with mutations in BRIP1 and BRCA2 also associated with PrCa development 19 , 20 , 36 , our report of mutations in RAD51C , FANCD2 and FANCI may increase to five the list of FA members involved in PrCa predisposition ., In our series , functionally deleterious mutations in FA genes represent 4 . 4% ( 2/45 ) of the familial/hereditary PrCa cases ( A criteria ) and 1 . 6% ( 1/64 ) of the early-onset PrCa cases ., Among the 504 non-prostate cancer cases diagnosed at our institution with the same NGS approach , three carriers of the same mutations were found , one with the FANCD2 mutation and two with the RAD51C mutation , with different cancers occurring in the families ., Further studies are required to determine the frequency of germline mutations in these genes in PrCa and other hormone-related cancers , as the pedigrees of both case HPC186 and case HPC447 , with the RAD51C and the FANCD2 functionally deleterious mutations , respectively , include relatives affected with BrCa and/or gynecological cancers ., To our knowledge , this is also the first report of heterozygous germline truncating mutations in CEP57 and RECQL4 as possible cancer risk factors ., CEP57 encodes a 57 kDa member of the CEP family of centrosomal proteins involved in MVA2 , a rare pediatric syndrome with high risk of development of childhood cancers 53 , 54 ., On the other hand , RECQL4 belongs to a family of five RecQ helicases RECQL1 , WRN ( RECQL2 ) , BLM ( RECQL3 ) , RECQL4 and RECQL5 55 ., Interestingly , monoallelic mutations in RECQL1 and in the Bloom syndrome gene BLM have been described as risk factors for BrCa 56 , 57 , although the BLM association has been contested by others 58 , 59 ., According to the Uniprot database , the RECQL4 frameshift mutation c . 2636del we here describe is not expected to affect the known functional domains of the protein , but more downstream ( C-terminal ) mutations in RECQL4 have been shown to cause RTS or GBS 55 and the C-terminal seems to be necessary for RECQL4 nucleolar localization through interaction with PARP-1 60 , therefore making very likely its deleterious nature ., Additionally , the absence of both mutations in public databases , namely ExAC , and in the 504 non-prostate cancer cases analyzed in our institution with the same NGS approach , may reflect their PrCa specificity ., In addition to the seven cases with truncating/deleterious mutations , we found “likely/potentially pathogenic” missense mutations in 11 PrCa families ., Taking into account the diversity and general high concordance of the in silico tools that were considered for the prediction of variant pathogenicity ( S2 Table ) , along with the low frequency found among the 710 healthy control cases screened ( S3 Table ) and with the fact that other missense mutations in ATM , CHEK2 and TP53 have been linked with cancer development 61 , 62 , it is plausible that the variants in ATM , BRIP1 , CHEK2 and TP53 here identified may explain PrCa susceptibility in the families carrying them ., The pathogenic nature of the CHEK2 mutation c . 349A>G found in two cases , was suggested in several studies , showing loss of DNA damage response and impaired activation due to lack of phosphorylation 48 , 49 ., Furthermore , this CHEK2 variant has been found in three of 694 BRCA1/BRCA2-negative BrCa families , two from the United Kingdom and one from the Netherlands , being described as a moderate to low penetrance variant 63 ., On the other hand , in a large case-control study gathering data from three consortia participating in the Collaborative Oncological Gene-environment Study ( COGS ) , the CHEK2 variant c . 349A>G was associated with increased BrCa risk ( odds-ratio 2 . 26 ) , but not with an increased risk for PrCa or OvCa 43 ., Segregation analysis and/or phenotypic evaluation in vitro would be useful to complement the available information concerning the pathogenicity of this and the remaining missense variants here identified ., For the variants found significantly associated ( or showing borderline significance ) with PrCa development , namely the ATM variant c . 8560C>T and the CHEK2 variant c . 349A>G , larger cohorts of familial/early-onset PrCa cases would be useful to define cancer risk estimates and the age-standardized PrCa risk attributed to these variants ., Looking at the overlap between the patients harboring truncating/functionally deleterious mutations and those harboring “likely/potentially pathogenic” missense variants , the ATM variant c . 8560C>T , found in three patients ( HPC3 , HPC186 and HPC332 ) , is the only variant overlapping with other mutations , namely in patient HPC186 , who carries the RAD51C frameshift mutation , and in patient HPC332 , who carries the MSH6 c . 1729C>T variant ., Immunohistochemistry analysis for MSH6 in the tumor of patient HPC332 showed normal MSH6 expression , thus reducing the likelihood that the MSH6 c . 1729C>T variant is a PrCa risk factor and rendering the ATM mutation c . 8560C>T the most likely risk variant in this patient ., Excluding the case HPC186 ( with co-occurrence of the RAD51C frameshift mutation ) , ATM represents the most commonly mutated gene in our series , eventually explaining increased risk of PrCa in seven cases ( ~5 . 8% ) , with CHEK2 being the second most frequently mutated gene ( four cases , ~3 . 3% ) ., Overall , functionally deleterious/“likely/potentially pathogenic” variants were found in 18 patients ( excluding the two families with missense mutations in MSH2 and MSH6 ) ., Of these , eight patients ( 44 . 4% ) fulfill the A criteria and 12 ( 66 . 7% ) fulfill the B criteria ( two cases complying with both ) , representing 17 . 8% and 13 . 9% of the samples enrolled in each group ., Seven of the 18 cases ( 38 . 9% ) were diagnosed at early age , representing 10 . 9% of the patients in the early-onset group ., Comparing clinicopathological data from patients harboring these variants with the group of patients without an identified “potentially pathogenic” mutation , no statistically significant associations were found ., In the context of this study , we identified one truncating variant in a gene that is included in the list of incidental findings recommended for return to patients after clinical sequencing by the guidelines of the American College of Medical Genetics and Genomics 64 ., The previously unreported MSH6 in-frame mutation c . 3846_3860del predisposes to Lynch syndrome ( OMIM #120435 ) , as it has been classified as pathogenic in two Lynch syndrome families in our institution , with evidence that included demonstration of loss of expression restricted to MSH6 in the colorectal tumors of carriers , a pattern also observed in the colon cancer of a relative of this patient who is also carrier of this in-frame MSH6 variant ., On the other hand , as no loss of MSH6 expression was observed in the prostate tumor , this MSH6 variant is unlikely to explain the PrCa predisposition in this family , which is most likely related to the RAD51C truncation mutation or the ATM missense mutation also found in this patient ., Even though targeted sequencing was performed under a research protocol and not as part of clinical sequencing , this incidental finding was reported to the patient during genetic counseling , as recommended by the guidelines of the American College of Medical Genetics and Genomics , and appropriate follow-up is being offered to the family as judged clinically appropriate ., In conclusion , we found functionally deleterious/“likely/potentially pathogenic” germline mutations in 18 of the 121 ( 14 . 9% ) familial/hereditary and/or early-onset PrCa cases selected for this study ., To our knowledge , this study is the first to report functionally deleterious germline mutations in the three FA genes RAD51C , FANCD2 and FANCI , and in two genes until now only associated with recessive disorders , CEP57 and RECQL4 ., Further data will be necessary to confirm the genetic heterogeneity of inherited PrCa predisposition hinted in this study ., This study is in accordance with the ethical standards of the Ethics Committee of the Portuguese Oncology Institute of Porto ( approval number 38 . 010 ) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards ., We selected 121 cases from our previously described series of 462 early-onset and/or familial/hereditary PrCa cases 17 , with two groups being considered: A ) cases with familial/hereditary PrCa , and B ) cases with early-onset PrCa and/or association with other types of cancer ., Among the cases in the A group , three criteria were defined:, 1 ) cases with at least three first-degree relatives with PrCa independently of the ages at diagnosis ,, 2 ) cases with two first-degree relatives with PrCa with average age at diagnosis ≤65 years and at least one of the affected cases diagnosed before the age of 61 , and, 3 ) cases diagnosed before the age of 61 with at least two first- or second-degree relatives with PrCa and average age at diagnosis of the three younger cases ≤65 ., Regarding the cases in the B group , three criteria were considered:, 1 ) cases diagnosed before the age of 56 years with at least three first- or second-degree relatives diagnosed with cancer and average age at diagnosis of the three younger diagnoses ≤55 ,, 2 ) cases diagnosed with second primary cancers besides PrCa and, 3 ) cases with relatives diagnosed with either early-onset and/or rare cancer types ( bilateral breast , male breast , brain ) and/or clustering of other cancer types ( e . g . breast , colon , or gastric cancers ) ., Cases previously identified as harboring pathogenic mutations in known PrCa predisposing genes ( HOXB13 , BRCA2 and MSH2 ) were excluded from this case priorization 17 , 27 ., DNA previously extracted from peripheral blood leucocytes by standard procedures 17 was quantified using Qubit Fluorometer ( Thermo Fisher Scientific , Waltham , MA , USA ) ., We used as control samples 710 healthy individuals ( 391 males and 319 females; mean age 55 . 1 years; SD±9 . 4 years ) , including 528 blood donors ( 285 males and 243 females ) from the Portuguese Oncology Institute of Porto with no personal history of cancer at the time of blood collection and 182 healthy relatives ( 106 males and 76 females ) with negative predictive genetic testing ( each from independent families ) ., We applied our previously established NGS pipeline 37 using the TruSight Rapid Capture target enrichment workflow and the TruSight Cancer panel , both from Illumina , Inc . ( San Diego , CA , USA ) ., For variant analysis , sequences were aligned to the reference genome ( GRCh37/hg19 ) using three different alignment and variant calling software: Isaac Enrichment ( v2 . 1 . 0 ) , BWA Enrichment ( v2 . 1 . 0 ) and NextGENe ( v2 . 4 . 1; Softgenetics , State College , PA , USA ) , as previously described 37 ., Briefly , for variant annotation and filtering , ., vcf ( variant call format ) files from the three software were imported into GeneticistAssistant ( Softgenetics ) and filtered for variant frequency in our in-house database , excluding variants present in more than 10% of the cases ., Additional variant selection included those with coverage >20x , alternative variant frequency between 30% and 70% ( excluding variants in mosaicism ) , and minor allele frequency ( MAF ) ≤0 . 1% 65 , 66 ., Synonymous variants and intronic variants at more than 12-bp away from exon-intron boundaries were excluded ., For MAF filtering , data was obtained from the 1000 Genomes Project Based on Project Phase III Data 67 , Exome Variant Server from NHLBI Exome Sequencing Project ( http://evs . gs . washington . edu/EVS/ ) , accessed in January , 2017 and Exome Aggregation Consortium ExAC ( http://exac . broadinstitute . org ) , accessed in January , 2017 databases , whenever available ., Variants assigned as not pathogenic , likely not pathogenic , of no clinical significance or of little clinical significance , according to public databases , namely ClinVar ( http://www . ncbi . nlm . nih . gov/clinvar/ , accessed in January , 2017 ) , Breast Cancer Information Core BIC ( https://research . nhgri . nih . gov/bic/ ) , accessed in January , 2017 ) , and InSiGHT ( via the Leiden Open-source Variation Database LOVD ( http://www . lovd . nl/3 . 0/home ) , accessed in January , 2017 68 ) , were discarded ., All the variants identified were validated by Sanger sequencing ., For this purpose , primers ( S6 Table ) were designed using the Primer-BLAST design tool from the National Center for Biotechnology Information ( NCBI ) 69 ., For PCR amplification , an initial denaturation step was performed at 95°C for 15min , followed by 35 cycles with denaturation at 95°C for 30s , annealing at appropriate temperature ( 58–62°C ) for 30s and extension at 72°C for 45s ., A final extension step at 72°C for 9min was included ., For the sequencing reaction , the BigDye Terminator v3 . 1 Cycle Sequencing Kit ( Thermo Fisher Scientific ) was used , according to the manufacturer’s instructions , and samples were run in a 3500 Genetic Analyzer ( Thermo Fisher Scientific ) ., For validation of the RECQL4 variant , primers and PCR conditions from Nishijo et al . were used 70 ., The TP53 variant was validated following the IARC ( International Agency for Research on Cancer ) protocol for direct sequencing ( http://p53 . iarc . fr/ , update 2010 ) ., Primers and PCR conditions for Sanger sequencing validation of MSH2 and MSH6 variants were kindly provided by Professor Michael Griffiths from the West Midlands Regional Genetics Laboratory , Birmingham Women’s NHS Foundation Trust , Birmingham , United Kingdom ., To explore the functional consequence of truncating/deleterious variants , MutationTaster 71 and Uniprot 72 were queried ., To infer the putative impact on splicing , the splice site predictors Human Splicing Finder 3 . 0 73 , MaxEntScan 74 , NNSPLICE 75 and NetGene2 76 were used ., To predict the biological impact of missense mutations , we looked at data from the predictor tools embedded in the NGS Interpretative Workbench from GeneticistAssistant , which includes the functional predictors SIFT , PolyPhen2 , LRT , MutationTaster , PROVEAN , FATHMM , CADD , MutationAssessor , MetaLR , MetaSVM and VEST3 , and the conservation analysis tools PhyloP , GERP++ , PhastCons and SiPhy , as previously described 37 ., To search for clinicopathological associations between mutation carriers and non-carriers , information on PSA at diagnosis , tumor staging and Gleason Score were gathered from medical records ( S4 Table ) and the Fisher’s exact test was used ., To evaluate the frequency in the general Northern Portuguese population of the missense variants identified in our series of PrCa patients we used KASP technology genotyping ( KBioscience , Herts , UK ) in our series of 710 healthy individuals , following manufacturer’s recommendations ., KASP assay primers ( S6 Table ) were designed using the Primer-BLAST design tool from NCBI and data were analyzed in the LightCycler 480 Software 1 . 5 . 0 .

Introduction, Results, Discussion, Material and methods

Considering that mutations in known prostate cancer ( PrCa ) predisposition genes , including those responsible for hereditary breast/ovarian cancer and Lynch syndromes , explain less than 5% of early-onset/familial PrCa , we have sequenced 94 genes associated with cancer predisposition using next generation sequencing ( NGS ) in a series of 121 PrCa patients ., We found monoallelic truncating/functionally deleterious mutations in seven genes , including ATM and CHEK2 , which have previously been associated with PrCa predisposition , and five new candidate PrCa associated genes involved in cancer predisposing recessive disorders , namely RAD51C , FANCD2 , FANCI , CEP57 and RECQL4 ., Furthermore , using in silico pathogenicity prediction of missense variants among 18 genes associated with breast/ovarian cancer and/or Lynch syndrome , followed by KASP genotyping in 710 healthy controls , we identified “likely pathogenic” missense variants in ATM , BRIP1 , CHEK2 and TP53 ., In conclusion , this study has identified putative PrCa predisposing germline mutations in 14 . 9% of early-onset/familial PrCa patients ., Further data will be necessary to confirm the genetic heterogeneity of inherited PrCa predisposition hinted in this study .

Prostate cancer ( PrCa ) is the most frequent cancer diagnosed in men worldwide , estimated to be responsible for the death of 27 , 540 men in the United States in 2015 ., Contrarily to other cancer types , the genetic contribution to the 10–20% of PrCa cases occurring in families with aggregation of the disease is largely unknown ., Germline mutations in the BRCA2 and the MSH2 breast and colon cancer predisposing genes , respectively , explain only about 1 . 5% of our early-onset/familial PrCa cases ., Taking advantage of recent deep sequencing technologies and an analysis pipeline established in our group , we have screened 121 PrCa patients with strong evidence of an hereditary component for mutations in 94 genes involved in several cancer predisposing syndromes ., We found truncating/functionally deleterious mutations in seven genes and “likely pathogenic” missense variants in four genes , of which five and one , respectively , have not been previously associated with PrCa predisposition ., We believe this study significantly contributes to the understanding of the genetic heterogeneity behind early-onset/familial PrCa .

cancer detection and diagnosis, deletion mutation, medicine and health sciences, cancer risk factors, genetic diseases, oncology, mutation, hereditary nonpolyposis colorectal cancer, nonsense mutation, mutation databases, frameshift mutation, autosomal dominant diseases, research and analysis methods, biological databases, clinical genetics, genetic causes of cancer, diagnostic medicine, heredity, database and informatics methods, genetics, biology and life sciences

journal.pcbi.1000340

Fuzzy Logic Analysis of Kinase Pathway Crosstalk in TNF/EGF/Insulin-Induced Signaling

A variety of modeling methods can be applied to understanding protein signaling, networks and the links between signals and phenotypes 1 ., The choice of modeling, method depends on the question being posed ( e . g . , mechanistic or phenotypic ) , the, quality and type of experimental data ( quantitative or qualitative ) , and the state, of prior knowledge about the network ( interaction map or detailed biochemical, pathway; Figure 1 ) ., Abstract, techniques are largely data-driven and aim to discover correlations among signals or, between signals and cellular phenotypes 2–4; these, methods include principal component analysis ( PCA ) and partial least-squares, regression ( PLSR ) ., Mechanistic differential equation-based models , in contrast , are, highly specified and dependent on extensive prior knowledge about components and, their interactions , but have the advantage that they capture temporal and spatial, dynamics at the level of individual reactions 5–9 ., Between, these extremes , modeling methods such as Bayesian statistics , hidden Markov models ,, and logic-based models have been used to construct graph-based representations of, influences and dependencies among signals and phenotypes based on experimental data, 10–18 ., An advantage of, these methods is their applicability to situations in which mechanistic information, is incomplete or fragmentary but the notion of a network of interacting biochemical, species is nonetheless informative ., Moreover , logic-based models use natural, language to encode common logical statements such as “if the kinase is not, active or the phosphatase is overexpressed , the substrate is not, phosphorylated” ., Logic-based models are commonly depicted as edge-node, graphs in which interactions among species occur at nodes , with gates specifying the, logic of the interactions based on a set of specified rules ., The identities of the, gates are typically determined based on prior knowledge or experimental observables, and the input-output relationships of each gate inferred from experimental data, 11 ,, 12 ,, 19–24 ., Among logic-based methods , the simplicity of Boolean models makes them attractive as, a means to render biological networks ., For example , a discrete-state representation, of the level of phosphorylation of insulin receptor substrate 1 ( IRS-1 ) at serine, 636 ( IRS ( S ) ) might use three input edges for time , TNF and EGF ( see below ) , one, output edge for IRS ( S ) , and one logic gate ( where “1” means, present or active , and “0” absent or inactive; Figure 2A ) ., Time is included as an, input variable to enable the representation of transient responses , following, cytokine treatment , for example ., In Boolean logic , interactions among inputs are, cast as combinations of elementary “AND” ,, “OR” , and “NOT” gates that generate logic, rules such as “ ( EGF OR TNF ) AND ( NOT ( time ) ) ” and are most easily, specified using truth tables ( Figure, 2B–C ) ., Truth tables consist of lookup values for the outputs, ( consequent value ) based on all possible combinations of input values ( antecedents ) ., Despite the appeal of Boolean models a two-state “on-off”, representation of many biological signals is quite unrealistic 25–27 ., In this work , we propose fuzzy logic ( FL ) as an approach to logic-based modeling with, the easy interpretability of Boolean models but significant advantages 28, including the ability to encode intermediate values for inputs and outputs ., We show, that FL can encode probabilistic and dynamic transitions between network states so, as to create simple and fairly realistic depictions of cell signaling networks 20–23 , 29–31 ., A key advantage of, logic-based approaches , also exemplified by FL , is the ability to construct models, ad hoc based on knowledge of network topology and data 32–36 ., Reverse engineering, models from data is an alternative and complementary approach , which is less biased, by a priori knowledge and assumptions , and is particularly useful, for identifying plausible topology and parameterization given quantitative data, gathered under several perturbations ., Here , we focused on building models by hand, because our goal was to test whether FL methods could be adapted to test a, priori knowledge and hypotheses against data to refine our, understanding of the network and generate testable hypotheses ., We complement our, initial model with model optimization to compare the effects of fuzzification ., Several means to refine Boolean models have been described , including kinetic logic, and the closely related piecewise-linear differential equations systems 22 , 37 , 38 ., Some, of these extensions rely on a differential equation system coupled to the Boolean, network to handle continuous variables ., The resulting models share common, steady-state behavior with the underlying Boolean system ( which is especially, useful , for example , in development and cell cycle studies ) 39 , but take longer to, simulate since they involve solving differential equation systems rather than, look-up tables ., Like fuzzy logic , dynamic Bayesian networks ( BN ) ( and the related, probabilistic Boolean networks 40 ) are able to handle data in a non-discrete, fashion , and have been used extensively to reverse engineer biological networks and, to model uncertainty in signaling networks 4 , 13 , 41 , 42 ., However , the, theoretical foundations are very different from those of FL: BNs are based on, probability distributions , in contrast to membership functions in FL ( see below ) ., Accordingly , the interpretation is also significantly different: BNs assign a, probability that a particular interaction exists ( with pre-defined weights ) , while, FL assigns rule weights to describe the interactions thought to be present ., We argue, that FL models represent a useful addition to the set of mathematical methods, available for analyzing complex cellular biochemistry ., The death-survival decisions made by mammalian cells in response to environmental, stimuli , such as those examined in this paper , are mediated by the integrated, activities of multiple receptor-dependent and cell-intrinsic processes that, coordinate opposing pro- and anti-apoptotic signaling ., We have previously described, a “cue-signal-response” ( CSR ) compendium of protein signals and, phenotypic responses in HT-29 human colon carcinoma cells treated with combinations, of tumor necrosis factor-α ( TNF ) , epidermal growth factor ( EGF ) , and insulin, 43 ., The compendium includes ten measurements of protein modification states, ( phosphorylation and cleavage ) and kinase activities for four proteins downstream of, TNF , EGF and insulin receptors collected over a 24 hr time period in biological, triplicate ., To date we have used PLSR to predict the phenotypic consequence of, perturbing the signaling network 44 and PCA to identify autocrine feedback circuits, 45 ., In this paper we explore the ability of a manually assembled multi-state FL model to, encode the dynamics of a complex intracellular signaling network ., We find that key, features of FL , such as non-discrete input-output relationships ( membership, functions – see below ) and the possibility that more than one relationship, can be invoked at the same time results in a remarkably intuitive representation of, biology ., It was therefore possible to generate new biological insight into the, regulation of IKK ( IkB kinase ) and MK2 ( mitogen-activated protein kinase-activated, protein kinase 2 ) kinases simply by inspection of the model ., A closer fit between, the FL model and data could presumably be achieved by automated regression ., As a, step in this direction we converted the multi-state FL model into a 2-state FL model, that could be calibrated against data ., The calibrated 2-state FL model exhibited a, better fit to data than a discrete model having the same degrees of freedom ., The, calibrated 2-state FL model also exhibited a better fit than the manually assembled, multi-state FL model , but only at the cost of less interpretability ., Overall we, conclude that manual assembly of FL models is an effective means to represent signal, transduction and derive biological insight; development of new approaches to, automated model fitting should also make FL models effective tools for, prediction ., Working from a normalized heat map of CSR data and the pathway scaffold from, Gaudet and Janes et al . ( Figures, 3–5 ) 43 , 44 ,, gates were manually constructed for signals such as phosphorylation , activation ,, or total protein levels ( Figure, 3 , Figure 4B ) ., These, intracellular proteins in the model include MK2 , c-jun N-terminal kinase ( JNK ) ,, extracellular signal-regulated kinase ( ERK ) , Akt , IKK , Forkhead transcription, factor ( FKHR ) , mitogen-activated protein kinase kinase ( MEK ) , IRS-1 , cleaved, caspase-8 ( Casp8 ) , and pro-caspase-3 ( ProC3 ) ., The first five measurements, characterize central nodes in five canonical kinase pathways governing, epithelial cell death: FKHR is a transcription factor downstream of Akt; MEK is, a kinase directly upstream of ERK; IRS ( S ) and IRS ( Y ) represent modifications of, insulin receptor substrate ( IRS ) by insulin receptor; and cleaved-caspase-8 is, the active form of the initiator caspase that cleaves caspase-3 , an effector, caspase responsible for degrading essential cellular proteins , activating CAD, nucleases and killing cells ., To illustrate how FL was used to model an intracellular signaling protein ,, consider the gate describing control of IRS-1 phosphorylation at serine 636, ( IRS ( S ) ) by EGF and TNF ( Figure, 2F–H ) ., For IRS ( S ) , the inputs were TNF concentration , EGF, concentration , and time , and the output was the level of IRS ( S ) phosphorylation ., The input and output activities were normalized between 0 and 1 for simplicity ., For example , in the IRS ( S ) gate , TNF concentrations of 0 , 5 , and 100 ng/mL were, normalized to 0 , 0 . 5 , and 1 as input values to the FL gate ( see Methods ) ., Because we do not explicitly model, biochemical processes such as receptor downregulation that make signals, transient , some of the FL gates had an input corresponding to time ( more, generally , this approach makes it possible to model dynamical processes using a, logical framework ) ., In the CSR data , “low” times refer to, early signaling responses ( 0–2 hr ) while “high”, times refer to late signaling events ( 2–24 hr ) ., Membership functions, were defined to transform input values to the DOM for each state ., For IRS ( S ) ,, the EGF input has low ( L ) and high ( H ) states ( Figure 2F ) ., When normalized EGF activity was, ∼0 , the gate assigned a high ( ∼1 ) DOM to L and low ( ∼0 ), DOM to H . As the EGF activity increased to 0 . 5 ,, DOM\u200a=\u200a0 . 5 for both L and H . The output level, classes ( L and H ) were treated as constants ( see Figure 2F ) ; MFs were unnecessary here because, gradation of the output was obtained during defuzzification ( see below ) ., Once, the membership functions had been defined , logic rules were listed as, “if A ( the antecedent ) , then B ( the consequent ) ” statements, using the inputs and output states as descriptors; e . g . , rule 2: if TNF is H and, time is L then IRS ( S ) is H ( Figure, 2F ) ., Each rule had an associated weight factor between 0 and 1 , which, was used to quantify the relative importance of the rules ., To compute the output of a gate for a given set of input values , we first, fuzzified the input variables ( see two examples in Figure 2G–H and described in text, below ) ., Next , each rule was evaluated , and a DOF was calculated as the minimum, of the DOMs for the inputs and the rule weight 28 , 46 ., Finally , the outcomes of each rule fired were resolved into a net output value, by defuzzification that involved computing the weighted average of the rule, consequences ( see Methods ) ., By way of, illustration , consider the two input value scenarios in Figures 2G–H ., In scenario 1 ( Figure 2G ) ,, EGF\u200a=\u200a1 ( that is DOM to MF, H\u200a=\u200a1 ) ,, TNF\u200a=\u200a0 ( DOM to MF, L\u200a=\u200a1 ) , and, time\u200a=\u200a0 . 27 ( DOM to MF, L\u200a=\u200a0 . 4 and, H\u200a=\u200a0 . 6 ) ., Rule 1 fired entirely ( output IRS ( S ), was L ) while rules 5 and 6 fired partially because time had partial membership, to L and H ( antecedents for rules 6 and 5 , respectively ) ; rules 2 , 3 , and 4 did, not fire to a meaningful extent ., Combining all these , the aggregate gate output, was ∼0 . 2 , an intermediate value between the full L output from rule 1, and the partial H output from rules 5 and 6 ., In contrast , scenario 2 ( Figure 2H ) shows a condition, ( EGF\u200a=\u200a0 ,, TNF\u200a=\u200a1 ,, time\u200a=\u200a0 . 19 ) that led to full firing of rule 4, ( though this rule has a weight of 0 . 25 ) , partial firing of rules 2 and 3 , and, negligible firing of rules 1 , 5 , and 6 ., The aggregate gate output in this case, was ∼0 . 5 ., To model CSR data 43 , eleven gates were constructed , each, comprising 2–4 inputs , 2–4 MFs per input , and 2–3, outputs ( see Figure 3 ) ., The, precise structure of each gate was based on the network scaffold , as described, above ( Figure 4A ) ., We aimed, for as few inputs , rules , and MFs as possible while still allowing a good fit to, data ., The parameter values for MFs and rules were fit manually to data but, future implementation of machine-learning algorithms or automated fitting would, improve the speed and accuracy of the process ( see below ) ., By way of, illustration consider the JNK and MK2 pathways , which are activated by stress, and cytokine treatment and are thought to be co-regulated following EGF or TNF, treatment ( Figure 4A , 47 ) ., During the course of constructing gates for JNK and MK2 , we found that the data, could be modeled without knowing whether or not cells had been treated with EGF, or insulin , suggesting that activation of JNK and MK2 was independent of ligand, addition ( Figure, 3B–C ) ., In some cases , gates based on the pathway scaffold were, insufficient to yield a reasonable fit to data and major changes were required, in the number and/or types of inputs ., For example , IRS-1 is the canonical, adapter protein downstream of the insulin receptor , though some of its many, phosphorylation sites are also substrates of other receptor kinases , including, EGFR 48 ., In modeling IRS-1 phosphorylation at two sites ,, tyrosine 896 ( IRS ( Y ) ) and serine 636 ( IRS ( S ) ) , we observed that both were, regulated by TNF and EGF but not by insulin ( Figure 3F and 3J ) ., The rules indicate that, both TNF and EGF treatment induce S636 phosphorylation while TNF inhibits, EGF-induced phosphorylation at Y896 ( see Text S1 ) ., During construction of an FL gate for Akt , we included inhibitory crosstalk from, ERK to Akt because it has been observed in several experimental settings 49–51 ., The introduction of, crosstalk greatly simplified the rule-base of the Akt gate , suggesting that this, crosstalk exists in HT-29 cells ( Figure 3I ) ., The mechanistic basis of crosstalk is not fully , and our, model includes a short time delay from ERK to the Akt gate input ., Negative, crosstalk from the ERK to Akt pathways may be the mechanism by which TNF, inhibits Akt phosphorylation upon insulin treatment , as observed by Gaudet et, al ., 43 ., A model with four inputs ( TNF , EGF , insulin , and time ) and describing the full, CSR dataset was constructed by joining together individual gates specified using, the approach described above ., Time delays were incorporated to model slow, processes such as the induction of transforming growth factor-α, TGF-α by TNF stimulation 45 ., TGF-α ,, which acts in an autocrine fashion ( not shown ) was united with the EGF input by, taking the maximum value across both signals at each point in time ( using the, “MAX” function ) , as these ligands bind, the same receptor and both affect MEK and Akt FL gates ( Figure 4B ) ., To compute model output , a, simulator stepped through small time steps , updating inputs to each gates at, successive steps ( see Methods ) ; model state, was then recorded at twelve equal time intervals corresponding to the, experimental time points ., Figure 5A depicts heatmaps of, the CSR dataset and the FL model , and shows that our FL model recapitulated most, major features of the CSR dataset across ten cytokine combinations ( Figure 5A ) ., For most inputs ,, the difference between simulation and experimental data were small , averaging, ∼2 . 2% , over the entire CSR data set ( as defined by the root, mean square deviation normalized by the mean of the data ) ., Common to all, predicted signals was the absence of a delay in activation after cytokine, stimulation ( Figure 5 ) ., To, model this delay would require an additional MF for several gates , a feature we, omitted for simplicity ., It was also challenging to model FKHR phosphorylation ., Even though Akt is known to regulate FKHR 52 , the model did not, effectively match data when Akt was the sole input to the FKHR gate; thus , we, modeled FKHR as having inputs from TNF , EGF , insulin , and time ( Figure 3H ) ., This suggests that, in HT-29 cells , FHKR is subject to more complex regulation than simply, activation by Akt ., One way to evaluate the performance of a model is to ask whether it can correctly, predict data that are not part of the training set ., Data describing the response, of HT-29 cells to co-treatment with TNF and C225 , an antibody that blocks ligand, binding to the EGF receptor , was not used to assemble the multi-state FL model ., We therefore asked whether the FL model could predict the effect of C225 as, compared to treatment with TNF alone ., Because EGFR is activated both by, exogenous EGF and autocrine TGF-α ( whose production is induced by TNF, 45 , 53 ) we modeled the, effect of C225 addition by disabling the MAX function, downstream of TNF and EGF ( recall that this gate is present to model activation, of EGFR not only by exogenous EGF but also by TNF-dependent release of, TGFα , which acts in an autocrine manner ) ., The model correctly predicted, that cotreatment with TNF and C225 would reduce Akt , MEK , and ERK signals as, compared to treatment with TNF-alone ( “−“ vs, “+”, C225 in Figure 6 ) ., However , the model did not predict, decreases in MK2 and JNK signaling because the MAX function downstream of EGFR, activity was not connected to the MK2 and JNK pathways , which are thought to be, downstream of TNF but not TGFα or EGF stimulation 45 ., We can reinterpret, our initial assumptions that TGFα signaling only affects Akt and ERK ., The other MAP kinases measured ( MK2 and more noticeably JNK ) exhibited less, activation in the presence of C225 ., Likewise , late IKK signaling was decreased, and slightly more caspases were cleaved compared to C225 alone , but these, effects were not predicted by our model ., The discrepancy between the model and, data suggest that MK2 , JNK , and IKK are activated in part by TNF via, TGFα by either a direct effect of EGFR or through crosstalk with the Akt, and ERK pathways ., Our model enabled us to predict some of the effects of C225 in, interfering with TNF signaling while providing context to revise our, understanding of TNF-induced signaling through EGFR in the MK2 , JNK , and IKK, pathways In the work described above , logic rules and membership functions for each gate, were established manually ., A better approach is to use training to optimize the, weights of all possible rules in a gate by minimizing the sum of the squared, differences between the experimental data and local model output ( see Methods ) ., Following optimization , logic rules, that are supported by the data should have weights near 1 , while, poorly-supported rules should have weights near 0 ., We tested the fitting, algorithm on the MK2 gate ., For such a gate , which has two MFs each for the two, inputs ( TNF and time ) and the output ( MK2 activity ) ,, 23\u200a=\u200a8 explicit rules are possible ., MK2 data from the 10 cytokine treatment conditions were used to optimize a, vector containing the 8 rule weights ., Our initial optimization attempt failed, because time-dependent MFs were not parameterized so as to capture rapid, increases in signals following cytokine treatment ., We had implicitly ignored, this discrepancy when fitting the model by hand ., To improve the automated, fitting procedure , an additional MF for time was included to represent, immediate-early responses , increasing the number of candidate rules to 12 ., Optimization yielded a gate with a good fit to data using only six rules with, weights near one ( Figure, 7A ) ., These six rules were identical to those assembled manually with the, exception of the new rule needed to represent immediate early signaling ( Figure 7B ) ., To test FL gate, regression with more rules , we applied the algorithm to the same MK2 data using, one additional membership function ( for medium activity levels ) and compared it, to an untrained model using the same MFs ., The training process created several, rules that were nearly identical to those introduced manually as well as several, new ones ( Figure, S1 ) ., The MK2 test case suggests that it is possible to optimize rule, weights as a means to fit logic rules without bias and is a first step towards a, more rigorous approach to logic-based modeling ., To compare FL and discrete models we converted our FL model to a multi-state, discrete model ( DL ) by leaving the rules , rule weights and MF thresholds the, same and changing the degree of fuzziness of the MFs so as to make the model, discrete ( Figure 2D , Methods ) ., Resulting FL and DL models are, therefore identical except in a single global parameter ( the degree of, fuzziness ) making direct comparison possible ., More than one rule could fire at, the same time in both the FL and DL model , making defuzzification necessary in, both ( see Figure, S2 ) ., Thus , the DL model was not a conventional Boolean model ., To measure the goodness of fit of FL and DL models , we computed the sum of, squared differences ( RSS ) and normalized RSS ( see Methods ) ., The FL model consistently exhibited a better fit to the, data than the DL model ( absolute deviation of 44 . 6 and 96 . 7 , and normalized, deviation of 0 . 035 and 0 . 076 , respectively ) ., When we compared simulated and, actual data we observed cases in with FL models were better than DL models ,, cases in which they were similarly effective and cases in which neither did a, good job in fitting data ., In general , DL models were less effective than FL, models in capturing intermediate activity levels ( Figure 5B ) ., For example , in the DL model ERK, activity alternated between low and high while in the FL model ERK activity was, graded , as it was in experimental data ( Figure 5A ) ., More striking breakdowns between, the DL model and data were observed for IRS ( S ) , JNK and Akt , ( Figure 5A ) ., For IRS ( S ), transient activation was missing from in the model for 1 of 5 cytokine, treatments and for JNK it was missed for 3 of 6 treatments However , DL models, effectively capture step functions and they are therefore well suited to sharp, transient signals ( Figure, 5C ) ., We also observed cases where both models failed to fit the data ,, especially when two peaks of activity were observed ( Figure 5D ) ., This failure to fit data could be, remedied by adding more input states for time and by altering the rules ( Figure S2 ) ., To ensure that the superior fit of the FL model ( as compared to the DL model ) was, not biased because the FL model ( and not the DL model ) was manually assembled ,, we independently optimized simplified FL and DL models ., We performed a global, optimization with 8-fold cross-validation of the rule weights in 2-state FL and, DL models ( see below , Methods , and Figures, S2 ) ., These models contain two states for each input and the output in, every gate ., Optimization of the 2-state FL model improved the estimated error, compared to the 2-state DL model ( with averages and standard deviations of, 0 . 030±0 . 005 and 0 . 040±0 . 006 , respectively , using a, normalized fitness measure ( see Methods and, Figure, S2 ) ., Additionally , we converted the 2-state DL model to BL by converting, the rule weights to a binary value ( 0 or 1 ) ., We repeated the optimization but, over binary rule weights for the BL and FL 2-state models ., The cross-validated, error of the binary-weighted FL model was ∼50% lower as, compared to the BL model ( 0 . 056±0 . 01 and 0 . 083±0 . 01 ,, respectively ) ., We therefore find that a standard Boolean model has poorer, performance than the discrete model ( DL ) studied here ( see Figure 2E , Discussion , and Figure S2 ) ., The improved ability of the DL model ( as compared to the BL model ) to predict, data following optimization on a training set suggests that continuous rule, weights confer noticeable flexibility to the models ., As a second means to evaluate the multi-state FL model we looked for new and, potentially testable biological insights ( see also Text S1 ) ., In this paper we describe the assembly and evaluation of a fuzzy logic model of, mammalian signaling networks induced by TNF , EGF , and insulin ., The logic gates and, their associated membership functions , which encode input-output relationships for, interactions among various species in the model , were generated based on study of, cellular responses to different cytokine treatments ., The gates were then linked, together based on prior knowledge of network topology and parameterized using, induction or an automatic fitting process that minimized the difference between, simulated and experimental trajectories ., The resulting models were interpretable, with respect to known interactions from the literature , and they generated dynamic, trajectories for various signals that were similar to experimental data ., We can, therefore conclude that efficient assembly of a FL network able to encode complex, experimental data is possible ., By building different versions of a FL gate , we were able to intuit potential, biological interactions that had gone unnoticed during data mining with other, analytic tools ., For example , the FL model suggested that MK2 and MEK are, co-regulators of ERK ., This offers a new explanation for the previously published, observation that MK2 has pro-survival effects 21 ., Similarly , a link, between EGF treatment and IKK inhibition suggests that EGF-induced downregulation of, the EGF receptor might interfere with IKK activation by inhibiting, TGF-α-induced IL-1α autocrine signaling , which is dependent on EGF, receptor activity ., Thus , FL modeling yields predictions about the strength and logic, of direct and autocrine-indirect processes ., In the future , the process of choosing, the best FL model can be made more rigorous than what we have undertaken here by, automating the fit of rules and membership function to data; this would obviously, make the process of extracting hypotheses from models more rigorous ., As a starting point for optimizing FL models , we show that it is possible to fit the, rules for individual gates to experimental data ., This raises the general possibility, that logic-based models can be improved by global fitting procedures 60 , 61 ., Optimization algorithms such as genetic algorithms and Monte Carlo simulations can, be used to fit membership functions and rule weights simultaneously ( Figure S2 ) ., However , a critical step in optimization of FL models will be the development of, objective functions that balance complexity and goodness of fit to data ., Because, different parameter types encode diverse degrees of freedom , designing a balanced, metric will be challenging ., Should a model be penalized equally for binary and, continuous parameters , or for additional rule weights versus another membership, function ?, Answering these questions will likely require application of theories such, as Minimum Description Length and Vapnik-Chervonenkis Theory 62 ., These methods employ, statistical learning methods ( Vapnik-Chervonenkis Theory ) or data compression, through Turing-style languages ( Minimum Description Length ) to quantify model, complexity ., We have already observed that the capacity of multi-state discrete logic, gates to effectively capture quantitative data features can be increased by, including a greater number of memberships ( states ) ( see Figure S3 ) ., Therefore , either fuzzification or inclusion of additional states can strengthen a, DL model ., A solid metric of model quality would make it possible to compare FL and, BL models rigorously as well as evaluate models of the same processes that differ in, topology or MFs ., The fuzzy logic framework supports several mechanisms for flexibility including the, slope and shape of the membership functions , rule weights , fuzzification and, defuzzification procedures , and rule structure ., Here , we limited our fuzzification, of logic models to a subset of possible FL functions ., We used only one degree of, membership and one membership shape for entire models and chose the simplest, fuzzification algorithms and rule structures ., Most of the flexibility in our FL, models , as compared to BL models , arose from fuzzy memberships and continuous rule, weights that enabled multiple rules to fire simultaneously ., By optimizing four, variants of the 2-state model ( discrete or fuzzy memberships and continuous or, binary rule weights , Figure S2 ) , we were able to demonstrate that much of the ability of the, FL models to fit the CSR data arose by allowing rule weights to be continuous and, not binary ., Thus , DL models may be a useful alternative to BL models ., If DL models, use quantized rather than continuous rule weights , they are likely to achieve a, similar flexibility of fuzzified logic models while offering the benefit of faster, optimization and easier interpretability with fewer degrees of freedom ., We built models by both manually and automatically fitting model parameters ., Though, the latter achieved better fits to data , it came at the expense of a loss of model, interpretability ., Model building methods that balance rigor of automatic, optimization with the intuition gained with hand-curated models will be a key step, forward ., This might be achieved by optimizing quantized rule weights instead of, continuous values , or by penalizing models for intermediate weights ., Use of a, processing algorithm that simplifies sets of optimized rules by excluding those with, low weights or merging similar rules would ease the interpretability gap between, manually and automatically assembled models ., Specialized software that offers a more, limited subset of FL capabilities would also streamline model development and, improve the computational time required for parameter optimization ., In conclusion , the current FL model of TNF/EGF/insulin-induced signaling in HT-29, cells begins to explore the potential of FL methods to model cell signaling, networks ., the future , the improvement of automated model fitting , a graphical-user, interface tailored to biological applications , and better means to mine and, incorporate literature data

Introduction, Results, Discussion, Materials and Methods

When modeling cell signaling networks , a balance must be struck between, mechanistic detail and ease of interpretation ., In this paper we apply a fuzzy, logic framework to the analysis of a large , systematic dataset describing the, dynamics of cell signaling downstream of TNF , EGF , and insulin receptors in, human colon carcinoma cells ., Simulations based on fuzzy logic recapitulate most, features of the data and generate several predictions involving pathway, crosstalk and regulation ., We uncover a relationship between MK2 and ERK pathways, that might account for the previously identified pro-survival influence of MK2 ., We also find unexpected inhibition of IKK following EGF treatment , possibly due, to down-regulation of autocrine signaling ., More generally , fuzzy logic models, are flexible , able to incorporate qualitative and noisy data , and powerful, enough to produce quantitative predictions and new biological insights about the, operation of signaling networks .

Cells use networks of interacting proteins to interpret intra-cellular state and, extra-cellular cues and to execute cell-fate decisions ., Even when individual, proteins are well understood at a molecular level , the dynamics and behavior of, networks as a whole are harder to understand ., However , deciphering the operation, of such networks is key to understanding disease processes and therapeutic, opportunities ., As a means to study signaling networks , we have modified and, applied a fuzzy logic approach originally developed for industrial control ., We, use fuzzy logic to model the responses of colon cancer cells in culture to, combinations of pro-survival and pro-death cytokines , making it possible to, interpret quantitative data in the context of abstract information drawn from, the literature ., Our work establishes that fuzzy logic can be used to understand, complex signaling pathways with respect to multi-factorial activity-based, protein data and prior knowledge .

computational biology/systems biology, cell biology/cell signaling

journal.pcbi.1000514

A Computational Analysis of ATP Binding of SV40 Large Tumor Antigen Helicase Motor

Helicases are a family of ATPase motors that couple the energy of ATP binding and hydrolysis to conformation changes , which in turn is coupled to the unwinding and translocation of DNA 1 ., Simian Virus 40 ( SV40 ) large tumor antigen ( LTag ) is an efficient hexameric helicase that belongs to the helicase superfamily III , as well as the AAA+ protein family ., The high resolution structures of LTag hexameric helicase in different nucleotide binding states have been previously reported 2 , 3 , including the Apo , the ATP-bound and the ADP-bound states ., These three structures reveal an iris-like motion of the hexamer helicase during the drastic conformational switches that are triggered by ATP binding and hydrolysis ., Accompanying the iris-motion of the LTag hexamer is the longitudinal movements of the six β-hairpins along the central channel ., Despite the advancement in LTag helicase studies mentioned above , the detailed paths for these conformational switches and the corresponding energetics associated with the ATP binding process are unknown , which can be simulated by a computational approach using molecular dynamics and targeted molecular dynamics ., Molecular dynamics ( MD ) propagates the molecular system under the laws of classic mechanics 4 , 5 , and is suitable for studying conformational changes ., However , the current computational capability restricts the size ( molecular weight ) of the studied system and the time scale of MD simulation ., For the studies of larger and more complex systems , targeted molecular dynamics ( TMD ) has been used to accelerate the simulation , which adopts an additional holonomic constraint on the physical potential to reduce the root mean square deviation between the current structure and the final ( targeted ) structure 6 ., TMD is suitable to calculate the transition pathways between two known protein conformations ., The combination of MD and TMD methods have been widely applied to the dynamics studies of various systems , such as the Ras p21 in the signal transition pathway 7 , F1-ATPase system 8 , 9 , the GroEL complex 10 , and the human a-7 nAChR receptor 11 ., Here we adopted TMD to calculate the whole transition pathway and used MD to simulate the accurate conformational change in certain key time slots ., In order to understand the energetic aspects of ATP triggered conformational changes of LTag hexameric helicase , we simulated the ATP binding process of LTag and the associated conformational changes ., We first built an Apo state with six ATPs placed 20 Å away from the binding pocket of the original Apo structure ., Then we used the TMD approach to calculate the transition pathway from the Apo state to the ATP bound state and examined the ATP binding process that powered this conformational transition ., The results suggest an ATP molecule goes through a three-step process before being “locked” inside the nucleotide pocket ., Meanwhile , the configurations of the binding pocket along the ATP binding pathway were evaluated by using the linear response approximation ( LRA ) version of the semi-macroscopic protein dipoles langevin dipoles method ( PDLD/S ) , a method that is capable of evaluating binding free energies significantly faster than the microscopic methods with comparable accuracy 12 ., In addition , the simulation results of the conformational transition reveal a refined pathway for the cooperative iris-like movement of LTag hexamer helicase previously observed in crystal structures ., There are three high-resolution LTag hexameric helicase structures corresponding to different nucleotide bound states 2 , namely , the Apo state ( PDB ID 1svo ) , the ATP bound state ( PDB ID 1svm ) and ADP bound state ( PDB ID 1svl ) ., The hexameric helicase structure reveals two stacked hexamer rings with a central channel ( Fig ., 1 ( A ) ) ., Each LTag subunit of the hexamer contains three structural domains , D1 , D2 and D3 ( Fig ., 1 ( B ) ) ., D1 is the N terminal Zn domain essential for LTag hexamerization 2 , 13 ., D2 is a typical AAA+ domain with Walker A or p-loop and Walker B motifs , which is important for ATP binding 2 ., D3 is composed mostly of long helices , which is sequentially interrupted by D2 roughly in the middle of D3 , while D1 is structurally well separated from D2/D3 ( Fig ., 1 ( C ) ) ., The binding pocket is located at the interface between two adjacent monomers ., The monomer with the P-loop at a given interface is named cis-monomer , and the other monomer forming the interface is named trans-monomer ., For ATP to bind to the nucleotide pocket , the only possible route is through the opening between the two neighboring monomers ( or subunits ) from the C-terminal end ( bottom ) ., The binding pocket residues on the cis-monomers can be divided into two groups , the I428 , D429 , K432 , T433 , T434 on the P-loop , and the N529 , D474 on the Sensor I motif ( Fig ., 1 ( C ) ) ., The binding pocket residues on the trans-monomers include the arginine finger tR540 ( t designates trans ) and lysine finger tK418 , and residues tR540 , tD502 and tR498 ., The ATP interacts with the cis-residues and trans-residues mainly through its phosphate group and the ribose ., The adenosine group inserts into the hydrophobic pocket formed between two helices , h9 and h13 , on the cis-monomer ( Fig ., 1 ( C ) ) ., There are three major conformational transition stages of the LTag molecular motor , which is associated with the ATP binding stage , followed by the ATP hydrolysis and the ADP releasing stages ., In this report , we focus on the study of the ATP binding stage ., We have built a pre-Apo state model by placing six ATP molecules 20 Å away from the nucleotide binding pocket , and an ATP docking stage model by putting the ATP in the binding pocket of the Apo state ., We simulated a 1 ns ( nanosecond ) pathway from pre-Apo state to the Apo ATP binding states and a 1 ns pathway from pre-Apo to the ATP docking state and finally to the ATP binding state ., To study the LTag helicase overall conformational changes during ATP binding , we took a closer look at the ATP binding pocket by analyzing the trajectory involving the ATP ligand binding process ., In this section , we studied the conformational changes of the cis and trans residues involved in ATP binding , the movement of ATP , and the dynamic hydrogen bond formation during the ATP binding process ., The results of this study suggest a cross-locking model of the binding pocket for ATP binding ., The coordination of Mg2+ plays an important role in the ATP binding ., Similar experimental study of F0F1-ATPase shows that the addition of Mg2+ will increase the binding affinity of the nucleotide and helps to proceed to the tight binding state 8 , 17 ., The binding pocket residues coordinate with the Mg2+ ion directly or through some intervening water molecules ., Among these intervening water molecules , the apical water , WAT1 , near the γPi helps to stabilize the pocket residues , and attack the γPi during hydrolysis ., In our simulation , we observed the coordination of Mg2+ with the intervening waters during the ATP binding procedure ( Fig . 6 ) ., The Mg2+ has a strong propensity to assume an octahedral coordination 17 ., During the ATP docking stage , the Mg2+ ion forms a complex of six-element structure with the β , γ oxygen and four water molecules ., The whole complex ( ATP-Mg2+ and five coordinated water molecules ) docks into the binding pocket until the WS ., There is a flattening stage in the distance profile between cis-residue T433 and the Mg2+ ( Fig . 6 ) , which indicates that the T433 is searching for a best position to attack the Mg2+ in the complex ., At the binding transition stage , T433 begins to attack the Mg2+ ., The invasion pushes one of the coordinated waters , WAT3 , close to its neighbor WAT2 , which forces WAT2 to leave the stable coordination position with the Mg2+ cation ( Fig ., 6 ( C ) ) ., As we can see from Fig . 6 ( A ) , there is a steep decrease in the distance between T433 and Mg2+ together with a sharp increase of the distance between WAT2 and Mg2+ ., On the other hand , the distance variation between WAT3 and Mg2+ is subtle ( Fig ., 6 ( B ) ) ., The stable coordination distance between Mg2+ and ligand is ∼2 . 0 Å ., The coordination transition indicates that hydration waters may not necessarily be stripped at once ., As in the case of F1-ATPase , the ATP may progressively exchange its hydrogen bonds with the hydration waters for hydrogen bonds with the ATP-pocket residues 8 ., When the ATP-Mg2+ complex diffuse near the pocket , the negatively charged phosphate group will interact with positively charged or polar amino acids , such as Arginine ( R540 , R498 ) , Lysine ( K418 , K419 ) and Asparagine ( D502 , D474 ) ., However , at the beginning , these charged groups may form hydrogen bonds with waters ., When the ATP-Mg complex comes in , the waters may act as temporary bridges that should be weakened and broken with molecular vibrations during ATP-Mg2+ binding , and eventually be replaced and expelled by the ATP-Mg2+ complex ., However , some of these water molecules will act as bridges via hydrogen bonds between the charged amino acids and the ATP-Mg2+ during the entire binding process ., The 2 . 0 Å crystal structures of LTag in different nucleotide bound states also reveal some of these fixed water molecules in the binding pocket before and after ATP-Mg2+ binding ., Here , we focus on the apical water and the water molecules coordinated with Mg2+ since they are directly related to the hydrolysis of ATP ., Our experimental result show that the apical water is unusually coordinated through four residues: two cis-residues D474 , N529 and two trans-residues tR540 and tR498 ( Fig ., 6 ( D ) ) ., There is no particular order of coordination observed during the binding procedure ., The distance between the four coordinated residues and the oxygen of the γPi group varies until the shrinking stage , when all the coordination distances converge to a stable hydrogen bond distance around 3 . 5 Å ., The coordination procedure could be considered as a shrinking cage for the apical water ( Fig ., 6 ( D ) ) ., The vibration of the water molecule decreases until the cage shrinks to the stable state , at which point the apical water will be in a position ready for the nucleophilic attack in ATP hydrolysis ., The PDLD/S-LRA method is used to evaluate the binding energy of a series of 20 key snapshots ( intermediate structures ) sampled from the TMD simulation trajectory ., The results give a rough binding pocket energy profile between the Mg-ATP complex and binding pocket ., Fig . 7 ( A ) and ( B ) show results using dielectric constants of 20 and 40 respectively ., The calculated trends do not depend on the choice of protein dielectric constant ., The energy profile starts at −6 kcal/mol , which corresponds to the interaction between Mg-ATP complex and the water from the beginning ., And we use −6 kcal/mol as a base line to measure the binding energy ., There is an energy barrier of 8 kcal/mol from WS to TS ( Fig ., 7 ( A ) ) ., The time corresponds to the Mg2+ coordination exchange , where the WAT2 ( Fig ., 6 ( C ) ) escapes from its stable position due to the invasion of residue T433 ., The coordination transition is similar to the transition from the Mg-ATP diphosphate coordination state and Mg-ATP tri-Phosphate coordination state ., In the diphosphate coordination state , the Mg2+ coordinates with ATP through β and γ phosphates ., In the triphosphate coordination state , the Mg2+ coordinates with ATP through all three phosphates ., The transition energy barrier is around 11 kcal/mol ( 18 KbT ) in the water 18 , slightly larger than our simulation results ., One possible explanation is that the conformation of the binding pocket protein may facilitate the coordination transition of Mg2+ by decreasing the barrier about 3 kcal/mol ., This is followed by an energy valley of 13 kcal/mol , which lasts throughout the binding transition stage , and ends at the beginning of the shrinking stage ., Then comes another energy barrier of 5 kcal/mol ., There are three hydrogen bonds formed with N529 at this time ( Fig . 4 ) ., One bond is formed with the γPi group oxygen , the other is formed with the tR498 and the third is formed with the apical water ., The adjustment in the shrinking stage helps to prepare the apical water to attack the γPi in the following ATP hydrolysis stage ., The energy profile stablizes at −12 to −14 kcal/mol , and the binding energy is about 8 kcal/mol ., The experimental result of the TNP adenine nucleotide analogues binding energy is −8 kcal/mol ( −33 kJ/mol ) 19 , which could be used as a reference of our simulation results ., Therefore , we conclude that our simulation results fall within a reasonable range , in comparison with the previous studies 8 ., The major energy barrier is in the binding transition stage ., Most of the binding energy is released during the docking and binding transition stage ., On the other hand , most of the domain scale conformational changes happen after the binding transition stage ( Fig . 5 ) ., The sequence may imply that the domain scale is triggered by the ATP binding ., Similar models have been reported , for example , in the F0F1-ATPase model , the energy transduction takes places during the binding transition stage as well 8 ., Some recent studies of similar T7 helicase 20 , 21 also reported that the global conformational change is triggered either by ADP release or by ATP binding ., The motor domain engages with DNA after ATP binding 16 ., The above simulation results indicate that the domain wise conformational changes happen in the ATP binding transition and the shrinking stages ., The most significant conformational change is the D2/D3 domain movement towards the D1 domain ( or D2/D3 folding ) ., The major folding movement occurs in the shrinking stage of the ATP binding process , with ∼20% occuring in the ATP binding transition stage ., Our previous work has illustrated a ∼17 Å movement on the tips of the β-hairpin 3 ., In this simulation study , we found that these two movements can be derived from the angled D2/D3 folding movement toward the N-terminal D1 domain , with an angle of approximately 20° ., And the hinge point for the angled folding movement is around the joint of helix h5 and h6 ( Fig ., 8 ( A ) ) ., From the bottom view ( Fig ., 8 ( C ) ) , the folding pushes the ATP-interacting cis-residues in an anti-clockwise direction to the neighboring trans-residues to form the cross-lock interactions to lock the ATP in the binding pocket ., Fig . 8 ( C ) illustrates an interesting movement of the β-hairpin during the folding ., The tip of the β-hairpin moves upward along the central channel with a screw motion , which is consistent with the simulation results in section 1 ., Because the LTag monomer conformational changes triggered by ATP binding occur in the context of a hexamer , the six monomers within a hexamer have to cooperate with each other during the conformational switch ., Our simulation shows that the most significant cooperative movement is the formation of the ATP binding pocket and the concomitant domain-wise folding of D2/D3 in the first transition stage ., The cis-residues for ATP binding sit in the front and face towards the folding direction ., The folding movement pushes the cis-residues to the position with the shortest locking distance ( bonding distance ) with the corresponding trans-residues of its anti-clockwise neighbor monomer for ATP binding ., At the same time , the folding movement of the neighboring monomer slides the trans-residues , which are located at the right side of the monomer ( Fig ., 8 ( E ) , to the contacting position for the incoming cis-residues ., At the end of the cooperative movement , the two sides of the ATP binding pocket reach the shortest bonding distance to form the cross-locking interactions for ATP binding ( Fig ., 8 ( E ) ) ., Accompanying the folding , six β-hairpins rotate and move along their slant axes as illustrated in Fig . 8 ( D ) and Fig . 8 ( F ) ., The ATP binding pocket is located at the base of the β-hairpin , thus the folding movement triggered by ATP binding could be amplified through the lever effects of six β-hairpins and transferred to the tip residues ., The binding of six ATPs is therefore coupled with both the screw movements of the six β-hairpins towards the N-terminal in the central channel and the collective angle folding movement of the six D2/D3 domains towards the D1 domains , like an iris of the camera ( Fig ., 8 ( E ) ) ., However , we could not perform reliable computational analysis of the nature and extent of the cooperativity between the subunits within a hexamer at this time due to the lack of the experimental kinetics data on the cooperativity of LTag helicase ., ATP binding and hydrolysis by the LTag helicase motor is essential ., We have performed a simulation study of the ATP binding process by LTag helicase in order to understand the energetics of ATP binding and the associated conformational changes for LTag helicase function in DNA unwinding ., Based on our simulation results , we propose a cross-locking model for the ATP binding procedure for LTag helicase ., The binding model can be divided into three main stages , namely , the docking stage from Apo state to the weak binding state , the binding transition stage from weak binding state to tight binding state , and the shrinking stage from the tight binding state to the ATP bound state ., The first two binding stages are similar to the binding zipper model of the F1-ATPase system ., During the ATP binding process , the Mg-ATP complexes diffuse to the binding pocket in the docking stage ., And the phosphate group begins to interact with the binding pocket residues , such as the P-loop , and forms the conformation of WS ., In the WS conformation , the bonding interactions between the three pairs of lock residues are not formed , and the three locks are fully open and the adenine group is completely outside the pocket ., WS progresses to the TS during the binding transition stage ., The Mg-ATP complex progressively forms hydrogen bonds with the residues in the binding pocket through the phosphate group and the ribose ., These interactions induce the conformational changes of both ATP and the lock residues around the pocket ., For the ATP , as the adenine inserts into the hydrophobic gap between h9 and h13 , the dihedral angle between the adenine/ribose and the phosphate group increases about 150 degrees ., The ATP also bends down to a right angle ., For the binding pocket , the three locks close sequentially , first lock1 ( Ribose and LYS419 ) , then lock2 ( ASP429 and LYS418/419 ) , and finally lock3 ( ASP474 and ARG540 ) ., The hydrogen bond analysis shows that the Mg-ATP complex first interacts with the P-loop and cis-residues , and then forms hydrogen bonds with the trans-residues ., The major stablizing hydrogen bonds begin to form with the cis-residues on the P-Loop and N529 , then trans-residues tK418 and tR540 ., This corresponds with the results of the mutation study 15 and the ATP binding observation from the F0F1-ATPase system 8 , 14 ., The number of hydrogen bonds increases linearly in the binding process , which is consistent with the results of the zipper binding model 8 ., The apical water is important for the nucleophilic attack in ATP hydrolysis ., Our simulation shows the position of the apical water is stabilized during the shrinking stage ., The intra-ring conformational change and the relocation of residues compress the “cage” space around the apical water , and after certain adjustments , the coordinated residues are stabilized near the apical water in the ATP binding stage ., At the end of the binding transition , the gate of the ATP pocket is fully closed ., Negatively charged side chains , such as ILE428 , ASP429 , and ribose bases , all gather outside the gate ., This may help to prevent the approach and binding of the other ATP ( Fig ., 3 ( TS ) ) ., In the binding transition stage , all the significant movement is concentrated in the binding pocket ., Only ∼20% of the domain-wise conformational changes occur at this stage , which includes a subtle D2/D3 domain movement ., The major domain-wise conformational changes ( ∼80% ) is accomplished in the shrinking stage ., It is interesting to note that the radius of the central channel in the C-terminal bottom portion decreases more than the middle portion during the ATP-binding triggered conformational change , which could mean that the D2/D3 upwards movement toward the N-terminal D1 domain may generate a pushing force for moving DNA through the central channel ., This movement is part of the iris-motion of LTag hexamer associated with the ATP binding and hydrolysis processes ., All the simulation is calculated using the CHARMM program package 22 and the binding energy profile is calculated by the POLARIS module of the Molaris program package 12 ., The CHARMM27 all-atom force field 23 and the TIP3 water model 24 is employed ., The cutoff radius for the non-bonded interactions is 14 Å ., The SHAKE algorithm is adopted to fix the hydrogen bond during the simulation 25 ., We have built two models for Apo state LTag helicase ., The first Apo structure is built for TMD simulation of the ATP binding procedure ., Six ATPs extracted from the ATP bound state are placed 20 Å away from the original Apo state helicase structure3 ., The O software program is used to adjust the ATP spatial position 26 ., The ATPs are relaxed for 10 ps at 300 K . We use the Dowser program to place the internal water for the Apo structure 27 ., A water sphere of 70 Å is built to wrap around the Apo structure ., 36 chloride ions and 28 sodium ions are used to neutralize the system ., We quenched the system for 10000 steps and then equilibrated for about 500 ps from 0 to 300 k , this is followed by another 200 ps equilibration for 300 K . The second model is built to verify the concerted model of ATP binding ., The system is built by replacing one of the Apo state monomer with the corresponding ATP bound state structure ., The position of the new monomer is decided by aligning the D1 domain to that of the original Apo state monomer ., The system is quenched for 10000 steps and equilibrated for 500 ps from 0 to 300 k ., The ATP bound state is scanned by the Dowser program to place the missing inner water ., A 70 Å TIP3 water sphere wraps the system ., Again , we quench the system for 10000 steps and equilibrated it from 0 to 300 K . The TMD simulation used an additional energy term based on the RMSD of the initial structure and final ( target ) structure ., The energy term has the form: , where k is the force constant ( 20 kcal·mol−1·Å−2 ) , RMSD ( t ) is the root means square distance of the current simulated structure from the target structure , and RMSD* ( t ) is the predefined target RMSD value at time t ., Since the forward and backward trajectory pathways are supposed to be the same , real crystal structure data provides a good starting point ., Therefore , our TMD simulation started from the equilibrated ATP bound coordinates and ends at the equilibrated Apo state for 1 . 5 ns ., The step size is 2 fs seconds ., This strategy is also employed in the previous E . coli MurD study 28 ., It is important to note that the Apo state monomer might not correspond to the ATP bound state monomer with the same segment name ., We aligned each pair of the monomers between the Apo and the ATP bound state and save the 15 ( ) pair-wise alignment scores ., Then align the six sequential monomers in the Apo state with those in the ATP bound state ., For example , the segment of Apo and ATP bound state are represented by ABCDEF and A′B′C′D′E′F′ respectively ., We first align the monomer sequence ABCDEF with A′B′C′D′E′F′ , and then align it to B′C′D′E′F′A′ , and next to C′D′E′F′A′B′ , and so on so forth ., The final alignment is the one with the best overall sequence alignment score ., In our study , we used the last 1 ns from the trajectory ., The extra 0 . 5 ns is removed since it is related with the surface diffusion when ATP approaching the Apo helicase , which is out of the current research ., To consolidate our results we have tried another two TMD simulations ., One is the normal pathway from the equilibrated Apo state to the equilibrated ATP bound state ., The conformational change is similar to the results above ., Another TMD simulation involves an intermediate Apo state with ATPs bound to the pockets ., The ATPs positions are decided by aligning the ATP monomer with the Apo monomer ., The intermediate Apo state is quenched and equilibrated in the same way as described above ., The TMD simulation starts from the ATP bound state , and goes through the intermediate Apo state and ends at the Apo state ., The simulation results are similar to the results presented above and therefore strengthen the results of our conformational pathway ., The PDLD/S-LRA ( Linear Response Approximation version of the semi-microscopic PDLD ) method is designed to effectively evaluate the protein-ligand binding free energies through a thermodynamic cycle that is a fast approximation of the rigorous Free Energy Perturbation ( FEP ) 29 ., The PDLD methods have been described in a series of theoretical papers , including the PDLD method 30 , the semi-empirical version , PDLD/S 31 and the fast approximation version , PDLD/S-LRA 32 ., PDLD methods have been widely applied in the related biological systems , such as the F1-ATPase 33 and HIV protease 29 ., Recently , we have also successfully applied the PDLD methods on the LTag DNA translocation analysis 34 ., We used the PDLD/S-LRA method to evaluate the 20 snapshots ( intermediate structures ) from the simulated TMD trajectory ., The PDLD/S-LRA method evaluates the change in electrostatic free energies upon transfer of a given ligand ( l ) from water to the protein by starting with the effective PDLD potentials; ( 1 ) where ΔGsol denotes the electrostatic contribution to the solvation free energy of the indicated group in water ( e . g . , denotes the solvation of the protein-ligand complex in water ) ., The values of the ΔGsols are evaluated by the Langevin dipole solvent model ., is the electrostatic interaction between the charges of the ligand and the protein dipoles in vacuum ( this is a standard PDLD notation ) ., This approach provides a reasonable approximation for the corresponding electrostatic free energies: ( 2 ) where the effective potential is defined in Eq ., 1 and and designate an MD average over the coordinates of the ligand-complex in their polar and non-polar forms ., It is important to realize that the average of Eq ., 2 is always done where both contributions to the relevant are evaluated at the same configurations ., That is , the PDLD/S energies of the polar and non-polar states are evaluated at each averaging step by using the same structure ., The 20 structures are sampled evenly from the initial docking stage , through the binding transition stage ( WS to TS ) , and the shrinking stage ( TS to ATP bound state ) ., All 20 structures are relaxed for 500 ps at 300 K . Each structure is then evaluated for 10 different runs ., The mean values of these 20 structural evaluations are connected as a rough energy profile ( Fig . 7 ) .

Introduction, Results, Discussion, Methods

Simian Virus 40 Large Tumor Antigen ( LTag ) is an efficient helicase motor that unwinds and translocates DNA ., The DNA unwinding and translocation of LTag is powered by ATP binding and hydrolysis at the nucleotide pocket between two adjacent subunits of an LTag hexamer ., Based on the set of high-resolution hexameric structures of LTag helicase in different nucleotide binding states , we simulated a conformational transition pathway of the ATP binding process using the targeted molecular dynamics method and calculated the corresponding energy profile using the linear response approximation ( LRA ) version of the semi-macroscopic Protein Dipoles Langevin Dipoles method ( PDLD/S ) ., The simulation results suggest a three-step process for the ATP binding from the initial interaction to the final tight binding at the nucleotide pocket , in which ATP is eventually “locked” by three pairs of charge-charge interactions across the pocket ., Such a “cross-locking” ATP binding process is similar to the binding zipper model reported for the F1-ATPase hexameric motor ., The simulation also shows a transition mechanism of Mg2+ coordination to form the Mg-ATP complex during ATP binding , which is accompanied by the large conformational changes of LTag ., This simulation study of the ATP binding process to an LTag and the accompanying conformational changes in the context of a hexamer leads to a refined cooperative iris model that has been proposed previously .

The Large Tumor antigen ( LTag ) encoded by Simian Virus 40 ( SV40 ) is a marvelous molecule that is not only a viral oncogene , but also an efficient molecular machine as a helicase that unwinds double helix DNA for genome replication , an essential process in all living organisms ., LTag hexameric helicase uses the energy of ATP to power its conformational switch for DNA unwinding ., Understanding how the LTag conformational switch is coupled to the energy from ATP usage by LTag to do the mechanical work of unwinding DNA is of great interest to biologists , and yet remains to be established ., Based on our previous high-resolution structures of LTag helicase in different conformational states , we simulated an LTag conformational transition pathway in the ATP binding process using the targeted molecular dynamics method ., Our simulation results suggest a three-step process for the ATP binding to the nucleotide pocket , in which ATP is eventually “locked” into the pocket by three pairs of “locker” interactions ., We have also quantitatively evaluated the energy profile of ATP binding using a special computational simulation technique ., Additionally , our simulation study of ATP binding by LTag and the accompanying conformational switches in the context of a hexamer leads to a refined cooperative iris model that may be used for DNA unwinding .

computational biology/protein structure prediction, computational biology/molecular dynamics

journal.pgen.1008155

Enhancing face validity of mouse models of Alzheimer’s disease with natural genetic variation

Alzheimer’s disease ( AD ) is the most common cause of adult dementia , with approximately 6 million Americans diagnosed with either clinical AD or mild cognitive impairment in 20171 ., Age is the greatest risk factor and currently we have the largest aging population that has ever been on this planet 2 ., Globally , there are 50 million people living with dementia and this number is expected to reach 152 million by 2050 ., Low- and middle-income countries are the hardest hit , comprising 66% of global cases 3 ., AD is pathologically characterized by the accumulation of beta amyloid ( β-amyloid ) plaques , neurofibrillary tangles , and widespread neuronal loss ., Another prominent feature is the neuroinflammatory response by a variety of cells including astrocytes and microglia ., Multiple studies have identified two forms of AD: familial AD ( FAD , also known as early-onset AD ) and sporadic AD ( also known as late-onset AD ) ., Widely-used mouse models of AD utilize FAD mutations in amyloid precursor protein ( APP ) and presenilin 1 and 2 ( PSEN1 and PSEN2 ) ., However , a recent review on the current status of AD clinical trials has suggested that the failure of these trials , in part , is due to the inability of current AD mouse models to translate to humans 4 ., While FAD mouse models have been vital to understand early drivers of amyloidosis , to date , they do not effectively model all hallmarks of AD , particularly frank neurodegeneration ., This has led some to question the utility of mouse models as preclinical models for AD and other diseases of complex etiologies ., Studies including the Dominantly Inherited Alzheimer’s Network ( DIAN ) and the Religious Order Study and Memory and Aging Project ( ROSMAP ) show significant variation in age of onset and rate of disease progression in individuals who inherit the same FAD mutations5 , 6 ., Furthermore , new work performing a genome-wide association study ( GWAS ) 7 on individuals with a family history of AD identified multiple novel variants ., This suggests that the underlying genetic contribution of many cases of FAD are also due to multiple interacting variants , not simply the single strong variants such as those in APP , PSEN1 and PSEN2 ., Therefore , the same is likely true in mouse models ., Murine models relevant to AD have been almost exclusively developed on a single genetic background , C57BL/6 ( B6 ) ., Few studies have assessed FAD mutations in a limited number of alternative genetic backgrounds including 129S1/SvImJ 8 , A/J and DBA/2J ( D2 ) 8–10 ., These studies showed genetic background altered β-amyloid deposition and seizure incidence , but modifications to neuronal cell loss were not reported ., Supporting the potential of incorporating genetic variation in AD mouse models , a recent study used F1 crosses between B6 and thirty classical inbred strains to show that the phenotypes observed from a heterozygous null mutation related to neurological function were not generalizable across strain 11 ., Interestingly , there were multiple cases in which there were inverse effects of the same allele on phenotypic outcomes ., Another recent publication showed greater transability of the mouse to human Alzheimer’s through the development of a new mouse panel known as the AD-BXDs ., This panel was developed by crossing congenic B6 5xFAD mice with BxD males ( B6xD2 ) , greatly increasing the genetic diversity in the context of 5 aggressive familial mutations ., Aging and characterization of these mice indicated a greater range in AD related phenotypes such as plaque pathology and cognitive deficits , and a greater transcriptional overlap with human AD 12 ., These studies highlight the likely huge potential for generating more translatable AD mouse models through the use of different genetic contexts ., Therefore , to take full advantage of the level of natural genetic variation available in mice , we employed genetically distinct wild-derived strains ., Historically , the lineage of commonly used classical laboratory strains can be traced to domesticated fancy mouse stock developed on a farm in Massachusetts in the early 1900s 13 ., Due to this , classical laboratory strains are undefined genomic mixtures of two or more subspecies of Mus musculus ( including Mus musculus domesticus and Mus musculus molossinus ) ., They exhibit limited inter-strain polymorphisms ( less than 5 million differences between a classical inbred strain when compared to B6/J ( 14 and Fig 1 ) , and do not represent any animal that exists in nature ., To overcome the limitations of classic laboratory strains , ‘wild-derived’ strains were introduced as research models in the 1980s ., Wild-derived strains are genetically distinct subspecies of Mus musculus ( e . g . Mus musculus musculus and Mus musculus castaneous ) ., Founders of each strain were caught from well-established wild mice populations from around the world ( see methods ) , and then inbred 15 ., Wild-derived strains show a much greater degree of genetic variation compared to B6 than other classical inbred strains do ( between 6 and 17 million differences ) including millions of private variations ., Importantly , the genetic variation encompassed in these strains and interactions of different gene networks evolved , thus , are likely physiologically relevant to the natural world ., This variation includes genes previously associated with AD including Apoe , Trem2 and Tyrobp , and these strains also show variation in phenotypes relevant to AD risk factors including cardiovascular health 16 , insulin secretion 17–19 , gut microbiota 19 , 20 and circadian rhythm 21 ., In this study , we hypothesized that incorporating FAD mutations into genetically distinct , wild-derived mouse strains would establish more clinically-relevant AD mouse models compared to those on classic laboratory strain backgrounds ., To test this , two commonly used FAD mutations ( APPswe and PSEN1de9 , herein referred to as APP/PS1 ) were introduced into three wild-derived strains representative of the three Mus musculus ( mus ) subspecies: WSB/EiJ ( WSB , M . mus domesticus ) , PWK/PhJ ( PWK , M . mus musculus ) , and CAST/EiJ ( CAST , M . mus castaneus ) ., Assessment of AD-relevant phenotypes showed that the effects of the APP/PS1 transgenes are strain-dependent and sex-dependent , with significant differences in amyloid deposition , neuronal cell loss and cerebral amyloid angiopathy ( CAA ) ., Transcriptional profiling and neuropathological assessment suggested myeloid cell responses are major contributors to the variation in AD phenotypes we observed in the wild-derived AD strains ., Three wild-derived AD mouse models were created by backcrossing for at least six generations the APP/PS1 transgenes from B6 to the genetically distinct substrains WSB , PWK and CAST ( Fig 1 ) ., The presence of both the APPswe and PSEN1de9 transgenes was confirmed by PCR ( S1 Fig ) ., For each strain , balanced cohorts of female and male wild type ( WT ) and APP/PS1 mice were established and aged to 6 months–an age window when the majority of plaques have seeded and are in an exponential growth phase in B6 . APP/PS1 22–24 ., APP/PS1 and randomized WT litter mate controls from each strain were tested sequentially in the following order: ( 1 ) PWK , ( 2 ) WSB , ( 3 ) CAST and ( 4 ) B6 ., For this first characterization of these new strains , a set of metabolic and functional assays were selected that spanned across a wide-range of AD-relevant phenotypes ., Significant strain- , sex- or genotype-specific differences were observed in body weight , body temperature and body composition ( S2 Fig ) ., In addition , significant differences were observed in activity measured using both piezoelectric floor monitoring and open field arenas ( S3 Fig ) ., WT mice from all three wild-derived strains were significantly more active than B6 WT mice ., Also , irrespective of genetic context , all APP/PS1 strains showed the previously reported increase in activity 25 compared to their WT counterparts ., Cognitive function was assessed in wild-derived strains and B6 using spontaneous alternation ( working memory ) and novel spatial recognition ( short-term memory ) in a Y-maze ., Given the increased behavioral ‘wildness’ 26 , 27 of the wild-derived strains , the Y-maze was modified to include specially fabricated covers ( see Methods ) to minimize likelihood of escape ., This was the first time that these tasks had been employed by us for either aging B6 mice or wild-derived strains of any age ., However , these tasks had been previously validated using young B6 male mice 28 and further validated here using PWK ( the first wild-derived strain to be tested ) ., For spontaneous alternation ( S4A–S4C Fig ) , percent alternation exceeded 50% for all strains irrespective of genotype ., Despite hyperactivity phenotypes observed in open field in wild-derived mice , there were no transgenic-related differences in activity levels as measured by total arm entries , thus , increased activity does not confound the interpretation of this task ., Furthermore , we found no correlation between number of arm entries and performance ., Therefore , these data suggest working memory was not affected by the APP/PS1 transgenes ., For novel spatial recognition ( S4D–S4F Fig ) , strain- , sex- and genotype-specific differences were observed ., For the PWK strain , a robust preference for the novel arm after a 30-minute delay was shown for both male and female WT and APP/PS1 mice indicating an intact short-term memory ., In contrast , for WSB females and CAST males , WT but not APP/PS1 mice showed a preference for the novel arm suggesting working memory was impaired in both female WSB . APP/PS1 and male CAST . APP/PS1 mice ., Highlighting the challenges of identifying tasks that can be performed by diverse strains , short-term memory using this task could not be determined for male WSB . APP/PS1 , female CAST . APP/PS1 , and male and female B6 . APP/PS1 , as the strain-matched and sex-matched WT counterparts were unable to perform the task ., Next , to assess neurodegeneration , NEUN+DAPI+ cell counts were performed across all strains , sexes and genotypes in a region of the superior cortex and in the CA1 region of the hippocampus , two brain regions commonly affected early in human AD ( Fig 2 , S1 Table ) ., Interestingly , even in the absence of the APP/PS1 transgenes , strain background was a significant driver of the overall neuronal cell number in the CA1 region ., Importantly , there was a significant loss of NEUN+DAPI+ cells in female WSB . APP/PS1 in the cortical region and CA1 compared to WT WSB females ., There was also significant loss of neurons in male and female CAST . APP/PS1 mice in the CA1 region ., There was no detectable NEUN+DAPI+ loss in either B6 . APP/PS1 ( as previously published in 10 , 29 , 30 ) or PWK . APP/PS1 strains in the two regions studied ., Despite the presence of neurodegeneration in CAST . APP/PS1 and female WSB . APP/PS1 , there was no evidence of increased tau pathology using AT8 , a marker of early tau hyperphosphyloration ( S5 Fig ) ., Amyloidosis was assessed in all four strains using ThioS staining , ELISA and Western blotting ., Surprisingly , numbers of cortical ThioS+ plaques were significantly decreased in all three of the wild-derived APP/PS1 strains in comparison with B6 . APP/PS1 ( Fig 3A–3C , S2 Table ) ., Numbers of hippocampal ThioS+ plaques were also significantly decreased with the exception of WSB . APP/PS1 females ., No plaques were observed in WT mice from any of the four strains in any brain region ., Plaque morphology appeared different between B6 . APP/PS1 and wild-derived APP/PS1 strains ., Specifically , there was an absence of small ThioS+ plaques in wild-derived APP/PS1 compared to B6 . APP/PS1 mice ., Despite the reduced numbers of plaques , there was a significant increase in Aβ42 ( measured by ELISA ) in both female CAST . APP/PS1 and WSB . APP/PS1 compared to B6 . APP/PS1 ( Fig 3D ) ., This increase cannot be accounted for simply by differences in mutant APP production as Western blotting using 6e10 ( antibody to human mutant APP ) showed similar APP protein levels across all strains ( Fig 3E ) with the exception of male PWK . APP/PS1 ( significant difference between male B6 . APP/PS1 and male PWK . APP/PS1 , p ≤ 0 . 01 ) ., Therefore , our data suggest that at 8 months , plaques in the wild-derived APP/PS1 strains may be further along in the rapid growth period previously defined for B6 . APP/PS1 mice 24 ., Another prominent amyloid phenotype observed in the wild-derived strains was ThioS+ vessels , suggesting the occurrence of cerebral amyloid angiopathy ( CAA ) ., Brain sections from all strains , sexes and genotypes were examined for the presence of ThioS+ vessels and by silver staining ., CAA was pronounced in vessels of CAST . APP/PS1 and WSB . APP/PS1 , but not B6 . APP/PS1 or PWK . APP/PS1 mice ( S6 Fig ) ., There was no evidence of vascular staining of ThioS in WT animals ., CAA has been associated with cerebrovascular damage in human AD and recent studies support a more prominent role of cerebrovascular decline in AD pathogenesis 31 , 32 ., To test the relationship between CAA and cerebrovascular integrity , brain sections from WSB . APP/PS1 were assessed as they showed the greatest percentage of ThioS+ vessels ., Cerebrovascular integrity was determined using antibodies to fibrin ( ogen ) , a protein that is ordinarily present in blood but its presence in the brain is indicative of blood brain barrier compromise ., Fibrin was present outside of the microvessels in brain sections from WBS . APP/PS1 , but not in B6 . APP/PS1 ( Fig 4 ) ., To provide insight into the strain-specific differences that may be driving the phenotype differences observed between strains , transcriptional profiling by RNA-seq was performed on the left brain hemispheres from WT and APP/PS1 male and female mice from all strains ( 93 samples in total ) ., Sequencing depth ( S7 Fig ) and expression levels of the APPswe and PSEN1de9 transcripts in APP/PS1 mice ( S8 Fig ) were consistent between strains ., Principle Component Analysis ( PCA ) identified strain as the greatest driver of gene expression variance across samples , consistent with the genetic distinctness of strains ( Fig 5A ) ., To identify modules of genes that were differentially expressed between groups , Weighted Gene Co-expression Analysis ( WGCNA ) was performed ., The majority of modules were driven by strain , independent of APP/PS1 genotype ( S9 Fig ) ., However , one module ( termed ‘light yellow’ ) was driven by APP/PS1 genotype and seen in all strain backgrounds ( Fig 5B ) ., This module contained 35 genes that are enriched for the Lysosome and Osteoclast Differentiation KEGG pathways ( Fig 5C ) ., The light yellow module included App and Psen1 supporting the fact that this module is likely an amyloid response module ., The majority of other genes in the module are expressed in myeloid cells ( either resident microglia and/or monocytes/macrophages ) ., Many of these genes have been previously implicated in AD-relevant processes such as amyloid deposition and synaptic loss including C1qa , Csf1r , Tyrobp , Cx3cr1 , Cd68 and Ctsz ., Importantly , DNA variations in two genes in the light yellow module , Trem2 and Cd33 , have previously been associated with human AD suggesting these genes may be early drivers of AD pathogenesis ., Assessment of the eigenvalues for the light yellow module revealed two major findings ., First , there was great variation in the eigenvalues when comparing WT samples between strains ., For instance , eigenvalues were lowest for WT samples from WSB and CAST ., This was reflected in the normalized expression levels of genes in the module ., Trem2 , Tyrobp and Ctss showed the lowest expression in WT samples from WSB and CAST ( Fig 5E ) –strains that showed neuronal cell loss in the presence of amyloid ( Fig 2 ) ., The second major finding was that there were marked differences in eigenvalues comparing WT to APP/PS1 samples ., The greatest difference between WT and APP/PS1 samples was observed in PWK ., Again , these differences were also observed at the level of individual genes within the light yellow module ( Fig 5B ) ., This suggests the ability of myeloid cells to respond to amyloid is strongly influenced by genetic context ., Together , these data suggest that there are intrinsic differences between myeloid cells in WT samples from different strains and that these cells respond differently to amyloid deposition ., Both these factors are likely critical in determining whether or not a strain is susceptible to amyloid-induced neurodegeneration ., A major and unexpected finding from the transcriptional profiling was that transcript levels of myeloid cell genes were significantly lower in WT WSB and CAST mice compared to B6 and PWK mice ., This suggests that myeloid cells vary between strains , even in the absence of amyloid ., To test this , IBA1+ myeloid cell numbers were determined ., There was a significant difference in the numbers of IBA1+DAPI+ cells in WT mice of different strains ( Fig 6 ) ., WT mice from CAST and WSB mice showed significantly fewer IBA1+ cells compared to B6 ( Bonferroni’s multiple comparison test vs B6: Male WSB p ≤ 0 . 0001 and CAST p ≤ 0 . 01; Female WSB p ≤ 0 . 01 and CAST p ≤ 0 . 05 ) ., This supports the transcriptional profiling data ( Fig 5 ) ., As expected , there was a significant sex and region-specific increase in IBA1+ cells in mice carrying the APP/PS1 transgenes compared to their WT counterparts ( Fig 6 , S3 Table ) ., Transcriptional profiling also predicted that plaque-mediated myeloid cell responses would differ between strains ., To assess this , the numbers of myeloid cells surrounding plaques were determined for each APP/PS1 strain ., For each mouse , the numbers of IBA1+DAPI+ cells ( myeloid ) were determined around five plaques of similar relative size in 6 mice per strain ( a total of 30 plaques/strain , Fig 7A and 7B ) ., The median number of IBA1+ myeloid cells per section was averaged per animal and then compared across strains ., For male animals , CAST . APP/PS1 had the greatest number of plaque-associated IBA1+ cells , while WSB . APP/PS1 had the least ., For female animals , CAST . APP/PS1 exhibited the greatest number of plaque-associated IBA1+ cells ., One myeloid cell response that has recently been highlighted as important is proliferative capacity , and there remains a debate regarding whether this is helpful or harmful in response to injury or in progression of neurodegenerative diseases 33 , 34 ., CAST . APP/PS1 mice showed the greatest numbers of myeloid cells around plaques despite having the fewest numbers of myeloid cells in WT animals ( Fig 6 ) ., To determine whether this could be due to myeloid cell proliferation , the proliferative marker KI-67 was used ., KI-67+IBA1+ cells in CAST . APP/PS1 mice were compared to B6 . APP/PS1 mice ., There were significantly more plaque associated KI-67+IBA1+ cells observed in CAST . APP/PS1 compared to B6 . APP/PS1 mice ( t ( 14 ) = 3 . 73 , p = 0 . 002 ) ( Fig 7D ) ., This suggests underlying differences in the proliferative capacity of myeloid cells between strains , and may be a factor in neuronal cell loss exhibited by CAST . APP/PS1 ., The work presented here highlights the value and power of increased genetic diversity within mouse models in order to gain insight into the complex etiologies of human disease ., Our work shows that in contrast to the B6 . APP/PS1 strain that has been widely used historically , CAST . APP/PS1 , WSB . APP/PS1 and PWK . APP/PS1 represent models that provide a new lens to understanding central features of human AD including amyloid-induced neurodegeneration , neuroinflammation , cerebrovascular integrity and cerebral amyloid angiopathy ., Overall , there were three major findings: ( 1 ) Female WSB . APP/PS1 showed significant hippocampal and cortical neuronal cell loss , whole brain elevated levels of Aβ42 , and cognitive impairment in a short-term memory task ., This was accompanied by substantial vascular amyloid deposition in the form of cerebral amyloid angiopathy accompanied by vascular compromise ., ( 2 ) CAST . APP/PS1 showed hippocampal cell loss and females exhibited whole brain elevated levels of Aβ42 ., ( 3 ) Transcriptional profiling corroborated by neuropathology identified strain-dependent baseline differences in the expression of neuroinflammatory ( primarily myeloid-related ) genes and in the magnitude of their response to amyloid ., Based on these findings , we predict that a major driver of the phenotype differences observed between strains ( e . g . neuronal cell loss and CAA ) is due to differences in neuroinflammation , particularly myeloid cells ., A substantial part of the observed variation in myeloid cell-driven inflammation has been linked to genetic differences between human populations 35 ., Wild-derived AD mouse models appear to show important strain- and sex-dependent differences in behavior and pathology that are similar to the human clinical population that show both sex-specific and ethnic differences in terms of prevalence and progression 36 ., However , a major challenge of this study was to develop a functional battery that could be used across the strains as they exhibit formidable differences in wildness compared to classical laboratory strains ., Wildness score is comprised of measurements of jumping , escape , struggle , squeaking and biting , and strains like B6 and D2 earn a score ranging between 0 . 21 and 0 . 66 , while wild-derived strains range from 1 . 35 for CAST 26 to 2 . 5 or greater for WSB 27 ., While it was important to include a range of functional assays , we anticipated there could be issues as traditional behavioral assays have been primarily optimized for typically behaving mice ( e . g . young male B6 ) and likely would not be optimal for testing wild-derived strains ., While B6 . APP/PS1 mice have previously been shown to exhibit deficits in assays such as Contextual Fear Conditioning as early as 6 months 37 , we chose to avoid aversive tasks due to the inherent difficulty in handling wild-derived mice and stress caused to the animals with repeated handling ., Instead , we chose tasks that utilized the animal’s natural exploratory drive such as y-maze tasks that assess spatial memory ., While spatial memory deficits in B6 or B6/C3H mice carrying the APP/PS1 transgenes are typically over 12 months of age 38 , 39 , it would be expected that earlier impairment would be identified in a sensitized genetic context ., To our knowledge , this is the first time that many of the tasks included in our battery were used to assess wild-derived mouse behavior , thus , all data is left intact ( no outliers removed ) and presented as individual data points ., Unfortunately , in this study , not all WT strains were able to perform the novel spatial recognition task , which may have been due to the length of the memory delay chosen ( 30 minutes ) ., This meant that short-term memory could not be assessed for some APP/PS1 strains ., Therefore , more extensive functional assays are still required at multiple ages to determine the utility of these strains for studies of cognitive impairment ., Given our experiences in this study , it may be necessary to develop and validate strain-specific cognitive assays due to inherent differences in wildness and age-dependent cognitive abilities ., This will be of particular importance in tasks that may require food restriction as these strains have vastly different metabolic rates ., In the age window tested , there were significant differences in plaque numbers and Aβ42 levels between strains ., Taken alone , plaque counts would suggest less amyloid in the wild-derived APP/PS1 mice in comparison with B6 . APP/PS1 ., However , the size distribution of plaques varied across the strains , with B6 . APP/PS1 exhibiting many smaller proximal deposits that may correspond to initial seeding of Aβ ., This is in contrast to plaques in the brains of wild-derived APP/PS1 mice that were of moderate size ., This could be indicative of a more advanced stage of amyloid deposition in 8 month wild-derived mice carrying APP/PS1 in comparison with B6 . APP/PS1 of the same age; and/or , suggest the presence of different conformations of amyloid fibrils ., Previous work has shown that identical peptide sequences are capable of forming into different conformations of amyloid fibrils , and that this difference can be detected by seeding efficiencies 40 ., The development of cerebral amyloid angiopathy ( CAA ) in WSB . APP/PS1 coupled with fibrin leakage ( Fig 4 ) suggests compromised vascular integrity and/or deficits in amyloid clearance ., It is possible that cerebrovascular damage ( measured here by fibrin leakage ) is downstream of amyloid ., Conversely , an inherent weakness in cerebrovascular structures in WSB mice may dispose mice to CAA ., While CAA has been reported many times before in AD models carrying APP mutations on a B6 background , typically it does not appear with complete banding until mice are over 14 months of age 41–43 ., Furthermore , the severity of CAA and risk of associated microhemorrhage progresses with age ., There is strong evidence to suggest that the earliest predictors of AD-susceptibility and onset are related to vascular and blood-brain-barrier integrity 44 ., The presence of severe CAA and neurodegeneration in WSB . APP/PS1 will allow mechanistic dissection of the relationship between CAA and neurodegeneration ., We found that strain is the greatest driver of gene expression variation in these mouse models , even more so than sex or APP/PS1 ., This is representative of the inclusion of millions of genetic differences created from wild-derived strains that have never before been explored in context of modeling AD ., Mus musculus , also known as the house mouse , are characterized as being commensal animals , meaning that they live in close association with humans , and even though they are able to adapt to a wide-range of environments , are dependent on human shelter or activity for their survival 15 ., Each distinct subspecies is from different geographical regions ( CAST was trapped in Thailand , WSB was trapped in eastern shore Maryland , USA and PWK was trapped in the Czech Republic ) , and evolved separately to survive alongside humans in the face of similar region-specific pressures ( exposure to pathogens or infection , climate , diet etc . ) ., Therefore , some of the genetic differences between mouse substrains may correspond with genetic variants in different populations of humans ., More likely however , the variations driving the phenotype differences will impact similar genes/pathways that are modified by genetic risk variants in the human population ., In support of this , many of the genes in the module identified by WGCNA ( Fig 5D ) have previously been implicated in human AD–including sporadic AD–either through genetic association , gene expression studies or functional studies ., Therefore , despite the presence of the APP/PS1 transgene artificially driving amyloid accumulation in these strains–the responses appear to be directly relevant to human AD ., This suggests that interventions tested in these new AD strains that target factors downstream of amyloid deposition but upstream of neurodegeneration would be expected to be clinically relevant to FAD and LOAD ., PWK . APP/PS1 is particularly intriguing as transcriptional data suggest it is the greatest responder to amyloid at 8 months and appears to be a resilient strain ( no neuronal cell loss detected ) ., These data may be consistent with a slower progression/transition from amyloid deposition to neuronal cell dysfunction which could become apparent at older ages , or representative of a neuroprotective signature ., Two additional and striking phenotypes may also be reflective of the substantial neuroinflammation in PWK . APP/PS1 mice ., First , during generation of the experimental cohort , PWK . APP/PS1 had to be separated from WT littermates at 3 months of age due to increased aggression ., Second , changes in activity in the piezoelectric chambers were observed in female PWK . APP/PS1 ( S3 Fig ) ., These may be behavioral manifestations of increased neuroinflammation in response to amyloid ., Agitation and circadian disruption are clinical symptoms that directly interfere with the ability of caregiving to occur in the home and have both been linked with neuroinflammation in humans 45 , 46 ., Transcriptional profiles in WT animals suggested that in comparison with B6 and PWK , CAST and WSB show lower baseline expression of the primarily myeloid-related genes in this module ., Cell counts confirmed that there was ~50% reduction in the number of IBA1+ cells in both CAST and WSB ., Natural inherent differences in neuroinflammation is important given the lack of studies into how genetic variation impacts glial cell development and homeostasis in the human population–an area that might be critical in predisposing to age-related diseases such as AD ., Similarly , while there have been renewed efforts to characterize the immune systems in wild-caught mice 47 , there still remains a dearth of knowledge regarding how genetic variation impacts myeloid cell and astrocyte function in inbred wild-derived strains ., This may be starting to change as recent work by Christopher Glass and colleagues 48 analyzed macrophages from five inbred mouse strains , including PWK ., As in our study , strain was the greatest driver of differences in gene expression in these macrophages ., Much of the foundation of mouse genetics has been focused on examination of a single genetic difference while holding all other genetic ( i . e . strain background ) and environmental influences constant ., Somewhere along the way , limited resources and a wise desire for standardization restricted this examination to only one or two laboratory strains , despite efforts more than 4 decades ago to develop mouse resources such as the wild-derived strains and periodic suggestions of researchers past to expand beyond one strain 49–51 ., Our study represents one of few studies to utilize natural genetic variation in mice to gain further insight in human AD ., For the first time , we show neurodegeneration and mixed pathology in wild-derived strains carrying the APP/PS1 transgenes ., Interestingly , our data suggests B6 is a ‘resilient’ strain when considering neurodegeneration ., This ‘resilience’ may be specifically driven by differences in myeloid-related neuroinflammation , and we predict that differences in myeloid cell biology in these new wild-derived AD mouse models will provide a much-needed platform for identification of novel genes/variants modifying susceptibility to neuronal cell loss ., One caveat of these new strains is that amyloid is driven by the APP/PS1 transgenes ., Transgenic overexpression of proteins can provide additional side effects and mutations in APP and PSEN1 which may not be ideal to uncover the mechanisms of sporadic AD ., A second caveat is that , despite neuronal cell loss , they appear to lack overt TAU pathology ( S5 Fig ) ., Therefore , further work is still needed to improve both the construct validity and the face validity of these new mouse models ., Research is now focused on inducing sporadic AD in mice in the absence of transgenic overexpression of familial AD mutations ., Mice differ from humans in both the APP and TAU proteins ., The human APP protein is generally considered to be more amyloidogenic than the mouse and the ratios of the 3R and 4R isoforms of TAU are balanced in human adults , but not in adult mice ., This may be a contributing factor to the apparent lack of TAU pathology in wild-derived APP/PS1 strains ., Multiple efforts , including our own , are improving the construct validity of AD mouse models by humanizing the App and Mapt loci and incorporating sporadic AD-relevant variants such as APOEE4 and TREM2R47H

Introduction, Results, Discussion, Materials and methods

Classical laboratory strains show limited genetic diversity and do not harness natural genetic variation ., Mouse models relevant to Alzheimer’s disease ( AD ) have largely been developed using these classical laboratory strains , such as C57BL/6J ( B6 ) , and this has likely contributed to the failure of translation of findings from mice to the clinic ., Therefore , here we test the potential for natural genetic variation to enhance the translatability of AD mouse models ., Two widely used AD-relevant transgenes , APPswe and PS1de9 ( APP/PS1 ) , were backcrossed from B6 to three wild-derived strains CAST/EiJ , WSB/EiJ , PWK/PhJ , representative of three Mus musculus subspecies ., These new AD strains were characterized using metabolic , functional , neuropathological and transcriptional assays ., Strain- , sex- and genotype-specific differences were observed in cognitive ability , neurodegeneration , plaque load , cerebrovascular health and cerebral amyloid angiopathy ., Analyses of brain transcriptional data showed strain was the greatest driver of variation ., We identified significant variation in myeloid cell numbers in wild type mice of different strains as well as significant differences in plaque-associated myeloid responses in APP/PS1 mice between the strains ., Collectively , these data support the use of wild-derived strains to better model the complexity of human AD .

Despite the rise in incidence of Alzheimer’s disease ( AD ) , it has been over a decade since a new drug treatment has been introduced ., Recently , a number of pharmaceutical giants have shut down their AD research units ., One issue that these companies and researchers have struggled with is the lack of translatability of preclinical studies to the clinic ., One aspect that has come under heavy scrutiny is whether the mouse can be an appropriate model for a complex human disease such as AD ., Current mouse models of AD have incorporated well-known early onset AD mutations on a single genetic background , C57BL/6J , which does not develop all features of human AD- namely marked neurodegeneration ., Here we sought to improve the utility and translatability of mouse models through the use of three genetically distinct , wild-derived inbred mouse strains , CAST/EiJ , WSB/EiJ and PWK/PhJ ., These mice encompass millions of genetic differences that have never before been explored in the context of modeling AD ., Wild-derived mice that carried the early onset AD mutations exhibited robust differences in immune response to amyloid , evidence of mixed pathology and early neurodegeneration , better recapitulating what happens in human AD than previous models .

medicine and health sciences, neurodegenerative diseases, gene regulation, population genetics, vertebrates, mice, animals, mammals, animal models, bone marrow cells, genetics of disease, model organisms, experimental organism systems, population biology, alzheimers disease, research and analysis methods, genetic polymorphism, animal studies, animal cells, gene expression, mouse models, dementia, mental health and psychiatry, rodents, eukaryota, cell biology, neurology, genetics, biology and life sciences, cellular types, human genetics, evolutionary biology, amniotes, organisms

journal.pcbi.1006299

Representational structure or task structure? Bias in neural representational similarity analysis and a Bayesian method for reducing bias

Functional magnetic resonance imaging ( fMRI ) measures the blood-oxygen-level-dependent ( BOLD ) signals 1 , which rise to peak ∼6 seconds after neuronal activity increases in a local region 2 ., Because of its non-invasiveness , full-brain coverage , and relatively favorable trade-off between spatial and temporal resolution , fMRI has been a powerful tool to study the neural correlates of cognition 3–5 ., In the last decade , research has moved beyond simply localizing the brain regions selectively activated by cognitive processes and the focus has been increasingly placed on the relationship between the detailed spatial patterns of neural activity and cognitive processes 6 , 7 ., An important tool for characterizing the functional architecture of the brain is representational similarity analysis ( RSA ) 8 ., This classic method first estimates the neural activity patterns from fMRI data recorded as participants observe a set of stimuli or experience a set of task conditions , and then calculates the similarity ( e . g . , by Pearson correlation ) between each pair of the estimated patterns ., The rationale is that if two stimuli are represented with similar codes in a brain region , the spatial patterns of neural activation in that region would be similar when processing these two stimuli ., When using Pearson correlation as a similarity metric , the activity profile of each voxel to all the task conditions is essentially viewed as one independent sample from a multivariate normal distribution in a space spanned by the experimental conditions , which is characterized by its covariance matrix ., Recently , it has been pointed out that RSA and two other approaches for understanding neural representational structure , namely encoding model 9 and pattern component modeling ( PCM ) 10 , are closely related through the second moment statistics ( the covariance matrix ) of the true ( unknown ) activity patterns 11 ., After the similarity matrix between all pairs of estimated activity patterns is calculated in a region of interest ( ROI ) , it can be compared against similarity matrices predicted by candidate computational models ., Researchers can also convert the similarity matrix into a representational dissimilarity matrix ( RDM , e . g . , 1 − C , for similarity C based on correlation ) and visualize the structure of the representational space in the ROI by projecting the dissimilarity matrix to a low dimensional space 8 ., Researchers might also test whether certain experimental manipulations change the degrees of similarity between neural patterns of interest 12 , 13 ., To list just a few applications from the field of visual neuroscience , RSA has revealed that humans and monkeys have highly similar representational structures in the inferotemporal ( IT ) cortex for images across various semantic categories 14 ., It also revealed a continuum in the abstract representation of biological classes in human ventral object visual cortex 15 and that basic categorical structure gradually emerges through the hierarchy of visual cortex 16 ., Because of the additional flexibility of exploring the structure of neural representation without building explicit computational models , RSA has also gained popularity among cognitive neuroscientists for studying more complex tasks beyond perception , such as decision making ., While RSA has been widely adopted in many fields of cognitive neuroscience , a few recent studies have revealed that the similarity structure estimated by standard RSA might be confounded by various factors ., First , the calculated similarity between two neural patterns strongly depends on the time that elapsed between the two measured patterns: the closer the two patterns are in time , the more similar they are 17 18 ., Second , it was found that because different brain regions share some common time courses of fluctuation independent of the stimuli being presented ( intrinsic fluctuations ) , RDMs between regions are highly similar when calculated based on patterns of the same trials of tasks but not when they are calculated based on separate trials ( thus the intrinsic fluctuation are not shared across regions ) ., This indicates that RSA can be strongly influenced by intrinsic fluctuation 17 ., Lastly , Diedrichsen et al . ( 2011 ) pointed out that the noise in the estimated activity patterns can add a diagonal component to the condition-by-condition covariance matrix of the spatial patterns ., This leads to over-estimation of the variance of the neural pattern and underestimation of correlation between true patterns , and this underestimation depends on signal-to-noise ratio in each ROI , making it difficult to make comparison of RDMs between regions 10 ., Recognizing the first two issues , several groups have recently suggested modifications to RSA such as calculating similarity or distance between activity patterns estimated from separate fMRI runs 18 , 19 , henceforth referred to as cross-run RSA , and using a Taylor expansion to approximate and regress out the dependency of pattern similarity on the interval between events 18 ., For the last issue , Diedrichsen et al . ( 2011 ) proposed PCM which models the condition-by-condition covariance matrix between estimated neural patterns as the sum of a diagonal component that reflects the contribution of noise in the estimated neural patterns to the covariance matrix and components reflecting the researcher’s hypothetical representational structure in the ROI 10 ., These methods improve on traditional RSA , but are not explicitly directed at the source of the bias , and therefore only offer partial solutions ., Indeed , the severity of confounds in traditional RSA is not yet widely recognized ., RSA based on neural patterns estimated within an imaging run is still commonly performed ., Furthermore , sometimes a study might need to examine the representational similarity between task conditions within an imaging run , such that cross-run RSA is not feasible ., The Taylor expansion approach to model the effect of event-interval can be difficult to set up when a task condition repeats several times in an experiment ., There also lacks a detailed mathematical examination of the source of the bias and how different ways of applying RSA affect the bias ., Researchers sometimes hold the view that RSA of raw fMRI patterns , instead of activity patterns ( β ) estimated through a general linear model ( GLM ) 20 , does not suffer from the confounds mentioned above ., Last but not least , the contribution of noise in the estimated neural patterns to the sample covariance matrix between patterns may not be restricted to the diagonal elements , as we will demonstrate below ., In this paper , we first compare the result of performing traditional RSA on a task-based fMRI dataset with the results obtained when performing the same analysis on white noise , to illustrate the severe bias and spurious similarity structure that can result from performing RSA on pattern estimates within imaging runs ., By applying task-specific RSA on irrelevant resting-state fMRI data , we show that spurious structure also emerges when RSA is performed on the raw fMRI pattern rather than estimated task activation patterns ., We observed that the spurious structure can be far from a diagonal matrix , and masks any true similarity structure ., We then provide an analytic derivation to help understand the source of the bias in traditional RSA ., Previously , we have proposed a method named Bayesian RSA ( BRSA ) , which significantly reduced this bias and allows analysis within imaging runs 21 ., BRSA is related to PCM in the sense that they both treat the true and unknown activity profiles of each voxel as a sample from a multivariate normal distribution and marginalize the true activity pattern in their analysis ., The critical difference is that PCM models the estimated activity patterns of each trial or task condition , in which complex spurious correlation structure could have already been introduced during the estimation , while BRSA directly models the raw imaging data ., Here , we further extend BRSA to explicitly model spatial noise correlation , thereby mitigating the second issue identified by Heriksson et al . 17 , namely the intrinsic fluctuation not modelled by task events in an experiment ., Furthermore , inspired by the methods of hyper-alignment 22 and shared response models 23 , we extend our method to learn a shared representational similarity structure across multiple participants ( Group BRSA ) and demonstrate improved accuracy of this approach ., Since our method significantly reduces bias in the estimated similarity matrix but does not fully eliminate it at regimes of very low signal-to-noise ratio ( SNR ) , we further provide a cross-validation approach to detecting over-fitting to the data ., Finally , we show that the learned representational structure can serve as an empirical prior to constrain the posterior estimation of activity patterns , which can be used to decode the cognitive state underlying activity observed in new fMRI data ., The algorithm in this paper is publicly available in the Python package Brain Imaging Analysis Kit ( BrainIAK ) , under the brainiak . reprsimil . brsa module ., Our previous version of Bayesian RSA method 21 with newly added modeling of spatial noise correlation is in the BRSA class of the module ., The new version described in this paper is implemented in the GBRSA class and can be applied to either a single participant or a group of participants ., Traditional RSA 8 first estimates the response amplitudes ( β ) of each voxel in an ROI to each task condition , and then calculates the similarity between the estimated spatial response patterns of that ROI to each pair of task conditions ., The estimation of β is based on a GLM ., We denote the fMRI time series from an experiment as Y ∈ R n T × n V , with nT being the number of time points and nV the number of voxels ., The GLM assumes that, Y = X · β + ϵ ., ( 1 ) X ∈ R n T × n C is the “design matrix , ” where nC is the number of task conditions ., Each column of the design matrix is constructed by convolving a hemodynamic response function ( HRF ) with a time series describing the onsets and duration of all events belonging to one task condition ., The regressors composing the design matrix express the hypothesized response time course elicited by each task condition ., Each voxel’s response amplitudes to different task conditions can differ ., The response amplitudes of one voxel to all conditions forms that voxel’s response profile ., All voxels’ response profiles form a matrix of spatial activity patterns β ∈ R n C × n V , with each row representing the spatial pattern of activity elicited by one task condition ., The responses to all conditions are assumed to contribute linearly to the spatio-temporal fMRI signal through the temporal profile of hemodynamic response expressed in X . Thus , the measured Y is assumed to be a linear sum of X weighted by response amplitudes β , corrupted by zero-mean noise ϵ ., The goal of RSA is to understand the degree of similarity between each pair of spatial response patterns ( i . e . , between the rows of β ) ., But because the true β is not accessible , a point estimate of β , derived through linear regression , is usually used as a surrogate:, β ^ = ( X T X ) - 1 X T Y ( 2 ), Similarity is then calculated between rows of β ^ ., For instance , one measure of similarity that is frequently used is Pearson correlation ., The similarity between patterns of condition i and j is assessed as, C i j = ( β ^ i - β ^ i ¯ ) ( β ^ i - β ^ j ¯ ) T n V σ β ^ i σ β ^ j ( 3 ), where β ^ i ¯ and σ β ^ i are the mean and standard deviation of the estimated pattern of condition i across voxels ., To demonstrate the spurious structure that may appear in the result of traditional RSA , we first performed RSA on the fMRI data in one ROI , the orbitofrontal cortex , in a previous dataset involving a decision-making task 24 ., The task included 16 different task conditions , or “states . ”, In each state , participants paid attention to one of two overlapping images ( face or house ) and made judgments about the image in the attended category ., The transition between the 16 task states followed the Markov chain shown in Fig 1A , thus some states often preceded certain other states ., The 16 states could be grouped into 3 categories according to the structure of transitions among states ( the exact meaning of the states , or the 3 categories , are not important in the context of the discussion here . ) We performed traditional RSA on the 16 estimated spatial response patterns corresponding to the 16 task states ., To visualize the structure of the neural representation of the task states in the ROI , we used multi-dimensional scaling ( MDS ) 25 to project the 16-dimensional space defined by the distance ( 1—correlation ) between states onto a 3-dimensional space ( Fig 1B ) ., This projection appears to show clear grouping of the states in the orbitofrontal cortex consistent with the 3 categories , suggesting that this brain area represent this aspect of the task ., However , a similar representational structure was also observed in other ROIs ., In addition , when we applied the same GLM to randomly generated white noise and performed RSA on the resulting parameter estimates , the similarity matrix closely resembled the result found in the real fMRI data ( Fig 1C ) ., Since there is no task-related activity in the white noise , the structure obtained from white noise is clearly spurious and must reflect a bias introduced by the analysis ., In fact , we found that the off-diagonal structure obtained from white noise ( Fig 1C ) explained 84 ± 12% of the variance of the off-diagonals obtained from real data ( Fig 1B ) ., This shows that the bias introduced by traditional RSA can dominate the result , masking the real representational structure in the data ., To help understand this observation , we provide an analytic derivation of the bias with a few simplifying assumptions 21 ., The calculation of the sample correlation of β ^ in traditional RSA implies the implicit assumption that an underlying covariance structure exists that describes the distribution of β , and the activity profile of each voxel is one sample from this distribution ., Therefore , examining the relation between the covariance of β ^ and that of true β will help us understand the bias in traditional RSA ., We assume that a covariance matrix U ( of size nC × nC ) captures the true covariance structure of β across all voxels in the ROI: β ∼ N ( 0 , U ) ., Similarity measures such as correlation are derived from U by normalizing the diagonal elements to 1 . It is well known that temporal autocorrelation exists in fMRI noise 26 , 27 ., To capture this , we assume that in each voxel ϵ ∼ N ( 0 , Σϵ ) , where Σ ϵ ∈ R n T × n T is the temporal covariance of the noise ( for illustration purposes , here we assume that all voxels have the same noise variance and autocorrelation , and temporarily assume the noise is spatially independent ) ., By substituting the expression for Y from Eq ( 1 ) into the point estimate of β ( 2 ) , we obtain, β ^ = ( X T X ) - 1 X T X β + ( X T X ) - 1 X T ϵ = β + ( X T X ) - 1 X T ϵ ( 4 ), which means the point estimate of β is contaminated by a noise term ( XT X ) −1 XTϵ ., Assuming that the signal β is independent from the noise ϵ , it is then also independent from the linear transformation of the noise , ( XT X ) −1 XTϵ ., Thus the covariance of β ^ is the sum of the covariance of true β and the covariance of ( XT X ) −1 XTϵ:, β ^ ∼ N ( 0 , U + ( X T X ) - 1 X T Σ ϵ X ( X T X ) - 1 ) ( 5 ) The term ( XT X ) −1 XT∑ϵ X ( XT X ) −1 is the source of the bias in RSA ., This bias originates from the structured noise ( XT X ) −1 XTϵ in estimating β ^ ., It depends on both the design matrix X and the temporal autocorrelation of the noise ϵ ., Fig 1F illustrates how structured noise can alter the correlation of noisy pattern estimates in a simple case of just two task conditions ., Even if we assume the noise is temporally independent ( i . e . , Σϵ is a diagonal matrix , which may be a valid assumption if one “pre-whitens” the data before further analysis 27 ) , the bias structure still exists but reduces to ( XT X ) −1 σ2 , where σ2 is the variance of the noise ., Since the covariance matrix of β ^ is biased , its correlation is also distorted from the true correlation structure ., This is because correlation is merely a rescaling of rows and columns of a covariance matrix ., Fig 1C essentially illustrates this bias structure after being converted to correlation matrix ( in this case , σ = 1 and β = 0 ) as this RSA structure , by virture of being derived for white noise , can only result from structure in the design matrix X . In reality , both spatial and temporal correlations exist in fMRI noise , which complicates the structure of the bias ., But the fact that bias in Fig 1C arises even when applying RSA to white noise which itself has no spatial-temporal correlation helps to emphasize the first contributor to the bias: the timing structure of the task , which is exhibited in the correlations between the regressors in the design matrix ., Whenever the interval between events of two task conditions is shorter than the length of the HRF ( which typically outlasts 12 s ) , correlation is introduced between their corresponding columns in the design matrix ., The degree of correlation depends on the overlapping of the HRFs ., If one task condition often closely precedes another , which is the case here as a consequence of the Markovian property of the task , their corresponding columns in the design matrix are more strongly correlated ., As a result of these correlations , XT X is not a diagonal matrix , and neither is its inverse ( XT X ) −1 ., In general , unless the order of task conditions is very well counterbalanced and randomized across participants , the noise ( XT X ) −1 XTϵ in β ^ is not i . i . d between task conditions ., The bias term B = ( XT X ) −1 XT Σϵ X ( XT X ) −1 then deviates from a diagonal matrix and causes unequal distortion of the off-diagonal elements in the resulting correlation matrix of β ^ ., These unequal distortions alter the order of ranking of the values of the off-diagonal elements ., Therefore , rank correlation between the similarity matrix from traditional RSA and the similarity matrix of any candidate computational model is necessarily influenced by the bias ., Conclusion based on such comparison between two similarity matrices or based on comparing a pair of off-diagonal elements within a neural similarity matrix becomes problematic , as long as the bias causes unequal distortion ., Furthermore , if the design matrices also depend on participants’ performance such as errors and reaction time , the bias structure could depend on their performance as well ., Comparison between neural representational structure and participants’ behavioral performance may also become problematic in such situations ., It is worth pointing out that the bias is not restricted to using correlation as metric of similarity ., Because structured noise exists in β ^ , any distance metrics between rows of β ^ estimated within imaging runs of fMRI data are likely biased ., We can take Euclidean distance as an example ., For any two task conditions i and j , the expectation of the distance between β i ^ and β j ^ is ∑ k = 1 n V ( β i k - β j k ) 2 + n V ( B i i 2 + B j j 2 - 2 B i j 2 ) , where B is the bias in the covariance structure ., Therefore , the bias n V ( B i i 2 + B j j 2 - 2 B i j 2 ) in Euclidean distance also depends on the task timing structure and the property of noise ., ( See Fig 1D ) ., In our derivations above , point estimates of β ^ introduce structured noise due to the correlation structure in the design matrix ., One might think that the bias can be avoided if a design matrix is not used , i . e . , if RSA is not performed after GLM analysis , but directly on the raw fMRI patterns ., Such an approach still suffers from bias , for two reasons that we detail below ., First , RSA on the raw activity patterns suffers from the second contributor to the bias in RSA that comes from the temporal properties of fMRI noise ., To understand this , consider that estimating activity pattern by averaging the raw patterns , for instance 6 sec after each event of a task condition ( that is , at the approximate peak of the event-driven HRF ) is equivalent to performing an alternative GLM analysis with a design matrix X6 that has delta functions 6 sec after each event ., Although the columns of this design matrix X6 are orthogonal and ( X 6 T X 6 ) - 1 becomes diagonal , the bias term is still not a diagonal matrix ., Because of the autocorrelation structure Σϵ in the noise , the bias term ( X 6 T X 6 ) - 1 X 6 T Σ ϵ X 6 ( X 6 T X 6 ) - 1 essentially becomes a sampling of the temporal covariance structure of noise at the distances of the inter-event intervals ., In this way , timing structure of the task and autocorrelation of noise together still cause bias in the RSA result ., To illustrate this , we applied RSA to the raw patterns of an independent set of resting state fMRI data from the Human Connectome Project 28 , pretending that the participants experienced events according to the 16-state task in Fig 1A ., As shown in Fig 1E , even in the absence of any task-related signal spurious similarity structure emerges when RSA is applied to the raw patterns of resting state data ., We then calculated the theoretical bias structure ( X 6 T X 6 ) - 1 X 6 T Σ ϵ X 6 ( X 6 T X 6 ) - 1 for each task sequence based on X6 of that sequence and Σϵ estimated as the average noise temporal correlation matrix of the resting state data of three other participants ( right figure of Fig 1E ) ., The off-diagonal elements of all the similarity matrices based on raw patterns were significantly correlated with the theoretical bias structure ( the largest Bonferroni-corrected p-value of Pearson correlation is 0 . 0007 ) and 51 ± 18% of the variance in the off-diagonal elements can be explained by the theoretical bias ., Second , averaging raw data 6 sec after events of interest over-estimates the similarity between neural patterns of adjacent events , an effect independent of the fMRI noise property ., This is because the true HRF in the brain has a protracted time course regardless of how one analyzes the data ., Thus the estimated patterns ( we denote by β ^ 6 ) in this approach are themselves biased due to the mismatch between the implicit HRF that this averaging assumes and the real HRF ., The expectation of β ^ 6 becomes E β ^ 6 = E ( X 6 T X 6 ) - 1 X 6 T Y = E ( X 6 T X 6 ) - 1 X 6 T ( X β + ϵ ) = ( X 6 T X 6 ) - 1 X 6 T X β instead of β ., Intuitively , X temporarily smears the BOLD patterns of neural responses close in time but ( X 6 T X 6 ) - 1 X 6 T only averages the smeared BOLD patterns without disentangling the smearing ., β ^ 6 thus mixes the BOLD activity patterns elicited by all neural events within a time window of approximately 12 sec ( the duration of HRF ) around the event of interest , causing over-estimation of the similarity between neural patterns of adjacent events ., If the order of task conditions is not fully counterbalanced , this method would therefore still introduce into the estimated similarity matrix a bias caused by the structure of the task ., Similar effect can also be introduced if β ^ is estimated with regularized least square regression 29 ., Regression with regularization of the amplitude of β ^ trades off bias in the estimates for variance ( noise ) ., On the surface , reducing noise in the pattern estimates may reduce the bias introduced into the similarity matrix ., However , the bias in β ^ itself alters the similarity matrix again ., For example , in ridge regression , an additional penalization term λβT β is imposed for β of each voxel ., This turns estimates β ^ to β ^ = ( X T X + λ I ) - 1 X T Y . The component contributed to β ^ by the true signal Xβ becomes ( XT X + λI ) −1 XTXβ ., As λ increases , this component increasingly attributes neural activity triggered by other task events near the time of an event of interest to this event’s activity ., Therefore , this method too would overestimate pattern similarity between adjacent events ., In all the derivations above , we have assumed for simplicity of illustration that the noise in all voxels has the same temporal covariance structure ., In reality , the autocorrelation can vary over a large range across voxels ( Fig 1G ) ., So the structured noise in each voxel would follow a different distribution ., Furthermore , the spatial correlation in noise means the noise in β ^ is also correlated across voxels ., Because noise correlation between voxels violates the assumption of Pearson correlation that observations ( activity profiles of different voxels ) are independent , the p-values associated with the correlation coefficients will not be interpretable ., Although we made these simplified assumption for ease of illustration , in the model development below , variation of auto-correlation across voxels and spatial noise correlation are both considered in our proposed method ., As shown above , the covariance structure of the noise in the point estimates of neural activity patterns β ^ leads to bias in the subsequent similarity measures ., The bias can distort off-diagonal elements of the resulting similarity matrix unequally if the order of task conditions is not fully counterbalanced ., In order to reduce this bias , we propose a new strategy that aims to infer directly the covariance structure U that underlies the similarity of neural patterns , using raw fMRI data ., Our method avoids estimating β ^ altogether , and instead marginalizes over the unknown activity patterns β without discarding uncertainty about them ., The marginalization avoids the structured noise introduced by the point estimates , which was the central cause of the bias ., Given that the bias comes not only from the experimental design but also from the spatial and temporal correlation in noise , we explicitly model these properties in the data ., We name this approach Bayesian RSA ( BRSA ) as it is an empirical Bayesian method 30 for estimating U as a parameter of the prior distribution of β directly from data ., Although Fig 3 shows that BRSA reduces bias , it does not eliminate it completely ., This may be due to over-fitting to noise ., Because it is unlikely that the time course of intrinsic fluctuation X0 and the design matrix X are perfectly orthogonal , part of the intrinsic fluctuation cannot be distinguished from task-related activity ., Therefore , the structure of β0 , the modulation of intrinsic fluctuation , could also influence the estimated U ^ when SNR is low ., For instance , in Fig 3F , at the lowest SNR and least amount of data ( top left subplot ) , the true similarity structure is almost undetectable using BRSA ., Is this due to large variance in the estimates , or is it because BRSA is still biased , but to a lesser degree than standard RSA ?, If the result is still biased , then averaging results across subjects will not remove the bias , and the deviation of the average estimated similarity structure from the true similarity structure should not approach 0 . To test this , we simulated many more subjects by preserving the spatial patterns of intrinsic fluctuation and the auto-regressive properties of the voxel-specific noise in the data used in Fig 3 , and generating intrinsic fluctuations that maintain the amplitudes of power spectrum in the frequency domain ., To expose the limit of the performance of BRSA , we focused on the lower range of SNR and simulated only one run of data per “subject” ., Fig 4A shows the quality of the average estimated similarity matrix with increasing number of simulated subjects ., The average similarity matrices estimated by BRSA do not approach the true similarity matrix indefinitely as the number of subjects increase ., Instead , their correlation saturates to a value smaller than 1 . This indicates that the result of BRSA is still weakly biased , with the bias depending on the SNR ., It is possible that as the SNR approaches 0 , the estimated U ^ is gradually dominated by the impact of the part of X0 not orthogonal to X . This bias is not due to underestimating the number of regressors in X0 ( see Part 6 The effect of the number of nuisance regressors on BRSA performance of S1 Material ) ., We leave investigation of the source of this bias to future work ., Empirically , the algorithm 35 we use to estimate the number of regressors in X0 yields more stable and reasonable estimation than other methods we have tested ( e . g . , 36 ) ., It should be noted that BRSA still performs much better than standard RSA , for which the correlation between the estimated similarity matrix and the true similarity matrix never passed 0 . 1 in these simulations ., The expected bias structure when spatial noise correlation exists is difficult to derive ., We used ( XT X ) −1 as a proxy to evaluate the residual bias in the estimated similarity using BRSA ., As expected , when the SNR approached zero , the model over-fit to the noise and the bias structure increasingly dominated the estimated structure despite increasing the number of simulated participants ( Fig 4B ) ., This observation calls for an evaluation procedure to detect over-fitting in applications to real data , when the ground truth of the similarity structure is unknown ., One approach to assess whether a BRSA model has over-fit the noise is cross-validation ., In addition to estimating U , the model can also estimate the posterior mean of all other parameters , including the neural patterns β of task-related activity , β0 of intrinsic fluctuation , noise variances σ2 and auto-correlation coefficients ρ ., For a left-out testing data set , the design matrix Xtest is known given the task design ., Together with the parameters estimated from the training data as well as the estimated variance and auto-correlation properties of the intrinsic fluctuation in the training data , we can calculate the log predictive probability of observing the test data ., The unknown intrinsic fluctuation in the test data can be marginalized by assuming their statistical property stays unchanged from training data to test data ., The predictive probability can then be contrasted against the cross-validated predictive probability provided by a null model separately fitted to the training data ., The null model would have all the same assumptions as the full BRSA model , except that it would not assume any task-related activity captured by X . When BRSA over-fits the data , the estimated spatial pattern β ^ would not reflect the true response pattern to the task and is unlikely to be modulated by the time course in Xtest ., Thus the full model would predict signals that do not occur in the test data , and yield a lower predictive probability than the null model ., The result of the full BRSA model on training data can therefore be accepted if the log predictive probability by the full model is higher than that of the null model significantly more often than chance ., Over-fitting might also arise when the assumed design matrix X does not correctly reflect task-related activity ., When there is a sufficient amount of data but the design matrix does not reflect the true activity , the estimated covariance matrix U ^ in BRSA would approach zero , as would the posterior estimates of β ^ ., In this case as well , the full model would perform worse than the null model , because the form of the predictive likelihood automatically penalize more complex models ., We tested the effectiveness of relying on cross-validation to reject over-fitted results using the same simulation procedure as in Fig 3 , and repeated this simulation 36 times , each time with newly simulated signals and data from a new group of participants in H

Introduction, Results, Discussion, Materials and methods

The activity of neural populations in the brains of humans and animals can exhibit vastly different spatial patterns when faced with different tasks or environmental stimuli ., The degrees of similarity between these neural activity patterns in response to different events are used to characterize the representational structure of cognitive states in a neural population ., The dominant methods of investigating this similarity structure first estimate neural activity patterns from noisy neural imaging data using linear regression , and then examine the similarity between the estimated patterns ., Here , we show that this approach introduces spurious bias structure in the resulting similarity matrix , in particular when applied to fMRI data ., This problem is especially severe when the signal-to-noise ratio is low and in cases where experimental conditions cannot be fully randomized in a task ., We propose Bayesian Representational Similarity Analysis ( BRSA ) , an alternative method for computing representational similarity , in which we treat the covariance structure of neural activity patterns as a hyper-parameter in a generative model of the neural data ., By marginalizing over the unknown activity patterns , we can directly estimate this covariance structure from imaging data ., This method offers significant reductions in bias and allows estimation of neural representational similarity with previously unattained levels of precision at low signal-to-noise ratio , without losing the possibility of deriving an interpretable distance measure from the estimated similarity ., The method is closely related to Pattern Component Model ( PCM ) , but instead of modeling the estimated neural patterns as in PCM , BRSA models the imaging data directly and is suited for analyzing data in which the order of task conditions is not fully counterbalanced ., The probabilistic framework allows for jointly analyzing data from a group of participants ., The method can also simultaneously estimate a signal-to-noise ratio map that shows where the learned representational structure is supported more strongly ., Both this map and the learned covariance matrix can be used as a structured prior for maximum a posteriori estimation of neural activity patterns , which can be further used for fMRI decoding ., Our method therefore paves the way towards a more unified and principled analysis of neural representations underlying fMRI signals ., We make our tool freely available in Brain Imaging Analysis Kit ( BrainIAK ) .

We show the severity of the bias introduced when performing representational similarity analysis ( RSA ) based on neural activity pattern estimated within imaging runs ., Our Bayesian RSA method significantly reduces the bias and can learn a shared representational structure across multiple participants ., We also demonstrate its extension as a new multi-class decoding tool .

learning, medicine and health sciences, diagnostic radiology, functional magnetic resonance imaging, engineering and technology, signal processing, social sciences, random variables, mathematical models, neuroscience, covariance, learning and memory, magnetic resonance imaging, simulation and modeling, cognitive psychology, mathematics, brain mapping, white noise, neuroimaging, research and analysis methods, imaging techniques, resting state functional magnetic resonance imaging, mathematical and statistical techniques, probability theory, psychology, radiology and imaging, diagnostic medicine, signal to noise ratio, biology and life sciences, physical sciences, cognitive science

journal.pcbi.1004379

Interdomain Contacts Control Native State Switching of RfaH on a Dual-Funneled Landscape

It has been more than 50 years since the protein-folding problem was first proposed 1 ., Since then , several experimental 2 , 3 and theoretical approaches 4 , 5 have deepened our understanding of the energy landscape that guides a protein to its unique , thermodynamically-stable three-dimensional structure , the so-called native state , required to carry out its biological function 6 ., However , the concept of the unique native state and the “one sequence/one fold” paradigm are challenged by transformer proteins 7 that are able to adopt multiple , highly-dissimilar but thermodynamically-stable configurations ., Several proteins capable of transforming into another native state in response to their cellular environment have been described , such as the ribosomal protein L20 from Aquifex aeolicus8 and the human chemokine lymphotactin 9 , the latter being extensively studied both experimentally 10 and computationally 11 ., In both proteins the native state switching involves transitions between unrelated regions: the unfolding of one region of the protein is accompanied by folding of a different region ., In other cases , such as the human mitotic spindle protein Mad2 12 , the structural transition involves conformational rearrangements where several secondary structure elements are maintained while the tertiary structure contacts are reorganized ., Another example is the membrane-fusion homotrimer glycoprotein hemagglutinin from the influenza virus , where a metastable fold is created by cleaving a precursor protein , which , upon release by changes in pH , undergoes a large-scale secondary , tertiary and quaternary structural rearrangement crucial for delivering the viral contents into host cells 13 , 14 ., Recently , an extreme case of a structural transformation has been described for the virulence regulator RfaH from Escherichia coli , which belongs to the NusG family of transcription elongation factors present in all three domains of life 15 ., These proteins contain an α/β N-terminal domain ( NTD ) that binds to RNA polymerase ( RNAp ) and acts as a processivity clamp that locks around the transcribed DNA 15 ., The NTD is connected through a flexible linker to the C-terminal domain ( CTD ) that in most NusG proteins is folded as a β-barrel 16 ., In contrast , though still connected by a flexible linker , the CTD of RfaH folds as an α-helical hairpin that is stabilized into tight association with the NTD through interdomain interactions 17 ( Fig 1 ) ., In this conformation , the CTD plays an autoinhibitory role by occluding the RNAp binding site of the NTD and preventing RfaH binding to the transcription complexes in the absence of a recruitment DNA signal ., Strikingly , NMR studies revealed that the isolated CTD folds into the five-stranded β-barrel structure seen in other NusG-like proteins ( Fig 1 ) 18 ., The ability of the CTD to refold from an α-helical hairpin into a β-barrel has been also evaluated in the context of the full protein by several approaches ., First , destabilization of interdomain interactions through disruption of the salt bridge between residues E48 from the NTD and R138 from CTD allows coexistence of both folds at equimolar equilibrium 18 ., Second , proteolytic cleavage of the flexible linker that connects both domains through an engineered TEV site wherein leads to refolding of the CTD into the β conformation 18 ., Finally , domain swapping of the CTD and NTD does not affect the structure and function of RfaH , reinforcing the idea that interdomain contacts are the key factor determining the CTD fold 19 ., These observations suggest that the CTD spontaneously refolds into a β-barrel upon domain dissociation ( Fig 1 ) , an event that is thought to be triggered when RfaH binds to its target ops ( operon polarity suppressor ) DNA 17 ., In this scenario , domain dissociation enables the protein to bind to the ops-paused RNAp and permits the conformational transition of the CTD towards the β fold , which binds to the ribosomal protein S10 similarly to E . coli NusG 18 ., Contacts with S10 are thought to enable the dramatic activation of RfaH-dependent operons by a combination of two mechanisms: recruitment of the ribosome to mRNA in lieu of a missing Shine-Dalgarno element 18 and subsequent coupling of transcription and translation that inhibits premature termination of RNA synthesis by Rho 20 ., The dramatic conformational change of RfaH constitutes an intriguing problem by itself , since the folding mechanism underlying the structural rearrangements that occur during the transformation process is currently unknown ., In addition , the detailed analysis of RfaH transformation will provide new insights about massive conformational changes towards alternative native or misfolded states that occur in other proteins ., In this regard , computer simulations can provide important information about these conformational changes and at the same time overcome many of the difficulties that may arise while following these structural rearrangements experimentally ., Studies of the structural transitions during the α-to-β conversion of the isolated CTD of RfaH using molecular dynamics with empirical force fields have been recently described 21 , 22 , which hint at the presence of partially unfolded intermediates on the folding pathway ., However , these simulations do not include the NTD of RfaH and thus neglect any involvement of the interdomain contacts shown to thermodynamically control the transformation process ., Inspired by this and by the fact that all the information required to determine the CTD fold is encoded by RfaH itself 19 , we investigated the dramatic conformational change of the CTD of RfaH in the context of the full protein using structure-based models 23 that have been developed based on the energy landscape theory 24 and the principle of minimal frustration 4 ., These models are biased towards the native state by the explicit inclusion of its topology into the energy Hamiltonian , such that all native interactions are stabilizing ., The robustness of these models has been demonstrated by the reproduction of the observed folding and binding mechanism of several proteins 5 , 25 , and their applications have been recently extended to the study of complex folding mechanisms by generalizing to multiple-basin energy landscapes 26–31 ., Using these dual-basin structure-based models , we were able to follow the reversible interconversion between the α and β folds of the CTD of RfaH in the context of the full-length protein ., Our results show that the structural transition between the folds is connected through an obligate intermediate , and that weakening of the interdomain contacts is sufficient to trigger the interconversion ., The structural features of the intermediate states described herein are consistent with local frustration and secondary structure propensity analysis of the CTD ., Moreover , our model allowed us to define the interdomain residues that are most responsible for controlling folding-upon-binding of the CTD into the α state ., These results are in excellent agreement with the current experimental evidence of the dramatic conformational transition of RfaH and provide new insights into its mechanism ., The folding of proteins is typically well described by structure-based models because a protein’s funneled energy landscape is selected to be consistent with the structure of the native state 4 , 32 ., In the case of RfaH , the structure of its CTD has been solved either in the context of the full protein by X-ray crystallography 17 or in isolation by NMR 18 , showing striking structural differences ., In the full protein , the folded state of the CTD corresponds to an α-helical hairpin that establishes extensive contacts with the NTD 17 ., However , the isolated CTD folds into a five-stranded β-barrel 18 observed in the homologous NusG-like transcription factors from bacteria , archaea and eukaryotes 33 ., Both folded states represent low free energy ensembles that the same sequence can fold into ., Therefore , in RfaH , evolution has selected a sequence that is consistent with two structures , which can be represented with a dual-basin structure-based model ., In this case , the enthalpy contributions from both folds are combined such that both structures of the CTD are explicit energy minima ., This dual-basin approach has been previously used to study the competing formation of symmetry-related native and mirror structures of Rop dimer 26 , 34 , 35 and the B domain of protein A 36 and the large-scale structural rearrangement of the human chemokine lymphotactin 29 and the influenza virus glycoprotein hemagglutinin 14 ., The thermodynamics of the dual-basin model of RfaH is consistent with experimental findings ( Fig 2 ) ., First , when connected to the NTD , the thermodynamic minimum of the CTD is the α fold 17 ( Fig 2A , εCIF=ε ) ., Second , when interaction with the NTD is removed , the CTD folds into β ( Fig 2C , εCIF=0 ) ., The β-fold is observed when the CTD is excised from the full RfaH protein by proteolytic cleavage of the interdomain linker 18 ., Finally , there exists an interface stability that allows for coexistence between the folds ( Fig 2B and S1 Fig ) ., Experimentally , both folds were detected when destabilizing mutations such as the NTD substitution E48S were introduced into the interface between the CTD in the α fold and the NTD 18 ., In the simulation , if the overall affinity between the NTD and CTD is reduced by uniformly lowering the strength of the interface contacts by ~50% , α and β are equally probable and exhibit transitions between the states ., The important role of the interface contacts in determining the fate of the CTD is readily understood by considering the differences in structure between the α and β folds ., The α-helical CTD forms a large interface with NTD , while the β-barrel buries many of the interface residues involved in these contacts ., Therefore , decreasing the strength of these contacts destabilizes α more than β ( Fig 2D ) ., Landscapes for additional intermediate levels of εCIF are shown in S2 Fig . At all levels of interface contact strength , a subset of CTD lies in intermediate configurations ., In the next section we discuss the role of these intermediates in the folding route connecting α and β ., Although the wild-type RfaH is only known to exist with the CTD bound in the α fold in the absence of the transcription elongation complex ( TEC ) , the use of domain-swapped 19 and single-residue 18 mutants provided strong experimental evidence of interconversion between the α and β folds of the CTD in the context of the full-length protein ., Domain swapping suggested that the protein can fold back into the α fold even when the CTD is the first element to be translated 19 ., The NTD substitution E48S destabilizes the interface such that the CTD coexists in both folds at equimolar equilibrium 18 ., Therefore , a model where the strength of the interface contacts is tuned so that both CTD folds are equally probable , as when εCIF=0 . 51ε , is not only useful for describing the interconversion pathway for the wild-type protein but also describes protein models that are experimentally realizable ., The energy landscape presented for RfaH when εCIF=0 . 51ε shows that its native basins are connected through obligate intermediate configurations ( Fig 2B ) ., As a control , to verify that the intermediate ensembles are not an artifact caused by our choice of dihedral mixing , we also performed simulations using a dual-basin dihedral potential as described elsewhere 14 ., This potential further stabilizes the intermediate ( S3 Fig ) ., The transformation process takes place in the context of the full-length protein and involves interactions between the domains ., The fraction of interface contacts QIF quantifies the level of interaction between NTD and CTD , while an RMSD difference , RMSDβ–RMSDα , measures the structural state of the CTD ( Fig 3 ) ., The free energy landscape along these coordinates shows that β and α are connected through two intermediate states , I1 and I2 , and that these intermediates are forming contacts with the NTD ( Fig 3A ) ., β and I1 are populated both at QIF = 0 and QIF > 0 , while I2 and α are only populated when interacting with the NTD ( Fig 3B ) ., α is fully populated when the fraction of interdomain contacts exceeds 75% ( Fig 3B ) ., Hence , our data suggests a three-state folding process β/I1 <-> I2 <-> α , where the interconversion between β/I1 and I2 occurs while interacting with the NTD ., To verify the kinetic relevance of our projection of the free energy landscape in Fig 3 , we performed a long constant temperature simulation and counted the transitions between the different ensembles ( S4 Fig ) ., Transitions only occur between states α <–> I2 , I1 <-> I2 and β <-> I1 , with the latter being most frequent , in line with the low free energy barrier separating these ensembles ., These transitions are consistent with the three-state folding process previously defined ., Additionally , the unfolded state is not sampled whatsoever in these simulations ., Recent simulations using implicit and explicit solvent force fields have suggested that the isolated CTD traverses an intermediate during kinetic simulations of the one-way α-to-β transformations 21 , 22 ., This result is consistent with the β <-> I1 dynamics that can transition without interacting with NTD ., Finally , it is worth noting that the presented folding landscape for RfaH differs from other transformer proteins such as lymphotactin , where stepping into the unfolded state is required 10 ., The intermediates emerge as low free energy combinations of native contacts contributed by the two input contact maps for RfaH ., To structurally describe the intermediate ensembles we determined which native contacts are formed ( Fig 3 ) ., A native interaction is considered formed in these ensembles if their contact probability is greater than 0 . 5 ., I1 is most similar to β ( Fig 3C ) ., Most of the interactions between strands β3-β4 , β1-β5 and a large portion of the contacts between strands β1-β2 and β2-β3 are established ( Fig 3C ) ., This is similar to previous depictions of the α-to-β conversion of the isolated CTD using Markov state models , where strands β2 , β3 and β4 are thought to be formed earlier during the transition towards the β state 22 ., I2 is most similar to α , having most of the interactions between strands β1-β2 and β2-β3 unformed ( Fig 3D ) ., Almost all of the non-local interactions between helix α1 and α2 are formed , but there is still partial unwinding of helix α2 , while α1 seems to be stable ., A higher probability of local contacts in helix α1 differs from molecular dynamics simulations of the isolated CTD , which suggested that this element is less stable 21 , 22 ., However , this discrepancy would be expected as I2 forms extensive interactions with the NTD that can modify its stability ., To gain insight into the intermediate ensembles predicted by the dual-funneled structure-based model , we estimated the local frustration of the α and β folds and the secondary structure propensity of the sequence of RfaH CTD using the protein frustratometer 37 and Jpred-3 38 webservers , respectively ., Local frustration analysis shows that most of the interactions that support robust folding ( i . e . minimally frustrated contacts ) of the α-helical state of RfaH CTD correspond to the non-local interactions between helix α1 and α2 , most of the local interactions of helix α1 and local interactions between residues 139–146 of helix α2 ( S5 Fig ) ., Interestingly , the C-terminal end of helix α2 suggests that this region is highly frustrated ( S5 Fig ) ., Thus , there is consistent evidence from both conformational entropy ( folding simulations using structure-based models ) and native state heterogeneity ( frustration ) for the structure of the intermediate I2 ( Fig 3D ) ., The β fold is highly consistent , only having a small amount of frustration localized in interactions between strands β2–β3 and β3–β4 ( S5 Fig ) ., These features of the β fold are also consistent with the overall structure of the intermediate I1 ( Fig 3C ) ., Lastly , secondary structure prediction based on the sequence of RfaH CTD suggest that residues 136–145 have some helical propensity ( S5 Fig ) , thus being consistent with the presence of helical local interactions that are featured by this region in the intermediate state I1 ( Fig 3C ) ., In line with our results , recent secondary structure prediction analysis of RfaH CTD 39 showed that residues 141–145 encompass a Leucine-rich region ( sequence LLLNL ) , which has a high propensity to adopt helical configurations , whereas the homolog region in NusG is mainly composed by valine and isoleucine , which are known to favor β structures 40 , 41 ., The same fragment is present in several unrelated structures solved in the Protein Data Bank and also exhibit an helical structure 39 , strengthening the idea that the sequence of RfaH CTD has some localized α-helical propensity and that this sequence motif can be used to identify other transformer proteins along the evolution of the NusG family ., To further validate the structural features of the intermediate states predicted by our dual-funneled model during native state switching of RfaH CTD , we performed targeted molecular dynamics ( TMD ) 42 of the α-to-β transformation of the full RfaH protein in explicit solvent ., It is worth noting that the reaction coordinate that steers RfaH towards the β fold in TMD is defined through the RMSD to the target structure , namely the βCTD , and hence there is no direct perturbation of the NTD-CTD interface interactions ., By use of a steering force of 672 kcal·mol-1·Å-2 over the 62 Cα atoms of the CTD , we collected 7 TMD simulations that each successfully reached a β-like fold , as indicated by measurement of the RMSD against the CTD in the β fold ( on average ~0 . 5 nm ) , totaling 140 ns of simulation ., As seen in Fig 4A , the α-to-β transition is accompanied by an increase in distance between the CTD and NTD domains , in a similar fashion as the increase in the fraction of CTD in the β fold at increasing domain distances observed in our dual-funneled models ( Fig 4B ) ., As illustrated in Fig 4C , our α-to-β TMD simulations show that a significant loss of helical structure is observed in helix α2 ( residues 142–151 ) between 8 ns and 12 ns of simulation , before dissociation of the CTD and NTD domains occur ., This observation is fully compatible with the structural features of the I2 intermediate described using dual-funneled models ( Fig 3D ) and with the local frustration analysis of the CTD in the α fold ( S5 Fig ) ., After dissociation , the CTD accumulates extended secondary structure content related to formation of β-strands , although some helical content is still present , thus being compatible with the I1 intermediate previously described ( Fig 3C ) ., Overall , our results provide good evidence that the transformation mechanism of CTD involves intermediate states that share structural features from both folds and that this process is not simply related to its topology , but a combination of the dual basin of CTD and its interface interactions with NTD ., The transformation event triggered by binding of RfaH to the ops-paused RNAp 17 is likely related to allosteric communication between the NTD ops binding site to the NTD-CTD interface ., Thus , understanding how the interface is involved in the transformation between α and β is crucial for understanding the activity of RfaH ., Both the β and I1 states can be populated in the absence of interface interactions ( Fig 3B ) ., Hence , the key binding step allowing the structural change corresponds to the β/I1 <-> I2 transition , since the I2 <-> α occurs with the CTD already bound to the NTD ., Therefore , we calculated the contact probability of each interdomain contact in the transition state ensemble ( TSE ) of this folding step to determine the residues responsible for binding between the NTD and CTD during the conformational change and enabling RfaH to act as a sequence specific regulator of gene expression ., All of the residues that are key for the binding TSE of the I1 <-> I2 step ( i . e . their contact probability is greater than 0 . 5 ) are located in the vicinity of residues E48 and R138 from the NTD and CTD , respectively ., In fact , most of the residues that form the β-hairpin of the NTD ( residues 30–52 ) are involved in binding of the CTD during this folding step ( Fig 5A ) ., Moreover , residue E48 , whose substitution by serine allows experimental observation of the α and β folds of RfaH in 1:1 equilibrium 18 , has a contact probability ( averaged over all contacts where this residue is involved ) of ~0 . 87 , being one of the highest probabilities among all of the NTD interface residues ., In the TSE , the NTD interacts with residues I129 , F130 , E132 , P133 , G135 , E136 , R138 and S139 from the CTD , which are located in the loop connecting helices α1 and α2 and in the first turn of helix α2 ., Remarkably , most of the side chains of these interface residues ( I129 , E132 , E136 , R138 ) are pointing towards the surface in the β fold , therefore being readily available to interact with the NTD ( Fig 5B ) ., This architecture allows β/I1 to interact with the NTD without unfolding the hydrophobic core , significantly lowering the overall barrier to transformation ., The only interface residue involved in the binding TSE that also forms extensive hydrophobic contacts in β is F130 , having the highest number of native contacts per residue in the β fold ( Fig 5C ) , and unfolding it likely creates the small barrier separating β/I1 and I2 ., We tested the importance of F130 on the stability of the CTD in the β fold by first defining the TSE of this fold using single-basin models ., Our simulations show that residues 105–107 , 121–123 , 134–143 and 148–154 have a contact probability in the TSE higher than 0 . 6 and define the folding nucleus ( Fig 5D ) ., It is important to note that these regions describing the folding nucleus of βCTD are also the firsts to exhibit β-strand formation in our TMD simulations ( Fig 4C ) ., Residue F130 has a slightly lesser contribution to the structure of the TSE by having a contact probability of ~0 . 5 in the single-funneled model ., We then performed an in silico mutation of F130 through deletion of all the native contacts that this residue establishes in the β fold ( named βF130 ) and performed simulations at εCIF=0 . 51ε , thus testing the role of residue F130 on the stability of the β fold and the structural transformation of RfaH CTD ., As shown by the contour plot in Fig 5E , removal of these contacts destabilizes the β fold and the intermediates , relative to the α fold of RfaH CTD ., Altogether , these results highlight the dual role of F130 in stabilizing the hydrophobic core of the β fold and interacting with the NTD to stabilize the intermediates ., RfaH is recruited to RNAp paused at the ops site 20 ., While the details of how RNAp and ops initially induce the dissociation of the α fold CTD are not known , we show that having RNAp bound to RfaH is sufficient to maintain the CTD in its β fold ., This is important since RfaH’s function of coupling transcription and translation requires the β-folded CTD to interact with the ribosomal protein S10 ., Interface contacts occluded by RNAp binding were identified by superimposing the NTD of RfaH with its archaeal homologue Spt5 from Pyrococcus furiosus , which forms a heterodimer with Spt4 and is bound to the RNAp clamp domain ( accession code 3QQC , Fig 6A ) 43 ., In this structure , residues 237–280 of the A’ subunit of P . furiosus RNAp form a coiled-coil equivalent to the β’CC of E . coli RNAp 43 ., Residues 255–265 located on the tip of the coiled-coil structure interact with Spt5 and are equivalent to residues 282–292 of E . coli RNAp β’CC , whose replacement by a glycine linker completely disrupts the interaction between RfaH and RNAp 17 ., In the resulting superimposition 53 out of 80 interface contacts are occluded , mostly in the vicinity of residue E48 ( Fig 6A ) ., Removal of these contacts from the dual-basin structure-based model mimics the effect of RNAp binding to the NTD , and leads to a strong destabilization of α ( S6 Fig ) ., At T = 0 . 92 TFβ and εCIF=ε , the populations in β and intermediate states are 76% and 21% respectively , with only 1% of the CTD in the α-helical fold ., Since binding of β’CC does not actually remove the affinity of the CTD for NTD , the interaction with β’CC is actually a biomolecular process where β’CC and CTD compete for the NTD ., We performed simulations where the β’CC of RNAp was explicitly included ., As illustrated in Fig 6B , β’CC competes with RfaH CTD to bind to the NTD when εCIF≤0 . 75ε , and effectively displaces the CTD when εCIF≤0 . 60ε ., Naturally , β’CC binding destabilizes the α fold of CTD by occluding its NTD interface ( Fig 6C ) ., Interestingly , the presence of β’CC raises the interface contact strength of the equilibrium between α and β from εCIF=0 . 51ε to εCIF=0 . 71ε ., If native RfaH has an equilibrium value of εCIF>ε ( since α is dominant in the NMR structure ) , this is consistent with the fact that RNAp alone does not bind RfaH ., Presumably , inclusion of the full RNAp with ops binding site would push the εCIF midpoint sufficiently above ε in order to shift the equilibrium towards bound β’CC and βCTD ., Unfortunately a structure including these interactions is not yet available ., These results can be sufficient to explain how RNAp is able to exclude the CTD from binding and favor its β fold ., The complex α-to-β structural conversion of RfaH-CTD in the context of the full protein can be addressed using dual-basin structure-based models that integrate the topology of both native states into a single Hamiltonian ., Our model is able to reproduce several features of this process that have been experimentally demonstrated or suggested from detailed molecular simulations , such as, i ) the disruption of interdomain interactions enables the coexistence of α and β;, ii ) the large structural change of RfaH as a three-state folding process β/I1 <-> I2 <-> α ., Our results also give new insights about how this folding mechanism is coupled with NTD-CTD binding , the structural features of the intermediate ensembles and the key interdomain residues that permit binding during the β-to-α transformation of the CTD ., Moreover , we propose that residue F130 , which stabilizes several interactions with the hydrophobic core of βCTD and is exposed towards the interdomain interface in the α fold , is key to control the stability of the β fold and the TSE that separates both native basins ., Overall , we find that in the presence of RfaH-NTD , the transformation mechanism of CTD is not simply related to its topology , but a combination of the dual basin of CTD and its interface interactions with NTD ., While most of these results arise from a structure-based model where the strength of the interfacial contacts has been homogeneously tuned to equally populate both folds , we also address a plausible scenario for the specific effect of RNAp after binding to the NTD ., Once interactions of ops with its binding site in RfaH have allosterically triggered domain dissociation and allowed RNAp to bind to the newly exposed NTD surface ( equivalent to reducing the strength of interdomain contacts below 0 . 75ε ) , steric hindrance of the formation of specific interdomain contacts by the RNAp β’CC favors the β fold of RfaH ., While our models overcome many of the challenges that can be found experimentally , the obtained results offer valuable starting points to guide in vitro experiments , such as mutational analysis of the NTD residues predicted to contribute for binding of the CTD and kinetic measurements of mutants of the F130 residue that would either lower the free energy barrier limiting the α-to-β conformational change or destabilize the β fold and favor the inactive state of RfaH , in order to gain a better understanding of the dramatic transformation of the CTD of RfaH ., Our simulations were performed using a coarse-grained structure-based model 5 generated using the SMOG server 44 , where each residue is represented by a single bead centered at the coordinates of its corresponding Cα atom ., In this model , bonds , angles and dihedrals are maintained by harmonic restraints , and non-bonded residues in contact in the native state are given attractive interaction while all other non-local interactions are treated as repulsive , as described in ref . 5 ., The terms r0 , θ0 , ϕ0 correspond to the values of bonds , angles and dihedrals in the native structure ., The parameters εr = 100ε , εθ = 20ε , εϕ = ε , εNC = ε weight the strength of each type of interaction ., The functional form of the contact potential is:, Vcontactssb=∑ij\xa0∈\xa0contactsεC5 ( σij0rij ) 12−6 ( σij0rij ) 10, ( 2 ), Where σij0 is the distance between the residue pair i , j Cα atoms in the native state and εC is the energy of the native contact ., The native contact maps for the full RfaH protein with its CTD in the α fold and for the CTD in the β fold were determined from structures deposited in the Protein Data Bank 45 with accession codes 2OUG and 2LCL , respectively ( Fig 1 ) ., Loop residues 101–114 not solved in the crystal structure of the full RfaH protein were modeled using MODELLER 46 and were given no native contacts in the α fold ., This approach is justified because small deletions , insertions , and substitutions in this loop do not affect RfaH function and thus presumable folding ( IA , unpublished ) ., The native contact map between residues separated in sequence by at least two amino acids ( i > j + 2 ) was determined from each structure using the shadow map algorithm 47 ., In order to account for the α-to-β conformational transition of the CTD of RfaH , the native contact potentials determined for both folds were combined as in Sutto et al 29:, Vcontactsdb=∑ij\xa0∈\xa0α\xa0non-interfacecontactsεC5 ( σijαrij ) 12−6 ( σijαrij ) 10+∑ij\xa0∈\xa0βcontactsεC5 ( σijβrij ) 12−6 ( σijβrij ) 10+∑ij\xa0∈\xa0α\xa0interfacecontactsεCIF5 ( σijαrij ) 12−6 ( σijαrij ) 10, ( 3 ), Where σijα is the distance between the residue pair i , j Cα in the α fold ( accession code 2OUG ) , σijβ is the distance between the residue pair i , j Cα in the β fold ( accession code 2LCL ) , εC is the energy of the native contacts in the α and β folds , respectively , and εCIF is the energy of the interfacial contacts formed between the NTD and CTD of RfaH in the α fold ( accession code 2OUG ) ., In our simulations , the energy of native contacts in the α and β folds were equally weighted ( εC = kBT* = ε ) , while the energy of the interdomain contacts εCIF was varied in the range {0 , ε} to investigate the interplay between binding interface contacts and folding ., The chosen sequence separation of two residues was adopted instead of the typical contact map definition of i > j + 3 due to two observations:, i ) simulations using the latter sequence separation gave rise to the presence of the intermediate I2 even when the strength of the interdomain contacts equaled ε ( S7 Fig ) , while relaxation rates derived from NMR experiments on the wild-type protein demonstrated tight domain interactions and preservation in solution of the inactive structure of RfaH solved by crystallography 18;, ii ) decreasing the strength of interdomain contacts on the dual-funneled model with a sequence separation of at least 3 residues significantly increased the population of the intermediates states , being higher than 70% when equilibrium between the α and β folds was achieved ( εCIF=0 . 70ε , S7 Fig ) , but there are no detectable intermediate configurations based on the signal from NMR experiments using the E48S mutant that reaches 1:1 equilibrium between both CTD folds 18 ., Theref

Introduction, Results and Discussion, Methods

RfaH is a virulence factor from Escherichia coli whose C-terminal domain ( CTD ) undergoes a dramatic α-to-β conformational transformation ., The CTD in its α-helical fold is stabilized by interactions with the N-terminal domain ( NTD ) , masking an RNA polymerase binding site until a specific recruitment site is encountered ., Domain dissociation is triggered upon binding to DNA , allowing the NTD to interact with RNA polymerase to facilitate transcription while the CTD refolds into the β-barrel conformation that interacts with the ribosome to activate translation ., However , structural details of this transformation process in the context of the full protein remain to be elucidated ., Here , we explore the mechanism of the α-to-β conformational transition of RfaH in the full-length protein using a dual-basin structure-based model ., Our simulations capture several features described experimentally , such as the requirement of disruption of interdomain contacts to trigger the α-to-β transformation , confirms the roles of previously indicated residues E48 and R138 , and suggests a new important role for F130 , in the stability of the interdomain interaction ., These native basins are connected through an intermediate state that builds up upon binding to the NTD and shares features from both folds , in agreement with previous in silico studies of the isolated CTD ., We also examine the effect of RNA polymerase binding on the stabilization of the β fold ., Our study shows that native-biased models are appropriate for interrogating the detailed mechanisms of structural rearrangements during the dramatic transformation process of RfaH .

To carry out their biological functions , proteins must fold into defined three-dimensional structures ., In most proteins , a single fold determined by the amino acid sequence , and sometimes influenced by environmental conditions , is believed to be suited for each protein’s dedicated task ., However , some proteins challenge this broadly accepted paradigm , adopting different structures that can enable diverse roles or trigger pathological responses , such as prion diseases ., Escherichia coli RfaH constitutes a dramatic example of this atypical behavior ., RfaH C-terminal domain folds into either a helical bundle that binds to the N-terminal domain and inhibits unregulated recruitment to the transcription complex or , in the presence of a specific DNA target , into a stand-alone β-barrel structure that binds to the ribosome and couples transcription and translation of RfaH-dependent genes ., To understand the mechanism of this structural rearrangement , we performed molecular dynamics using a model where the stabilizing interactions from both folds are integrated ., Our results argue that this transformation requires destabilization of the domain interface , is favored by interactions between the N-terminal domain of RfaH and RNA polymerase , and proceeds via a bound intermediate state that connects both folds .

journal.pcbi.1003861

Reactive Searching and Infotaxis in Odor Source Localization

The efficiency of male moths searching for females is astonishing ., In spite of serious difficulties , as , for example , large distances , sharp time constraints , and sparse discontinuous clues , their olfactory pheromone system usually guarantees a successful encounter 1 , 2 ., Far from the pheromone emitting female , odor plumes consist of sparsely distributed pheromone patches 3 , leading to rare , intermittent detections 4–6 ., This mating race is not only fascinating in itself but also particularly convenient to study the chain linking perception to action , and for the investigation of search tasks in general ., The advantages are the rich adaptive behavioral repertoire of insects generated by a relatively simple neuronal system 7 , 8 , a clear instinct-based task , and its suitability for testing and comparing different types of search strategies ., Moths , as well as other insects , have developed a specifically adapted behavior , in addition to a specialized neuronal subsystem for the processing of pheromone information 9 , 10 ., Experimental evidence indicates a two-step behavioral strategy 1 , 5 , 11–13: sensing a pheromone patch induces an upwind surge 5 , towards the pheromone emitting source ( the female ) ., Upon loosing the scent , they switch to crosswind ( zigzag ) casting 11 , 12 , 14–16 , or looping or spiraling 14 , 15 , 17 , 18 ., Spiraling is typically done by walking insects ., An important factor is the olfactory stimulus 4 , 5 , 14 , e . g . , the pheromone dose or the pulsation frequency ., The latter relates to another important factor , the presence of an air flow in odor-modulated anemotaxis ., Inspired by the observations detailed above , various models of reactive search strategies have been suggested and modified 2 , 19–21 ., They are based on predefined movement sequences which are typically triggered by odor perceptions ., Such biologically inspired strategies can be employed to locate pheromone sources 22 or other odors 23–26 , given appropriate sensors ., In general , bio-inspired methods are widely discussed to overcome the challenges in chemical sensing 26 , 27 ., A powerful alternative to reactive searching is the more sophisticated and computationally rather expensive approach of using cognitive strategies , e . g . 6 , 28 , 29 with respect to searching without continuous ( or smooth ) chemical gradients ., Such methods produce an adaptive behavior as current perceptions are weighted by past clues and actions , i . e . , learning and memory are typically involved ., The infotaxis strategy 6 is based on Bayesian inference to maximize the information gain about the location of the source in a turbulent medium ., Originally implemented as a simulation , infotaxis can be used in a real-world set-up in combination with a robot in order to track a thermal source 29 , 30 ., Such experimental tests of theoretical models are particularly important for applied research as they point out real-world issues and limitations that mere simulations cannot account for ., This is true for both studies involving reactive 8 , 26 and cognitive strategies 31 , 32 , as well as for gradient based underwater chemo-orientation using a biomimetic robot lobster 33 ., Another example for such studies is biologically-inspired chemical plume tracing using an autonomous underwater vehicle 23 , 24 , 34 ., Moreover , the interplay between robotics and insects offers the possibility to investigate the insects behavior 8 , 35–37 while specifically modifying the experimental conditions ., Using such an approach , we compare the strategies detailed above: reactive versus cognitive searching in dependence of the stimulation strength for a turbulent air flow ., Our cyborg , a male moth Agrotis ipsilon mounted on a mobile robot 22 , has to find the pheromone source located two meters upwind from the starting position , see Fig . 1 ., The turbulent air flow yields sparsely distributed pheromone patches ., Their detections , recorded from the antenna , control the robotic movements via three biologically motivated reactive search strategies and one cognitive infotaxis strategy 6 ., The reactive strategies are derived from an analysis of electrophysiological recordings from the macroglomerular complex ( MGC ) , the first cerebral relay for the processing of pheromone information perceived by olfactory receptor neurons ( ORNs ) in the moths antennal lobe 9 , 38 ., They show triphasic neuronal responses ( On , inhibition , Off ) to pheromone stimulation 22 , 39 ., The origin of the inhibitory phase was the focus of 22 who employed reactive searching implemented on a cyborg in order to provide evidence that reactive searching could be mediated by multiphasic responses ., We here provide a detailed comparison of cognitive versus reactive searching , in dependence of the stimulus dose , focusing on the analysis of the resulting trajectories ., Based on the occurrence of multiphasic responses , we consider the strategies detailed in Fig . 2: ( sp ) combines surge and arithmetic spiral casting , ( za ) and ( ze ) combine surge and a two-step casting sequence composed of crosswind casting , i . e . , zigzagging followed by arithmetic or exponential spiraling , respectively ., The single movement sequences are motivated by biological findings on behavioral insect data: straight upwind surge 2 , 5 , zigzagging with an increasing step size 2 , 40 , and spiral casting 15 , 18 ., In addition , we contrast our search trajectories to those resulting from behavioral experiments where a walking silkmoth tracks a pheromone source in a wind tunnel 36 ., We consider in this article the following questions: is complex cognitive searching superior to using simple reactive strategies ?, What is the influence of the stimulus strength ?, How can the resulting search trajectories be characterized and compared ?, Do they show common features and how do they relate to behavioral data ?, We expected a clear and overall predominance of infotaxis since the algorithm involves memory and learning , and it has been proven to be very successful in computer simulations 6 , 41 , 42 ., We found , however , that reactive searching can be more efficient if it includes a response reminiscent of casting ., We first present the biological motivation of the implemented reactive search strategies ., We then detail the results of using these strategies , as well as the infotaxis algorithm , in cyborg experiments ., On the one hand , we focus on basic features as success rate and efficiency , expressed by the corresponding trajectory lengths ., On the other hand , we also aim at a more detailed characterization of these search paths , including a qualitative comparison to behavioral data 36 ., Finally , we discuss our findings , in particular in terms of which approach is more efficient , under which circumstances , and why ., We consider the electrophysiological MGC recordings of 8 multiphasic neurons and 6 monophasic neurons ( Fig . 2 , in total 58 and 53 single trials , respectively ) ., The Off phase in multiphasic neuronal responses to pheromone stimulation in the MGC is apparent in the firing rate 9 , 22 , 39 ., We find that it is also apparent in the firing regularity and firing reliability which are characterized by the coefficient of variation and the Fano Factor , respectively ., Fig . 2A shows the spike times of seven trials of a multiphasic MGC neuron ., Upon pheromone stimulation , firing abruptly increases from the baseline ( Bl ) and produces an On response , followed by an inhibitory phase ., After inhibition , neuronal firing restarts at a higher rate than Bl activity which then slowly decays back to baseline ., We call this transient intermediate phase the Off response ., The Off is apparent in the Peri-Stimulus-Time-Histogram ( PSTH , the average firing rate over time ) , in terms of spiking regularity and reliability , as shown in Fig . 2B ., Both the coefficient of variation and the Fano Factor are close to one during irregular and unreliable Bl firing , drop to approximately zero during the On , and show intermediate values during the Off while gradually reapproaching one ., In contrast , the correlation coefficient , representing spike time precision ( Methods , part 1 ) , exhibits no Off phase ., Both the increase in synchrony , i . e . spike time precision , from baseline ( 0 ) to On ( 1 ) and the decrease back to zero are abrupt ., For comparison , we also analyze monophasic MGC responses to pheromone stimulation , shown in Fig . 2C ., Such neurons also respond with an On , i . e . , an abrupt increase in firing ( data not shown , see 22 , Fig . 2 ) , similar to the On response of multiphasic neurons: there 1 , 0 , 0 ) ., Yet , for monophasic neurons the increased spiking during the On switches immediately back to baseline spiking , there is neither an inhibitory , nor an Off phase ., Both firing regularity and reliability switch directly back to their baseline values: 0 . 7 and 1 , respectively ., On the right side of Fig . 2 we present the reactive search strategies associated with the different in the neuronal MGC responses to pheromone stimulation ., Note that the wind is assumed to blow from the top and that mean wind direction and speed are fixed ., Basically , we distinguish between two assumptions: the sp strategy neglects the Off while the other two strategies ( za and ze ) comprise a zigzag casting sequence representing the Off in our multiphasic neurons ( see Methods , part 2 ) ., The sp strategy could thus model the behavior based on the activity of monophasic neurons ., As initial movement , we choose arithmetic or exponential spiraling which represents baseline firing ., We assume that each detection event initiates a straight upwind surge representing the On ., If there is no subsequent detection , the movement changes either into Off zigzagging followed by baseline spiraling , or it directly switches back to baseline spiraling ( sp , using arithmetic spirals ) ., Note that the cyborg stops zigzagging after a fixed period of 19 s ( if there is no further detection ) : as we record from the antenna and not from the MGC , we do not have access to the length of the Off ., Our za and ze strategies combine zigzagging with either arithmetic or exponential spirals 43 in order to test whether exponential spiraling yields an increases in efficiency 22 , Suppl . Information ., We do not consider reactive searching without spiral movements as we expect relatively high failure rates due to the missing downwind component 25 ., Consequently , the agent cannot reorient appropriately after passing ( and missing ) the source ., This happens , for example , whenever detections occur close to , but downwind and laterally shifted with respect to the source position ., We now briefly introduce the infotaxis algorithm 6 ., It uses Bayesian inference to localize the source of an odor plume in a turbulent medium ., It combines two mechanisms: exploitation ( i . e . , approaching the source based on perceived information ) and exploration ( i . e . , maximizing the information gain ) ., The exploration mode predominates if there is nearly no information available ., Long periods of time with no odor encounter broaden the posterior distribution and compel the agent to explore the environment in large patterns ., On the contrary , if many detections indicate that the source is close , the exploitation mode triggers more localized movements ., Starting from a prior probability distribution for the location of the source , the agent accumulates information while exploring its environment ., Both odor detections and non-detections contribute to the ongoing update of the estimated spatial probability distribution ., Infotaxis has been shown to function very well in computer simulations ( see Introduction ) : typical failure rates are close to zero , even for ‘no wind’ or ‘no stimulus’ conditions ( the agent continues to search until the source is found ) ., Its movements are discretized steps and it allows only for four movement directions ., For each step , the direction is determined as that which maximizes the entropy reduction in the probability distribution of the source location ., More details are given in Methods , part 3 ., We now compare the results of our cyborg experiments , i . e . , search trajectories obtained for three reactive search strategies and infotaxis stimulated with different pheromone concentrations ( no pheromone , minimum , medium , and maximum concentration , see Methods , part 2 ) ., The number of successful and failure trials are given in Table 1 ., With Fig . 6 we aim at a qualitative comparison between our strategy driven trajectories and behavioral data provided by the Kanzaki-Takahashi Laboratory 36 ., There , a silkmoth ( Bombyx mori ) walking in a wind tunnel tries to locate a pheromone source ( relatively high pheromone concentration ) located 60 cm away from the starting position 8 , 36 , 37 ., Large parts of the trajectories are estimated to lay inside the odor plume ., We show three arbitrarily selected behavioral search paths and the corresponding track-angle distribution ., These are much broader than our histograms ( Fig . 4C , 4D and 5C ) but there is a central ( surge ) peak , as well as a second peak due to zigzagging , though here occurring at 100° and non-symmetric ( probably due to the low number of samples ) ., In order to permit a qualitative mapping , we selected two typical examples of reactive paths , as well as two representative examples of cognitive search paths , shown in Fig . 6A and 6B , respectively ., Since the behavioral data ( Fig . 6C ) was obtained for a relatively strong stimulation , we here neglected cyborg data obtained for minimum or no stimulation , as well as particularly long or widespread paths ., Behavioral trajectories exhibit no spirals but there are some circular sections , e . g . , far from the source in trial 3 , as well as some looping sequences ( trial 2 ) ., The lack of spirals in behavioral compared to our reactive trajectories ( Fig . 6A ) could be due to the small distance between start and source ( 0 . 7 m ) or due to cumulative navigational errors , similar to our odometry errors ., For example , it has been suggested that the homing paths of desert ants are actually distorted spirals 18 ., The insect-controlled trajectories in Fig . 6C are clearly dominated by zigzag patterns , but without increasing lateral amplitudes — as we assumed ( Methods , part, 2 ) for our za and ze strategies 2 , 40 ., Our reactive search paths contain many straight upwind movements representing surge sequences ( Fig . 6A ) ., Such isolated straight upwind sections also emerge in infotaxis trajectories , e . g . , the dark green path in Fig . 6B , even though not as a consequence of detections but rather related to exploration movements ., The behavioral trajectories , however , become less curvaceous towards the source; straightness occurs rather as a global instead of an isolated local feature ( Fig . 6C ) ., In terms of turning left and right while globally moving upwind , they resemble the cyan cognitive path in Fig . 6B ) although horizontal movements in behavioral paths decrease towards the source ., The latter feature , however , is also observable for the population of reactive za and ze trajectories ( Fig . 4B ) ., After having substantiated the existence of an Off firing phase in multiphasic MGC responses to pheromone stimulation we defined two basic reactive strategies ( Fig . 2 ) : a simple one composed of surge and spiral casting ( sp ) , and two more complex strategies including an additional Off zigzag phase following the surge sequence ( za , ze ) ., We applied these reactive strategies , as well as the cognitive infotaxis algorithm , in robotic experiments enabling our cyborg to locate a pheromone source ., Reactive searching with Off zigzagging yielded the shortest trajectories , independent of the pheromone dose ( Fig . 3 ) ., Infotaxis is less efficient but ensures slightly higher success rates , while reactive searching using only spiral casting was least efficient ., The effect of the pheromone dose on success rates and path lengths was not as clear as expected ., With respect to reactive strategies a higher dose led to shorter trajectories ( Fig . 3C ) , to smaller deviations from the optimal path ( Fig . 3F ) , and to more upwind surge ( Fig . 4C and 4D ) ., In terms of cognitive searching , however , the minimum dose yielded the shortest path — in spite of rather large deviations from the optimal path ( Fig . 3D and 3G ) ., Moreover , there was no effect of the dose on the track-angle histograms but the absolute number of turns increased with the dose ( Fig . 5C and 5B ) ., Cognitive strategies are based on complex algorithms that involve memory and learning ., Given the naive assumption of a linear relationship between costs in terms of complexity and profit , we expected infotaxis to be superior to reactive strategies — which is not what we found ., We now take a closer look at some factors characterizing our search task: the distance between starting position and source , the strength of the wind and the pheromone dose ., A higher dose induces more pheromone patches and thus augments the probability of pheromone detections ( cf . Fig . 5D ) ., Shorter distances to the source , as well as a stronger wind should have a similar effect ., More frequent clues , in turn , should facilitate the search task ., In general , infotaxis trials enabled many detections due to a slowly moving robot ( Methods , part 2 ) , i . e . , a longer time span to perceive pheromone patches ., This is particularly true at medium and maximum doses where the cyborg starts to detect very early and then advances by zigzagging along the centerline ( Fig . 5A , right ) , generating further detections ., Despite so many clues , the resulting trajectories are astonishingly long ., We therefore suspect that our experimental set-up with only 20 steps between agent and source is not very appropriate for cognitive searching ., It provides a rather simple task compared to the actual capabilities of infotaxis in computer simulations 6 , 30 , 41 , 42 with more than 100 steps between agent and source ., The basic difference between our real-word setup and infotaxis simulations is that we are confined to searching relatively close to the source because of experimental constraints whereas there are no such restrictions in computer simulations ., Indeed , in more dilute conditions ( lowest dose and ‘no pheromone’ condition ) , robot paths resemble typical infotactic trajectories previously observed in simulation ., With respect to our experimental conditions , reactive searching that includes Off zigzagging is obviously the optimal solution , in particular when combined with exponential spiraling ., It has been already suggested that such a two-phasic casting yields shorter trajectories than spiraling only 22 , but only for the maximum dose ( cf . Introduction ) ., Here , we demonstrate that this reduction is larger for more demanding tasks , and we investigate the dose dependency in terms of movement directions ( Fig . 4 ) ., The explanation is as follows ., The uncertainty about the location of the source is larger in the direction perpendicular to the wind than in the axis of the wind since the default search direction is upwind ., Hence , zigzag movements perpendicular to the wind are more efficient than isotropic spirals as the potential information gain is larger ., But what could be the origin of the Off in multiphasic MGC neuron firing ?, We hypothesize that it is an effect of the long tail of the pheromone response of ORNs 9 , 38 , 39 ., A high number of ORNs connect to fewer ( projection ) neurons in the MGC 9 , 10 ., ORNs respond to a brief pheromone pulse with a rather steep increase in firing which then slowly decays back to baseline spiking 39 ., The duration of this decay exceeds the end of the inhibitory phase of MGC neurons ., Moreover , the slightly increased reliability during the Off ( ) could also be explained by this hypothesis: there is still enough parallel ORN input to induce reliable spiking over different trials , but it is not enough synchrony to obtain precise spike timing ( ) ., Stimulus On and Off responses are typically reported for distinct neurons ., A famous example is the On and Off cells in the visual system of vertebrates , e . g . , in the cats visual cortex 44 , that enhance contrast information ., The term “Off cell” typically refers to a neuron that primarily responds to a reduction in stimulation strength or to the end of a stimulation period ., With respect to insect olfaction , separated On and Off ORNs in the cockroach antenna have been reported to encode opposite changes in the concentration of fruit odors45 , 46 ., The Off response of certain neurons in the silkmoths antennal lobe even contains information about the odors identity 47 ., Another example , related to behavioral switches , are two distinct neuronal populations in basal amygdala of mice that signal ‘fear on’ and ‘fear off’ , respectively , initiating the appropriate behaviors 48 ., In this article , however , On and Off originate from the same neuron , but emerge one after another ., As usual , the On encodes the stimulus onset , i . e . , a pheromone detection , while the Off signals that there has been no subsequent detection ., If there are several subsequent detections ( pulsation ) , there are also several subsequent ( independent ) On responses , each followed by an inhibitory phase 22 , 39 , 49 ., Thus , in agreement with the conventions described above , our Off encodes the loss of a stimulus — while additionally carrying some timing information ., We propose to consider Off zigzagging as a form of short-term memory ., This concept is in good agreement with our hypothesis on its origin , the long-lasting ORN responses ., Hence , the Off phase indicates that there has been a recent pheromone detection which has just been lost ., In this case a behavioral switch to crosswind zigzagging is appropriate because the agent is probably inside the plume and just needs to locate the centerline 37 , 50 ., If , however , the scent has been lost a longer time ago ( 30s in our experiments ) , the corresponding movements should include a downwind component in case the agent already passed the source ., We here assumed spiraling 22 , 43 but looping 14 , 15 or zigzagging with an angle ° 2 , 13 would also be adequate ., In this respect , reactive searching composed of surge and two-phasic casting establishes a kind of compromise strategy: predefined movements that include memory about the timing of the last detection ., According to our results , such an approach seems to be the best choice for searching inside or close to the pheromone plume , i . e . , if the agent is located downwind not too far from the source ., Obviously , our study cannot answer the question which search strategy moths actually use ., However , we would like to stress that a moth using infotaxis would require to develop a cognitive map capturing information about all previous detections and their spatial positions with respect to the agents locations ., Instead , we provide some evidence on how efficient , adequate , and realistic various strategies are compared to behavioral data ., Behavioral trajectories are shorter than all robotic paths — except for reactive zigzagging with maximum stimulation ( Fig . 6E ) ., Thus , under the given experimental conditions , insect behavior is generally more efficient than infotaxis and also than reactive searching ., The problem with most biological data is that timing and positions of odor detections are not known ., In any case , close to the source there are many detections ., Then , infotaxis is particularly inefficient since it yields too many sharp turns , the trajectories being dissimilar to behavioral ones ., The latter are not necessarily completely straight as assumed for our surge movements but exhibit a few turns of small curvature 36 ., This is probably the reason for the superiority of zigzagging with maximum stimulation that yields very straight trajectories ( pink in Fig . 6A ) ., If there are only a very few detections , the exploration term of infotaxis yields rather straight upwind paths with some embedded loops ( dark green in Fig . 6B ) ., In contrast , behavioral paths show mostly zigzagging ( with sharp turns and large lateral amplitudes ) , as well as some spiraling which is by far not as regular and long-lasting as assumed for our reactive strategies ., Therefore , as alternative approach for reactive searching , we propose to prolong zigzagging and to introduce a dependency of both the lateral displacement and the turning angle on the recent number of perceived detections: the more recent detections the smaller the angle and the step size ., Moreover , zigzag sequences are not only potentially related to multiphasic MGC responses , they have also been suggested to be linked to the so-called flip-flop activity of descending neurons in the silkworm 17 , 51 ., The big advantages of reactive searching are its proximity and adaptability to real-world biological data , as well as its simplicity in terms of computational requirements ., Nevertheless , we speculate that cognitive searching , whether used in nature or not , is more appropriate if the agents starting position is far outside the odor plume ( dilute condition ) ., Neurons from the macroglomerular complex ( MGC ) were recorded from male Agrotis ipsilon Hufnagel during pheromone stimulation of the antennae ., The pheromone stimulus is a blend of three components ( ratio 4∶1∶4 ) : ( Z ) -7-dodecenyl acetate , ( Z ) -9-tetradecenyl acetate and ( Z ) -11-hexadecenyl acetate ., The stimulation lasted 200 ms . There are 4 to 10 trials for each neuron ., Extracellular recordings were performed by inserting two glass electrodes filled with Tucson ringer into the MGC ., After amplification the signal was band-pass filtered ( 0 . 3 to 5 kHz ) and sampled at 16 kHz ., Spike sorting ( R-package SpikeOMatic ) yielded single neuron signals ., For more details see 38 , 39 ., We here analyzed the responses of 8 neurons that exhibited clear multiphasic responses ( Fig . 2A and 2B ) : an excitatory On peak in the Peri-Stimulus-Time-Histogram ( PSTH , the average firing rate over time ) , followed by an inhibitory phase and finally a more or less pronounced tonic excitatory Off phase 9 , 22 , 39 ., For analysis , the recordings were subdivided into the following time intervals: baseline Bl , On , Off1 , Off2 Off5 ( Off6 for neuron 1 ) ., The separation between Bl and On onset was based on a segmentation algorithm described 22 , Off1 starts directly after the inhibitory phase ., The Off interval length ( 2 . 3 or 2 . 5 s ) was chosen in a way that smooth changes during the Off are detectable and otherwise as large as possible ., For comparison , we also investigated 6 monophasic neurons ( separated in six time intervals ) , i . e . , neurons that showed simply an On ( Fig . 2C ) ., The Off occurrence is usually based on an increased firing compared to baseline activity 22 , 39 ., For a better characterization in terms of separating between Off and Bl , we calculated the following measures , see Fig . 2B and 2C: ( ) The coefficient of variation was computed by dividing the standard deviation of the inter-spike-interval ( ISI ) distribution by its mean ., It is =\u200a0 for regularly spiking neurons and =\u200a1 for irregular Poissonian spiking ., To be relatively independent of slow variations in the firing rate we used a local version , i . e . , considering only two adjacent ISIs at a time 52 ., ( ) Typically the correlation coefficient characterizes the synchrony in neuronal firing ( pairwise calculation ) : =\u200a1 if two neurons fire synchronously , =\u200a0 if firing is completely asynchronous ., Since we consider several trials of one neuron ( instead of synchronous trials of different neurons ) , here characterizes the spike time precision from trial to trial ( cf . 22 ) ., The bin size was 0 . 05 ms for all pairs ., ( ) The Fano Factor is calculated from the population activity , i . e . , the variance of the firing rate divided by its mean ., 1 indicates unreliable neuronal firing , =\u200a0 means reliable firing ., The bin size was 0 . 125 ms . This analysis was done in Matlab ., Tethered moths A . ipsilon were mounted on a Khepera III robot ( K-Team , Vallorbe , Switzerland ) , see Fig . 1B ., The insect body was immobilized inside a styrofoam block while the head was free in order to record the electroantennogram ( EAG ) 22 , see Fig . 1 , zoom 1 ., For electrical contact , the last 2–4 segments of one antenna were cut off and inserted into a glass pipette ( Fig . 1 , zoom 2 ) clamped by a micromanipulator and filled with ( in mM ) 6 . 4 KCl , 340 glucose , 10 Hepes , 12 MgCl2 , 1 CaCl2 , 12 NaCl ., A silver wire inside the glass pipette served as recording electrode while another wire , the reference electrode , was inserted into the neck ., The sensor was approximately 16 cm above the ground ., An EAG acquisition board was embedded on the robot ., The EAG signal was transmitted wireless via WIFI to a remote computer in order to be used as input for a MGC neuron model ., This neuron simulation was performed in real-time ( time steps =\u200a0 . 01 ms ) using SIRENE , a C-based neural simulator ( http://sirene . gforge . inria . fr ) ., Neuron simulation , pheromone detection and robot control were performed in separate threads ., A graphical user interface ( written in Qt/C++ ) visualized both EAG input and neuron output ., For more details see 22 , 53 ., Our cyborg ( i . e . , the robot using the antenna of a tethered moth as pheromone sensor ) had to locate the pheromone source in an arena of 4 m length and 2 . 5 m width , see Fig . 1A ., The whole set-up was placed in a Faraday cage ( height 1 m ) which was open to the upwind side ., We assumed that the source was found whenever the cyborg entered a disk of 20 cm radius centered at the source ., The cyborg always started at ( x , y ) = ( 0 , 0 ) m , the pheromone source was at ( 0 , 2 ) m ., A fan was placed at ( 0 , 7 ) m providing a relatively constant wind in -y direction with an average velocity of 0 . 880 . 3 m/s ( measured at the source location , 23 cm above the ground , i . e . , the height of the center of rotation , with a hot wire anemometer Testo 425 ) ., The mean wind velocity was the same in all experiments and it was given to the robot as a fixed parameter ., The airflow was rather turbulent than laminar ., Additional wind velocity measurements one and two meters downwind from the source ( on the centerline and on its left and right side ) typically yielded a standard deviation between 20% and 30% of the corresponding mean values ., We also estimated the Reynolds number to be , indicating turbulence ., The source was a filter paper strip ( approximately 5 cm long and 1 . 5 cm wide ) with 10 µl of pheromone solution dropped on its tip , located approximately 16 cm above the ground ., We used three different doses: minimum =\u200a0 . 1 µg/µl , medium =\u200a0 . 3 µg/µl and maximum =\u200a1 µg/µl of main pheromone component ( ( Z ) -7-dodecenyl acetate ) ., The plume contour ( indicated by black dashed lines in Fig . 4A , 4B and 5A ) is defined as the parabolic region where 90% of all pheromone detections occurred ., For trials that lasted longer than 3 min , the filter paper was replaced every trial , otherwise , we used one filter paper for two consecutive trials ., The cyborg trajectories were recorded using path integration provided by the odometry tracking module of the Khepera III Toolbox ( http://en . wikibooks . org/wiki/Category:Khepera_III_Toolbox ) ., We employed three r

Introduction, Results, Discussion, Methods, Analysis of the search trajectories

Male moths aiming to locate pheromone-releasing females rely on stimulus-adapted search maneuvers complicated by a discontinuous distribution of pheromone patches ., They alternate sequences of upwind surge when perceiving the pheromone and cross- or downwind casting when the odor is lost ., We compare four search strategies: three reactive versus one cognitive ., The former consist of pre-programmed movement sequences triggered by pheromone detections while the latter uses Bayesian inference to build spatial probability maps ., Based on the analysis of triphasic responses of antennal lobe neurons ( On , inhibition , Off ) , we propose three reactive strategies ., One combines upwind surge ( representing the On response to a pheromone detection ) and spiral casting , only ., The other two additionally include crosswind ( zigzag ) casting representing the Off phase ., As cognitive strategy we use the infotaxis algorithm which was developed for searching in a turbulent medium ., Detection events in the electroantennogram of a moth attached to a robot indirectly control this cyborg , depending on the strategy in use ., The recorded trajectories are analyzed with regard to success rates , efficiency , and other features ., In addition , we qualitatively compare our robotic trajectories to behavioral search paths ., Reactive searching is more efficient ( yielding shorter trajectories ) for higher pheromone doses whereas cognitive searching works better for lower doses ., With respect to our experimental conditions ( 2 m from starting position to pheromone source ) , reactive searching with crosswind zigzag yields the shortest trajectories ( for comparable success rates ) ., Assuming that the neuronal Off response represents a short-term memory , zigzagging is an efficient movement to relocate a recently lost pheromone plume ., Accordingly , such reactive strategies offer an interesting alternative to complex cognitive searching .

The moth mating race is a suitable model case for studying the efficiency of various search strategies and to compare them to real-world behavior ., All there is to guide olfactory navigation are simple sporadic clues , i . e . , single pheromone detections ., Thus , a pheromone seeking male relies on a specifically adapted behavior where action selection is triggered by simple perceptional events ., They switch between stereotypical movement sequences , as , for example , upwind surge and crosswind casting ., This behavior can be either a consequence of cognitive processing or a reactive reflex of fixed action patterns ., Suggesting a direct relationship between neuronal central activity and such action patterns , we combine and implement them as reactive strategies ., We also employ infotaxis , an artificial intelligence algorithm specifically developed for searching in turbulent odor plumes ., Using these strategies in cyborg experiments , we obtain and compare the resulting search trajectories ., Our results indicate that complex , computationally expensive search strategies like infotaxis are not necessarily better than simple reactive ones ., With respect to our set-up , reactive searching yields the shortest trajectories if and only if it includes a crosswind zigzagging phase that represents a short-term memory ., Thus , already a minimal bit of simplistic memory can produce very efficient goal-directed behavior .

systems biology, machine learning, behavioral neuroscience, computer and information sciences, cognition, artificial intelligence, computational neuroscience, single neuron function, olfactory system, biology and life sciences, sensory systems, computational biology, computerized simulations, cognitive science, neuroscience, animal cognition, coding mechanisms

journal.pcbi.1004257

Identifying Driver Genomic Alterations in Cancers by Searching Minimum-Weight, Mutually Exclusive Sets

Somatic genome alterations ( SGAs ) such as somatic mutations , somatic copy number alterations and epigenomic alterations are major causes of cancers1–3 ., In general , SGAs in a tumor can be divided into two types: those that affect cellular signaling proteins , perturb the cellular signaling system , and eventually contribute to cancer initiation and progression are called driver SGAs; and those that do not directly contribute to cancer development are designated as passenger SGAs ., A fundamental problem of cancer-genome research is to identify signaling pathways that , when perturbed by driver SGAs , lead to cancer development or affect clinical outcomes for patients ., Identification of such pathways will not only advance our understanding of the disease mechanisms underlying cancer , but will also provide guidance for the precision treatment of cancer patients ., In a cell , signaling pathways detect and transmit cellular signals to maintain cellular homeostasis; often , such signals eventually regulate the transcription of genes in order to initiate certain biological processes ., For example , the signal transmitted by a growth factor usually leads to the transcription of genes involved in cell proliferation ( Fig 1 ) ., As such , the impact of an SGA affecting a signaling protein in a tumor often manifests as an expression signature embedded in the expression profile of the tumor ., For example , a mutation leading to constitutive activation of the epidermal growth factor receptor ( EGFR ) gene may lead to over-expression of its target genes ., Thus , the gene expression profile of a cell at a given time reflects the state of its cellular signaling system , although it is a convoluted response to all active signals ., Inferring the state of an individual pathway requires the deconvolution of the signals embedded in its gene expression data ., The Cancer Genome Atlas ( TCGA ) has collected the most comprehensive genome-scale data to date , including somatic mutations , copy number variations , and gene expression from a large number of different types of cancers ., By simultaneously capturing SGAs and gene expression data from each tumor , the TCGA data reflect the cause and readout of perturbed signaling pathways , thus providing a unique opportunity for studying cancer pathways ., Identification of perturbed and critical signaling pathways using the TCGA data is challenging in the following ways ., First , as a cancer cell usually hosts dozens to hundreds of SGAs , we need to identify the small number of driver SGAs among the large number of passenger SGAs within a given tumor ., Current approaches for identifying driver SGAs mainly concentrate on those that occur beyond random chance in a patient population4–7 ., These approaches would fail to find low-prevalence SGAs that affect a specific cancer-driving pathway ., The second challenge is caused by the heterogeneity of mutations in tumor cells , in that few tumors have identical SGA patterns ., One reason for this phenomenon is that a signal pathway can be perturbed in multiple locations8 , and different tumors may share a common aberrant pathway but exhibit SGAs affecting distinct proteins on the pathway ., Thus , it is a challenge to determine if distinct SGA events in different tumors affect a common pathway ., Finally , a cancer results from perturbations in multiple pathways2 , and distinct combinations of pathway perturbation underlie the heterogeneity of cancers2 in terms of clinical phenotype ., Thus , it is a challenge to determine , among multiple SGAs and multiple aberrant pathways in a tumor , which SGA affects which pathway ., Researchers have developed various approaches to search for driving pathways using TCGA by exploiting different properties of tumor cells9–11 , including mutual exclusivity12–14 , which is the observed phenomenon that SGA events affecting the proteins within a signaling pathway seldom co-occur in a tumor ., A natural explanation for this phenomenon is that , if one mutation is sufficient to perturb the signal of a pathway and leads to the development of cancer , perturbation of other proteins is not required , and therefore co-occurrence of perturbations is seldom observed ., This property is observed in different types of cancer cells and pathways4 , 13 , 15 , 16 ., While it is the case that SGAs affecting the genes within a common pathway tend to be mutually exclusive , the reverse may not necessarily be the case , that is , finding a set of mutually exclusive SGAs does not ensures that their corresponding proteins are in the same pathway ., Current mutual-exclusivity-based methods12–14 , 16 , 17 concentrate on finding a set of SGAs of size k , such that the set covers as many tumors as possible while minimizing overlapping cover ( thus maximizing mutual exclusivity ) ., As the numbers of tumors and SGAs examined by contemporary genome technology increase , it becomes increasingly easier to find a set of unrelated SGAs that covers a certain number of tumors while exhibiting mutually exclusivity due to the heterogeneity of tumors ., To address this shortcoming , Zhao et al14 further considered the co-expression of SGA-affected genes in order to enhance the search of pathways ., The intuition underlying their approach is that proteins within a signaling pathway tend to be co-expressed , so the correlation of the expressions of the candidate genes of a pathway can be used as another objective function to guide pathway search ., Both mutual exclusivity and co-expression of SGAs are auxiliary properties of a signaling pathway , but they are not sufficient to indicate whether the SGAs affect a common cellular signal; therefore , they are useful but not the optimal objective function to guide the search for a signaling pathway ., In this study , we propose a novel signal-oriented framework for searching cancer pathways by combining gene expression with SGA data ., The premise underlying our approach is as follows: since the state of a signaling pathway can be reflected by the expression state of a set of genes it regulates ( i . e . , its signature ) , the task of searching for a pathway can be formulated as a search for a set of SGAs that collectively exhibit strong information with the state of a gene expression signature ., Under such a setting , mutual exclusivity and co-expression of SGA-affected genes can further be used as auxiliary objective functions to constrain the search space and to enhance the confidence of the results ., This approach addresses the fundamental task of the pathway discovery—finding a set of SGAs perturbing a common signal ., We applied this novel framework to the ovarian cancer and glioblastoma data from TCGA , followed by systematically evaluating the impact of the signal-oriented approach on the search for driving pathways and comparing the performance of our exact algorithm with that of heuristic or stochastic algorithms ., We show that the signal-oriented approach provides a general framework in which different pathway-searching algorithms can be combined with different signal-oriented objective functions ( beyond those discussed here ) to explore new directions for studying cellular signaling systems ., Data on somatic mutation , copy number alteration , and gene expression from 568 ovarian cancer tumors and 513 glioblastoma tumors , as well as 8 normal control samples from ovarian tissue and 10 cases of normal brain samples were downloaded from the TCGA4 , 5 ., For each tumor , we considered a gene as being differentially expressed if its expression value increased or decreased at least 3-fold in comparison to the median value of the gene in the control samples ., We defined a gene as affected by an SGA event if it had a non-synonymous single nucleotide variation in its coding region , and/or an insertion or a deletion; we also labeled a gene as affected by an SGA event if it had copy number alteration ( with the GISTIC22 score ≥ 2 and z-score ≥ 1 . 64 or GISTIC score ≤ -2 and z-score ≤ -1 . 64 , where the z-score is obtained by the z-transformation of the expressions of the gene across all tumors ) ., Hence , only the copy number alterations that affected gene expression ( with a p-value of 0 . 05 ) are included ., We removed the genes that exhibit both amplifications and deletions ( with the smaller fraction being over 10% ) in a given set of tumors that are supposed to have a common signal perturbed as inconsistent genes ., For example , if a gene X is affected by copy alteration , where 85% of events are amplification and 15% are deletion ( thus smaller fraction is over 10% ) , we would remove this gene from the consideration ., This is a relatively conservative consideration of those genes with consistent copy number alteration direction as potential drivers ., Finally , if two genes exhibit perfect correlation ( co-amplified or co-deleted whenever altered ) , we treat these genes as one common genomic alteration ., These procedures lead to a tumor-by-gene binary matrix recording differentially expressed genes , and a tumor-by-gene binary matrix recording SGA events in tumors ., To deconvolute signals embedded in the gene expression data , we hypothesized that if a set of genes performs coherently related functions and tends to be co-regulated in multiple tumors , the genes are likely regulated by a common signaling pathway as a module ., To find such modules among the cancer tumors , we employed a knowledge-driven data mining approach , developed in our previous studies19–21 , which consists of two major procedures:, 1 ) identifying functionally coherent gene subsets among the differentially expressed genes in each tumor , such that each gene subset is annotated by a GO term that summarizes the function of the genes; and, 2 ) further identifying the gene subsets that are differentially expressed in multiple tumors , which is formulated as the dense bipartite subgraph finding problem ., Researchers have used set cover23 , 24 or the extension of the set cover – module cover 25 model to find gene subsets that are differentially expressed because of pathway perturbation ., Our two-step method differs from these in that genes included in each solution subset are both functionally coherent and co-expressed in a considerable number of tumors ., We also obtained the tumors that perturb the pathways regulating the expression of each gene subset ., To identify functionally coherent gene subsets , we first found a tumor’s differentially expressed genes and grouped them into non-disjoint subsets by mining their annotations 19–21 ., This was achieved by representing the hierarchical structure of GO terms as a directed acyclic graph and searching for genes annotated with closely related GO terms ., We first associated genes to the GO terms according to annotations of the genes ., We then iteratively merged highly specific GO terms and their associated genes to their parent GO terms according to a procedure 19 that strives to minimize the loss of semantic information during the process ., In this fashion , we can group genes annotated with closely related terms into a set annotated with a more general GO term that retains the information of the original annotations ., We stop the procedure if a further merge leads to a non-coherent gene set , according to a quantitative metric that assesses the statistical significance of functional coherence of the gene set 19 , 21 ., This procedure enabled us to partition differentially expressed genes from each tumor into non-disjoint , functionally coherent subsets ., Next , we further identified the functionally coherent gene subsets that are affected in multiple tumors ., We modeled this problem as a dense bipartite subgraph ( DBSG ) finding problem , of which the detailed algorithm was introduced in our previous work21 ., Briefly , we pooled gene subsets annotated with a common GO concept across all tumors and constructed a bipartite graph , in which the vertices on one side represent the pool of differentially expressed genes sharing the GO annotation , and the vertices on the other side represent the tumors; an edge between a gene and a tumor indicates that the gene is differentially expressed in the tumor ., We then searched for a subset of genes that are co-differentially expressed in multiple tumors ., We formulated our task as follows: find a maximum dense bipartite subgraph such that each gene must be connected to at least 75% ( a connectivity ratio ) of all tumors in the subgraph and each tumor must be connected to at least 75% of all genes in the subgraph ., Thus , each DBSG consists of a set of genes , i . e . , an RM , and a set of tumors in which the RM is perturbed ., To search for the candidate members of a signaling pathway regulating an RM , we aimed to find a set of SGA-affected genes that has the following properties:, 1 ) the SGA events affecting the genes cover as many as possible of the tumors in which the RM of interest is perturbed;, 2 ) the SGA events carry strong information with respect to the expression state of an RM , or , in other words , the SGAs are significantly enriched in tumors in which RMs have been perturbed; and, 3 ) each tumor is covered by only one gene in the solution set ( thus mutually exclusive . Note: SGA events are only mutually exclusive among tumors in each DBSG ) ., By assigning a weight to each SGA to reflect the amount of information the SGA carries with respect to the state of the RM , we formulate this computational problem as the weighted mutually exclusive maximum set cover problem , a variant of the well-known set cover problem in algorithm theory26 ., To assess the strength of association of an SGA-affected gene ( a genome locus ) with the state of an RM , we apply a hypergeometric test to compute the enrichment of the SGA events of the gene in tumors within which the RM has been perturbed27 ., We then set the log p-value of the SGA enrichment analysis as the weight for the gene; thus , a set of SGAs with a smaller total weight tends to carry more information with respect to the RM when compared to another gene set with a greater total weight ., Fig 2C illustrates the problem setting as follows ., Of the 16 tumors and 5 genes in a dataset ( Fig 2C ) , 6 tumors are included in a DBSG ., We define that a tumor is covered by a gene if an SGA affecting the gene occurs in the tumor; we then represent each gene by the subset of tumors in the DBSG it covers ., In our example , g1 = {T1 , T2} , g2 = {T3 , T4 , T5} , g3 = {T1 , T4 , T5} , g4 = {T1 , T3} , and g5 = {T2 , T5 , T6} ., For each DBSG , we define the set of all tumors in the DBSG , X , as the ground set; for example , in the figure , X = {T1 , T2 , T3 , T4 , T5 , T6} ., We define, F, as the set of the candidate genes; in our example ,, F, = {g1 , g2 , g3 , g4 , g5} = {{T1 , T2} , {T3 , T4 , T5} , {T1 , T4 , T5} , {T1 , T3} , and {T2 , T5 , T6}} ., We define w: F→ ( -∞ , ∞ ) ; the function w gives weight to each gene ., Given a subset of genes ,, F’⊂F, , if no two elements in, F’, have any common element , i . e . , if no two genes cover the same tumor , we then say that, F’, is mutually exclusive; the weight of, F’, is, ∑S∈F’, w ( s ) ., The problem’s goal is to find a mutually exclusive subset of, F, that covers a maximum number of elements of X ( i . e . , that covers a maximum number of tumors ) ., If we find two or more solutions , e . g . , {g1 , g2} and {g4 , g5} , that cover the maximum number of elements , we choose the solution with the minimum weight , {g1 , g2} ., This is the formal definition of the weighted mutually exclusive maximum set cover problem ., As is the case with the formulations of other studies on mutual exclusivity12–14 , 16 , 17 , our problem is NP-hard ( see the proof in a separate technical report28 ) ., Previous studies used heuristic or stochastic algorithms 12–14 , 16 , 17 to handle the mutually exclusive set cover problem or its variants , but such algorithms do not guarantee the finding of optimal solutions ., In this study , we developed an exact algorithm , called the Weighted Mutually Exclusive Maximum Set Cover algorithm ( or the ME algorithm ) , that guarantees the finding of exact optimal solutions ., The algorithm uses parameterized techniques29 , such that the running time is an exponential function of a parameter that can be bounded by a small number for certain specific applications , instead of the exponential functions of large input sizes that are generally used for solving general NP-hard problems ., Because our problem involves about 600 tumors and 30 , 000 genes , an exponential function of any of these input sizes would be intractable ., However , using the parameterized technique , we developed an algorithm whose worst time complexity is O* ( 1 . 325m ) , where m , the parameter , is the number of candidate genes when we search driver SGAs affecting a pathway ., In fact , the actual running time of the algorithm is much smaller than that of the worst time complexity; the algorithm can finish our computation task on a workstation in 5 to 10 minutes , even when m is 200 , which is sufficiently large candidate size for searching driver SGAs that perturb a signaling pathway ., We refer to a set of solution SGAs as perturbation module ( PM ) to reflect that they may perturb a common signaling pathway ., Our method adopts a branch-and-bound principle: the algorithm first finds a subset in, F, , and then branches on it ., Due to the mutual exclusivity constraint , if any two subsets in, F, overlap , then at most only one of them can be chosen into the solution ., For example , suppose that the subset S intersects with other d subsets in, F, ; then , if S is included into the solution , S and other d subsets intersecting with S will be removed from the problem , and if S is excluded from the solution , S will be removed from the problem ., We continue this process until the resulting sub-problems can be solved in constant or polynomial time ., Let T ( m ) be the number of computations needed when call the algorithm with m subsets in, F, , then we can obtain the recurrence relation T ( m ) ≤ T ( m− ( d+1 ) ) +T ( m−1 ) ., As if d = 0 for all subsets in, F, , the problem can be solved in polynomial time ( all subsets in, F, will be included into the solution ) , in the recurrence relation , d ≥ 1 ., Therefore , we can obtain T ( m ) ≤ 1 . 619m , which means the problem can be solved in O* ( 1 . 619m ) time ( note: Given the recurrence relation T ( k ) ≤ ∑0 ≤ i ≤ k-1 ciT ( i ) such that all ci are nonnegative real numbers , ∑0 ≤ i ≤ k-1 ci > 0 , and T ( 0 ) represents the leaves , then T ( k ) ≤ rk , where r is the unique positive root of the characteristic equation tk- ∑0 ≤ i ≤ k-1 citi = 0 deduced from the recurrence relation30 ) ., We improved the algorithm’s running time by carefully selecting subsets in, F, for branching ., As the proof of the algorithm is very involved , we present the details in the technical report28 ., We have implemented all algorithms for the paper ., Supplement results and codes for algorithms can be found at: http://pitttransmed-tcga . dbmi . pitt . edu/mutuallyExclusive/ ., Using the integrated knowledge-mining and data-mining approaches , we identified 88 dense bipartite subgraphs ( DBSGs ) from the ovarian cancer tumors ., Each DBSG includes a response module ( RM ) consisting of at least 10 genes that are differentially expressed in 30 or more tumors ., Based on our functional coherence analysis , the genes in an RM were functionally related to each other ., To further corroborate these results , we also evaluated the RMs using the Ingenuity Pathway Analysis ( IPA— http://www . ingenuity . com/ ) ; each of our RMs was found to significantly overlap with at least one of the IPA networks ., For example , we found 55 RMs , of which more than 90% of their genes overlapped with at least one network from the Ingenuity network database ( results are presented in a supplementary website so that researchers can browse the RMs and their driver SGAs ) ., As an example , an RM that consists of 11 genes ( AURKA , CCNB1 , CHEK1 , COL5A1 , EPHB3 , NEK2 , PSRC1 , STMN1 , TACC3 , THBS2 , TWIST1 ) that are up-regulated in 62 tumors ., The biological processes in which the genes are involved are summarized by the GO term GO:0051128: Regulation of Cellular Component Organization ( note: because genes in each RM are annotated by a GO term , we use GO term IDs to name RMs and PMs; we also use U_ or D_ to indicate whether genes in the RM are up-regulated or down-regulated , respectively ) ., For example , the designation “RM U_GO0051128” means that the genes in the RM are up-regulated and that they are annotated by the GO term GO:0051128; “PM D_GO0009611” is the PM that regulates the RM D_GO0009611 ) ., We found that 10 of those 11 genes are in an IPA network ( Fig 3 ) that is labeled as “Cellular Assembly and Organization , Cellular Function and Maintenance , Cell Morphology” ., Previous laboratory studies show that 9 of these genes play important roles in tumor initiation and progression in different types of cancers ., For example , AURKA was found overexpressed in the early stage ovarian tumors , therefore suggesting that the alteration of AURKA could be an early event of ovarian cancer31 ., High levels of AURKA expression is closely correlated to poor survival of patients with ovarian cancer 32 ., The proliferation-related targets AURKA and CCNB1 were overexpressed in clinical ovarian tumor specimens33 ., Our predicted results corroborate with the established roles of AURKA and CCNB1 in cancers ., In addition , many references also show that overexpression of seven of the remaining nine genes in the RM are related to cancers ( S1 Table ) ., Another example of a cancer-related RM , annotated with the GO term GO:0010564 ( Regulation of Cell Cycle Process ) , includes 10 genes ( BIRC5 , CCNB2 , CDC7 , CDKN2A , CENPE , CENPF , CHEK1 , NEK2 , TIMELESS , UBE2C ) that are up-regulated in 140 tumors ., All of those 10 genes are in an IPA network related to “Cell Cycle , DNA Replication , Recombination , and Repair , Cellular Assembly and Organization” ( see Supplement ) ., A literature search shows that 9 of 10 genes in the module are related to several types of cancers ( S2 Table ) ., Among them , expression of CENPE and CCNB2 correlates with worse clinical outcomes of patients with breast or ovarian cancers 34–36 ., The fact that the genes in these RMs are functionally coherently related and co-regulated in multiple tumors from different types of tumors indicates that their aberrant expression is likely regulated by a common signal; thus , expression state of an RM can be utilized as the readout of the state of a hidden signal , allowing the search for the SGA events perturbing the signal ., One interesting observation in this RM is that the CDKN2A , a tumor suppressor , is overexpressed ., By checking the copy number alteration data , we found that the overexpression of CDKN2A in most of 140 tumors were not caused by the gene amplification ., The similar phenomenon was observed in large number of tumors in other tumor types , such as GBM and HNSC , where the CDKN2A was overexpressed in almost all tumors without CDKN2A amplifications ., The explanation of this phenomenon needs the further study from cancer biologists ., This framework was also applied to TCGA data of glioblastoma multiform ( GBM ) , the most malignant cancer in the brain , and we identified 101 RMs ., Comparing RMs in GBM with ones in ovarian cancer , 38 RM pairs were annotated by an identical GO term in both ovarian cancer and GBM , among which 17 modules are significantly overlapped ( p-value and q-value of overlapping < 10–4 , S3 Table ) ., For example , the RMs annotated with U_GO0007067 ( mitotic nuclear division ) found from GBM and ovarian cancer have 18 and 17 genes respectively , in which 15 genes are in common , and the union of the two RMs includes 20 genes ., As expected , literature studies indicate that almost all the above genes and most of other significant overlapping RMs are related to cancers ( S4 Table ) , including those involved in U_GO0009611 ( response to wounding ) and U_GO0006974 ( cellular response to DNA damage stimulus ) ., Thus the approach of searching RMs as reflections of perturbed cellular signals is generalizable to different types of cancers and capable of finding cancer-related RMs ., We further investigated if the expression states of RMs are relevant to patients’ clinical outcomes , we dichotomized GBM patients from TCGA according to the expression state of each RM , followed by survival analysis ., We found the expression states of 25 RMs are associated with significant differences in patients’ survival ( p-values < 0 . 05 and q-values < 0 . 05 for Kaplan-Meier analysis , see S5 Table and S1 Fig ) ., We also applied these methods to the breast cancers from TCGA for searching RMs ( data not shown ) ., We used the breast cancer RMs as features for predicting survival of the patients studied by Curtis et al 37 in an open research challenge ( the DREAM 7 Challenge ) , in which RMs were found to be highly predictive of patient survival 38 ., Therefore , the expression states of RMs reflect the states of cancer cells , which is determined by cellular signal transduction pathways ., We hypothesized that the differential expression of an RM in a tumor is due to the aberrant signal resulting from pathway perturbation by at least one of the SGAs ( somatic genome alterations ) observed in the tumor ., Since somatic copy number alterations are common in ovarian cancers ( which may contribute to differential expression of genes ) , we first examined if identified RMs are driven by copy number alterations ., For each RM , we treat differential expression of a gene in a tumor as a differential expression event ., We also define that it is driven by a copy number alteration event if the gene is copy number altered in the tumor ., We then calculated the fraction of copy-number-alteration-driven differential expression events for each module and averaged them across the RMs , which shows that , on average , only 3 . 4% of differential expression events are likely driven by copy number alteration ., The results support our hypothesis that the differential expression of an RM is driven by a pathway rather than by direct copy number alterations ., As such , a reasonable strategy for identifying the signaling pathway regulating the RM is to pool the tumors in which the RM is differentially expressed and further search for a subset of the SGA events in these tumors that carries the strongest information with respect to the expression state of the RM ., We refer to a module of genes affected by such SGAs as a “perturbation module” ( PM ) ., When given a DBSG , we first identified all SGA events observed in the tumors within it , and then calculated enrichment of SGAs affecting a gene using a hypergeometric test , assigning the enrichment p-value as the weight of the gene ., We applied the ME algorithm to identify an optimal PM for each DBSG , using up to 200 SGAs with the lowest p-values as input , a sufficiently large number when considering that most known biological pathways contain around tens of proteins ., The sizes of the returned PMs ranged from 3 to 14 genes , with an average of 7 . 14 ., Since our algorithm strives to include SGAs that are specifically enriched in the tumors in a DBSG , the genes in a PM as a whole are highly enriched in the tumors in a DBSG , with enrichment p-values ranging from 5 . 68×10−4 to 7 . 16×10−25 ( median: 8 . 97×10−14 ) ., Since SGA events are in effect randomized perturbatons performed by nature , a strong correlation between SGA events and the expression state of an RM suggests that SGAs influence gene expression rather than the reverse direction ( differential gene expression causing SGAs ) ., Thus , genes in each PM identified in our study are likely members of the signaling pathway perturbed in tumors that underlie the differential expression of the genes in an RM ., Though experimental validation of the results could be conclusive , it is extremely costly ., Therefore , we validated our results by comparing them to the existing knowledge using the IPA package , with the understanding that while the knowledge base of the IPA may be incomplete , it is , nonetheless , an accessible approach ., Our findings indicated that most PMs were significantly associated with different diseases and/or disorders ( 65 PMs with both p-values and q-values of at most 0 . 001 , with a median of 9 . 21×10−4 ) ; among them , 30 PMs were related to cancers with both p-values and q-values smaller than 0 . 001 ., We further investigated whether the identified PMs could be mapped to known signaling pathways , concluding that , indeed , many PMs were enriched in known pathways , including 51 PMs that were enriched in a known pathway with both p-values and q-values of at most 0 . 01 ., As an example , we examined the PM corresponding to the RM studied in the previous section , U_GO0051128 , which consists of 6 genes ( CCNE1 , RB1 , FRMD1 , COLIM4 , MAST3 , RNF139 ) ., The genes in the PM are enriched in the IPA pathway “Estrogen-mediated S-phase Entry” ( Fig 4A ) , with a p-value of 1 . 83×10–5 ., This PM has two genes ( RB1 and CCNE1 ) in the well-characterized RB1 cancer pathway that plays important roles in ovarian cancer tumorigenesis ( Fig 4B ) 5 ., Golgi integral membrane protein 4 ( GOLIM4 ) is a type II Golgi-resident protein that involves in processing proteins synthesized in ER and assist in the transport of protein cargo through the Golgi apparatus39 ., These transported proteins include ones shown in Fig 4A ., Ring finger protein 139 ( RNF139 ) is a multi-membrane spanning protein with ubiquitin ligase activity ., RNF139 interacts with tumor suppressor VHL and JAB140 , the latter is responsible for the degradation of tumor suppressor CDKB1B/p27CIP1 in this pathway ( Fig 4A ) ., GOLIM4 and RNF139 in this PM were found as potential cancer drivers in various types of cancers 41–43 ., Additionally , MAST3 , RB1 , and CCNE1 in this PM are critical in regulating cell cycle 44–46 ., As shown in Fig 4C , the mutually exclusive pattern of the SGAs affects genes in this PM identified from the 62 tumors in which this RM was perturbed ., It is of highly significance that our algorithm predicts that the amplification of CCNE1 gene conveys the identical information as to the mutation or deletion of RB1 ., As shown in Fig 4B , the protein encoded by CCNE1 inhibits that of RB1 ( Fig 4B ) ; both amplification of CCNE1 and mutation/deletion of RB1 have the same effect on a common signal , leading to aberrant regulation of cell cycle entry and thereby causing cancers ., Indeed , 6 out of 11 genes in the RM U_GO0051128 are related to cell cycle ., When searching the PASTAA ( http://trap . molgen . mpg . de/PASTAA . htm ) database for enriched transcription factor binding sites of genes in this RM , the binding site of E2F1 ( a transcription factor directly downstream of RB1 ) was the most significantly enriched region in the promoters of the genes in this RM ( S6 Table ) ., Thus , our algorithm correctly identified a perturbed signaling pathway and its downstream target genes ., Fig 4D further shows that proteins encoded by the genes in the PM U_GO0051128 directly interact with other well-known oncogenes , such as TP53 , MDM4 , CCND1 , CCND2 , MYC , E2F3 , E2F5 , BRCA2 , PTEN , MET , and COPS5 ( S7 Table ) ., Thus , the results indicate that the signal-oriented approach leads to biologically sensible findings ., However , a challenging issue of handling copy number alteration is that a set of genes can be co-amplified ( co-deleted ) within highly overlapping but not perfectly identical copy number alteration fragments in different tumors ., Thus , it would be difficult to differentiate the signals of such alterations ., For example , RNF139 , TRMT12 , ZNF572 , SQLE , MYC and other genes are often co-amplified but not perfectly correlated

Introduction, Materials and Methods, Results, Discussion

An important goal of cancer genomic research is to identify the driving pathways underlying disease mechanisms and the heterogeneity of cancers ., It is well known that somatic genome alterations ( SGAs ) affecting the genes that encode the proteins within a common signaling pathway exhibit mutual exclusivity , in which these SGAs usually do not co-occur in a tumor ., With some success , this characteristic has been utilized as an objective function to guide the search for driver mutations within a pathway ., However , mutual exclusivity alone is not sufficient to indicate that genes affected by such SGAs are in common pathways ., Here , we propose a novel , signal-oriented framework for identifying driver SGAs ., First , we identify the perturbed cellular signals by mining the gene expression data ., Next , we search for a set of SGA events that carries strong information with respect to such perturbed signals while exhibiting mutual exclusivity ., Finally , we design and implement an efficient exact algorithm to solve an NP-hard problem encountered in our approach ., We apply this framework to the ovarian and glioblastoma tumor data available at the TCGA database , and perform systematic evaluations ., Our results indicate that the signal-oriented approach enhances the ability to find informative sets of driver SGAs that likely constitute signaling pathways .

An important goal of studying cancer genomics is to identify critical pathways that , when perturbed by somatic genomic alterations ( SGAs ) such as somatic mutations , copy number alterations and epigenomic alterations , cause cancers and underlie different clinical phenotypes ., In this study , we present a framework for discovering perturbed signaling pathways in cancers by integrating genome alteration data and transcriptomic data from the Cancer Genome Atlas ( TCGA ) project ., Since gene expression in a cell is regulated by cellular signaling systems , we used transcriptomic changes to reveal perturbed cellular signals in each tumor ., We then combined the genomic alteration data to search for SGA events across multiple tumors that affected a common signal , thus identifying the candidate members of cancer pathways ., Our results demonstrate the advantage of the signal-oriented pathway approach over previous methods .

journal.pgen.1004300

Heterogeneity in the Frequency and Characteristics of Homologous Recombination in Pneumococcal Evolution

The evolution of many bacterial species is largely driven by horizontal exchange of sequence ., Often , this can be attributed to the movement of autonomously mobile genetic elements ( MGEs ) ., Many of those are able to insert into the host chromosome through site-specific recombination mediated by an integrase ., However , in ‘naturally’ transformable species that possess a competence system , exogenous DNA can be imported from the environment and integrated into the chromosome through homologous recombination ( HR ) ., This process was first discovered in Streptococcus pneumoniae ( the pneumococcus ) , representing some of the earliest work on molecular genetics 1 ., Initially , recombination was considered by many microbiologists to be interesting but rare ., However , later population-based studies demonstrated that it can have a quantifiable impact on population genetic structure of many bacteria , including S . pneumoniae 2–4 ., Additionally , as this mechanism only requires that the acquired DNA is homologous at the ends , recombination allows for the cassette-like transfer of highly variable genes , such as those that encode for the pneumococcal capsule 5 , 6 , in a process originally defined as ‘homology-directed illegitimate recombination’ 7 ., This has important clinical consequences , as this exchange of sequence has played a crucial role in the development of pneumococcal antibiotic resistance 8 , as well as the ‘switching’ of capsule types that can result in vaccine escape 9 , 10 ., The rate at which the recombination process occurs is of importance when considering the adaptation of the bacterium to clinical interventions ., The simplest null expectation is that HR is a homogeneous process across the species ., However , recent findings suggest that homogeneity of recombination is unlikely to capture the dynamics of horizontal sequence exchange in pneumococci ., In particular , heterogeneity has been observed in the rates at which different genotypes accumulate sequence diversity through HR ., Analysis of multilocus sequence typing data identified a subset of ‘hyper-recombinant’ pneumococci that were more likely to be resistant to a number of antibiotics 11; similarly , comparison of lineages within a single population found significant variation in the observed rate of HR 12 ., Second , in vitro work has found that the frequency of recombination events occurring across the genome in isogenic recipient bacteria varies with the concentration of donor DNA , suggesting the environment is likely to influence the process of sequence transfer 13 ., Similarly , extensive exchanges between pneumococci over short time periods have also been observed in clinical isolates , sometimes with important phenotypic consequences 14–16 ., Third , variation has been observed in the rate at which pneumococci undergo transformation in experimental systems 17 , 18 ., Therefore more detailed quantification of the observed contribution of HR will be invaluable in defining and understanding the behaviour of distinct lineages under different conditions ., This in turn should help us understand how recombination contributes to the overall rate of diversification , and how it drives adaptive changes in pneumococcal populations ., The opportunity for such an analysis is presented by the recent whole genome sequencing of two international collections representing contrasting pneumococcal genotypes ., The first is a set of 241 pneumococcal genomes of the recently emerged pandemic multidrug resistant lineage , PMEN1 19 ., This lineage appears to have originated in Europe in 1970s , and in the following decades spread rapidly across the world ., The ancestral serotype of this lineage , serotype 23F , has switched to new capsules by HR which have resulted in its evasion of the 7-valent vaccine introduced in the early 2000s ., The second lineage is a set of serotype 3 isolates belonging to clonal complex 180 ( CC180 ) 20 ., Serotype 3 , which causes disease associated with high levels of mortality , has been recently included in the expanded 13-valent conjugate vaccine formulation ., The CC180 lineage appears to be older than PMEN1 , yet there is little evidence of it having undergone homologous recombination in recent decades , with the consequence that it is generally susceptible to antibiotics and has not altered its serotype ., Hence these two genotypes , PMEN1 and CC180 , are highly distinct both in terms of their phenotypes and evolutionary dynamics ., This work describes the fitting of different mathematical models of sequence exchange to the HR identified in the PMEN1 and CC180 datasets in order to identify and characterise and heterogeneity evident in the process ., This resulted in the identification of two different classes of HR in both lineages: micro-recombination and macro-recombination ., Potential underlying mechanistic explanations for this observation , and the implications for bacterial evolution , are discussed ., The analysis presented here is based on the inference of individual HR events , as previously described by Croucher et al . 19 ., Briefly , this approach identifies independent HR events as clusters of SNPs in a genealogy reconstructed from whole genome alignments ., Removal of those events allows to establish a clonal tree based on vertical transmission of SNPs ., The inference for the PMEN1 lineage was based on an alignment of sequences , resulting in a genealogy with branches and homologous recombinations , whereas the inference for the CC180 lineage was based on an alignment of sequences , resulting in a genealogy with branches and homologous recombinations ., Let label the branches , and let be the number of HR events assigned to branch , such that ., For a given branch , let label the recombination events , and let be the length of genetic tract , in DNA base pairs , replaced by the HR event ., We define the recombination rates in our models as rates per unit of branch length ., Thus , their interpretation depends on the chosen measure of branch length ., Since our model structure is generic with respect to this choice , by default the branch length is measured by years estimated using a dated genealogy based on a relaxed molecular clocked estimated using Bayesian methods ., ( The results for alternative branch lengths are given in Tables 4–5 , Figures 8–9 and Text S2 . ), We thus use a statistical modelling approach to explain the number and size of HR events on a branch of length given the genealogy of a lineage ., We use a modelling approach to test whether recombination in S . pneumoniae is heterogeneous with regard to its rate or length distribution ., Four models were devised to account for patterns observed in the data:, ( i ) recombination is homogeneous in frequency and in size ( Model 1 ) ;, ( ii ) recombination is heterogeneous in frequency or in size , with heterogeneity modelled as deviation from the null model 1 ( Model 2 ) ;, ( iii ) recombination is heterogeneous in frequency and size , and is modelled by two independent and homogeneous processes of recombination with different frequency and size: micro-recombination and macro-recombination ( Model 3 ) ; and, ( iv ) recombination is heterogeneous in frequency and size , as in model 3 , but the heterogeneity in frequency is independent from the heterogeneity in size ( Model 4 ) ., The models were fitted by the maximum likelihood method , namely maximising the log-likelihood function given in Text S1 ., This was done using optimization functions NMaximize or FindMaxiumum in Mathematica 8 . 0 ., The comparison between four different models was performed using the Akaikes Information Criterion , adjusted for finite degrees of freedom ( AICc ) ., We considered one model to be a better fit than another when the difference in AICc was less than 10 ( ) ., The best model was chosen as the one with the lowest value of ., If multiple models were the best fit to the data , the model with the smallest number of parameters was chosen as the best by the rule of maximum parsimony ., Goodness of fit was determined by verifying the ability of the model to replicate the data under re-simulation ., To that end , marginal distributions of frequency and size of the simulations were compared to the equivalent marginal distributions of the data ( see Results ) ., The details of the simulations are described in Text S3 ., In brief , an ancestral sequence of S . pneumoniae was chosen as the earliest isolate of PMEN1 known 19 , 22 ., A forward , discrete-time simulation was designed to simulate the evolution of the lineage , including diversification through recombination simulated through incorporating homologous sequence from other publically available pneumococcal genomes ., We assumed that at every time step the sequence acquired a single base substitution , and could diversify into two independently diversifying lineages with a constant probability ., Each sequence also had a probability of being sampled at each timestep , after which it stopped evolving ., The simulation was stopped when the population reached a maximal number of sequences , ., At each timestep , recombination occurred as prespecified by one of the four models: A , B , C or D . In Model A , recombination occurred homogeneously across the genome , with lengths of recombinations following a geometric distribution ., In Model B , heterogeneity ( micro/macro-recombination ) was introduced in frequency but not the size ., In Model C , heterogeneities in both frequency and size were correlated , as described in Model 3 above ., In Model D , heterogeneity was also introduced in both frequency and size but the two were treated as independent variables for each recombination ., Each model was run three times , giving 12 simulations overall ., To study the process of HR in the evolutionary history of the two lineages , PMEN1 and CC180 , we fitted mathematical models which describe how recombination events are distributed along the branches of the evolutionary tree of each lineage of S . pneumoniae ., The procedure of model fitting is described in detail in Text S1 ., The phylogenies of both lineages have been constructed as described previously in 19 , 20 based on vertically inherited point mutations , and were shown to be highly consistent with a molecular clock ., Recombination events were reconstructed such that they were associated with particular branches of the phylogeny 19 ., To remove events that may have been introduced through the movement of MGEs in PMEN1 , rather than being mediated by HR , any events affecting the prophage remnant , prophage MM1-2008 or ICE Sp23FST81 were not considered in this analysis 22 ., Likewise , for CC180 , these MGEs included the OXC141 prophage locus and a single putative integrative and conjugative element ( ICE ) 20 ., The distribution of recombination events on the phylogenetic trees of both lineages is summarised in Fig . 2 ., The simplest model considered is that recombination events occur as a homogeneous point Poisson process through time with rate , so that the number of events occurring on a genealogical branch of length is Poisson distributed with mean , and that event sizes are geometrically distributed , with the mean length of genetic tract replaced by recombination for each event being base pairs of DNA ( see Fig . 1 and Methods ) ., This model failed to capture clear heterogeneities in both the rate and size of events in PMEN1 ( Fig . 3A–C & Table 1 ) , and the same was true for the CC180 lineage ( Fig . 4A–C & Table 2 ) ., A standard way to empirically describe heterogeneity is to quantify over-dispersion of the distribution of interest ., To quantify heterogeneity in frequency and size in both lineages , we extended the approach in model 1 ., The extension of Poisson and geometric distribution is in both cases a negative binomial distribution with parameter , which reduces to a geometric distribution for and to Poisson for very large values of ( see Fig . 1 ) ., A model based on a negative binomial distribution of events per branch with mean and dispersion coefficient , and a negative binomial distribution of event sizes with mean bp and dispersion coefficient fit the data much better than the homogeneous , Poisson-based model for the PMEN1 dataset ( ; Fig . 3D–F & Table 1 ) and also for the CC180 dataset ( ; Fig . 4D–F & Table 2 ) ., This demonstrates that both the recombination rate and recombination event size are heterogeneous processes , but gives little insight into the potential mechanisms generating heterogeneity ., Heterogeneity in the recombination rate suggests that recombination sometimes occurs in discrete saltations rather than at a homogeneous rate ., We further observed a correlation between the frequency of recombination events and their size ( Fig . 2C and 2F ) ., We thus modelled the recombination process by a mixture of two , homogeneous recombination processes ., The first process , which we refer to as micro-recombination , leads to single small replacements ., The second process , which we refer to as macro-recombination , leads to multiple synchronous or near-synchronous larger replacements ., We assumed that the micro-recombination process is described by the same parameters and as in the null model; the macro-recombination process occurs at rate , in which multiple tracts of DNA are incorporated into the genome by HR simultaneously ( or at least in a short period of time compared to the genealogical branching process , so that these end up assigned to a single phylogenetic branch ) ., We model the number of gene segments incorporated per macro-recombination event by a Poisson distribution with mean , and the event sizes are geometrically distributed with mean length of genetic tract replaced by recombination for each event being bp ( see Fig . 1 ) ., In this model , the heterogeneity in rates is generated dynamically through the process of near-simultaneous recombination events , but this model alone does not generate excess heterogeneity in the size distribution of recombination event ., The mixture model 3 provided a much better fit than the homogeneous model 1 for both PMEN1 lineage and CC180 lineage ( and , respectively ) ., It also provided a better fit than the heterogeneous model 2 ( and ) , although results of comparing non-mechanistic descriptions of heterogeneity ( Model 2 ) to mechanistic models ( Model 3 ) should be interpreted with caution , since mechanistic models are likely to be more useful even for equivalent goodness of fit ., ( See also Figures 3G–I and 4G–I , Tables 1 , 2 and 3 . ), A key property of the mixture model ( Model 3 ) is that it generates correlation between the rate of recombination and the size of recombination events , since macro-recombination events , when they occur , are simultaneously larger and more numerous ., To test whether this correlation was supported by the data , we compared the mixture model to a model identical in every respect , except for this correlation between rate and size ( the uncorrelated mixture model 4 ) ., The resulting model fitted the data less well than the mixture model , with for PMEN1 data ( Fig . 3J–L & Table 1 ) and for CC180 data ( Fig . 4J–L & Table 2 ) ., In summary , the mechanistic mixture model 3 fit to the data well and generated novel mechanistic insight ., These results were not dependent on the units used to measure branch length ( see Methods and Text S2 ) ., Maximum likelihood estimates of the parameters and univariate 95% confidence intervals are given in Table 3 ., We then used this best fit model to determine the probability that each of the recombination events was generated either by micro-recombination or by macro-recombination ., We found that of 615 events detected in PMEN1 lineage , 136 were likely to have been generated by micro-recombination , and 389 were likely to have been generated by macro-recombination , with the remainder indeterminate ., In CC180 lineage , of 79 events , 14 were likely to have been generated by micro-recombination , and 64 were likely to have been generated by macro-recombination , with only one event indeterminate ., The location of each event along the pneumococcal genome as well as in the inferred phylogeny of PMEN1 and CC180 lineage is shown in Figure, 5 . This figure shows the heterogeneity of recombination in the phylogenies of both lineages , where certain branches exhibit multiple , long macro-recombinations , whereas short , micro-recombinations tend to be more randomly distributed ., This can also be seen in supplementary Figures 10 and 11 in Text S2 , where an alternative distribution of recombination events in both lineages ( i . e . , all independent recombination events along the genome sorted by branch length ) is shown ., Finally , the distribution of micro- and macro-recombination events as a function of their length and the inferred number of SNPs is given in Figure, 6 . The figure shows that the inferred SNP density of micro- and macro-recombinations varies by approximately one order of magnitude , suggesting that the actual rate of micro-recombination may be considerably higher than that detectable through these data ( but see Discussion ) ., In PMEN1 , 10 serotype-switching events were observed 19 ( i . e . , those which induced a change from the serotype 23F to a different one ) , and all those events were found to be with 100% posterior probability likely to have been the result of macro-recombination ., More generally , to examine whether recombinations at major antigen loci are likely macro-recombinations , we counted the number of recombinations spanning or overlapping five major antigen loci in PMEN1 ( pspA , capsule biosynthesis locus , or cps , pclA , psrP and pspC ) and three major antigen loci in CC180 ( pspA , cps , and pspC ) ., Of 171 such detected recombinations in PMEN1 , 93 were likely to have been generated by macro-recombination ., By contrast , in CC180 only 4 recombinations at major antigens were found , however all 4 of them were likely to have been generated by macro-recombination ., To assess our method of detecting heterogeneity of recombination in the genetic data we designed a simulation framework where we evolved a pneumococcal lineage over time with four prespecified mechanisms of recombination , and examined how well we can distinguish between those mechanisms ( see Methods and Text S3 ) ., Specifically , we designed analyses in which the PMEN1 reference genome diversified into a sample of related sequences through discrete time-steps as specified by one of four different simulation frameworks ( Models A–D ) ., We then reconstructed the evolutionary history of the lineage , with recombination events mapped onto the phylogeny , as described above and in 19 ., We next fitted our four models of recombination ( Fig . 1 ) to assess which of them best explains the underlying mechanism of diversification ( see Tables 6–7 in Text S3 ) ., In the first simulation ( A ) , recombination was simulated as a homogeneous process , and the homogeneous model 1 was the best fit ., In the second simulation ( B ) , the distinction between micro-recombination and macro-recombination was introduced but only based on frequency and not size , and in these cases model 3 was the best fit to the data ., However , there was no significant difference in the size distributions between the two modes of recombination , contrasting with the fits to the genomic data ., In the third simulation ( C ) , a full mixture model of micro- and macro recombination was considered , and again model 3 was the best fit , with the likelihood of each model fits being of the same order of magnitude as in PMEN1 and CC180 data ., Finally , in the fourth simulation ( D ) , an uncorrelated mixture model was assumed with independent heterogeneity in frequency and size ., In this case , in two runs there was no significant difference in the fit of model 3 and 4 , while in the third model 4 was a much better fit to data than model 3 ., These simulations thus demonstrate that the observation of model 3 fitting the genomic data best , with a dramatic difference in lengths between the micro- and macro-recombinations , is unlikely to be an artefact of the method used to detect recombination , or the models formulation We next investigated whether the obtained results can explain recent observations of recombination in the pneumococcus using whole genome data ., The near-simultaneous import of multiple fragments through transformation has previously been observed between a donor and recipient during a chronic infection in vivo in one patient 14 , and also inferred through reconstructing the history of another lineage , sequence type 695 15 ., In the study by Hiller and colleagues 14 , 16 recombination events varying in size from 0 . 4 kb to 235 kb ( mean of 15 kb ) were unidirectionally transferred from one donor strain into a recipient strain during an infection followed over a period of seven months ., The observation that , in each case , multiple long recombinations had occurred over a defined short period suggested these examples might represent clear examples of the macro-recombination process ., We found the size distribution of macro-recombinations to be in accordance with the one observed by Hiller et al . for both PMEN1 ( see Fig . 7A ) and CC180 lineage ( see Fig . 7B ) ., On the other hand , the study by Golubchik et al . identified 53 recombination fragments in 5 vaccine escape recombinant lineages , ranging in size from 0 . 4 kb to 90 kb ( mean of 10 kb ) ., Although the distribution of recombination sizes inferred by this analysis of re-sequencing data did not resemble any of the distributions defined by the models of recombination presented here , it nevertheless suggests a strikingly heterogeneous recombination process ( see Fig . 7C and 7D ) ., A more formal approach would be needed to determine whether this is due to an actual recombination heterogeneity or due to another factor like the method used to infer recombination , or vaccine-induced selection ( see also Discussion ) ., Finally , it has been demonstrated that multiple fragments of DNA can be imported by a member of the PMEN1 lineage during a single period of competence for transformation under controlled conditions 13 ., While the overall distribution of sizes observed was similar to that reconstructed as happening during the lineages diversification , there was less variation in the range of detected sizes ., The discrepancy between the size distributions from the transformation experiment and the one observed in the PMEN1 lineage ( see Fig . 7E ) points to some interesting questions about varying conditions under which pneumococci undergo recombination during their evolution ( see Discussion ) ., Perhaps unsurprisingly , the predicted size distribution of the CC180 lineage was even less consistent with the distribution of recombinations from the in vitro experiment ( see Fig . 7F ) ., One hypothesis that could explain the observed difference between micro- and macro-recombination could be the effect of mismatch repair ( MMR; see also Discussion ) ., MMR inhibits the acquisition of polymorphisms through transformation , but in the pneumococcus becomes saturated upon the import of around 150 SNPs 23 , 24 ., Thus micro-recombinations could be acquired under the constraint of this system , whereas macro-recombination could represent the acquisition of sequence unlimited by MMR ., In accordance with this hypothesis , when we divided branches of the phylogeny on the basis of the most common mechanism of recombination occurring on them , those on which micro-recombination predominated generally imported fewer than 150 substitutions in total , while those on which macro-recombination was more common typically acquired many more than this ( see Figures 12–13 and Text S2 ) ., We also examined whether there were differences in the types of substitutions introduced by micro- and macro-recombination , as MMR varies in the efficiency with which is repairs different mutations ., We found that macro-recombinations were enriched for ‘low efficiency markers , which are repaired most effectively by MMR both in PMEN1 ( ) , and in CC180 ( ) ., Interestingly , no association between the type of marker and the type of recombination was observed in the simulated pneumococcal sequences with preassumed micro- and macro-recombination mechanism ( see Table 8 and Text S2 ) ., Our analysis shows that both analysed lineages of Streptococcus pneumoniae , the multi-drug resistant PMEN1 and the older but less diverse CC180 , have likely evolved under two distinct homologous recombination processes ., The first process , which we call micro-recombination , occurred at a homogeneous clock-like rate and gave rise to isolated small genetic replacements ., The second process , which we call macro-recombination , was more erratic , giving rise to large , multiple synchronous ( or near-synchronous ) replacements ., While in PMEN1 we found both micro- and macro-recombinations to have occurred at a similar rate ( every 17 years ) , in the less rapidly diversifying CC180 lineage micro-recombination was more frequent than macro-recombination ( once in 340 years vs . once in 770 years ) ., Overall , recombination was much more heterogeneous in CC180 ., Furthermore , the difference in sizes between micro- and macro-recombination was found to be greater in CC180 ( 0 . 03 kb vs . 14 kb ) than in PMEN1 ( 0 . 6 kb vs . 9 kb ) ., Finally , the number of simultaneous recombinations imported during macro-event was smaller in PMEN1 than in CC180 ( 2 . 3 vs . 15 ) ., The best fit parameters , together with the 95% confidence intervals , are summarised in Table 3 ., The principal caveat in this analysis is that it is dependent on the correct identification of both the genealogy and the recombinations in the original analysis of the PMEN1 and CC180 lineages 19 , 20 ., The main evidence given for the correct identification of the recombinations is that their removal from the set of base substitutions used to construct the phylogeny results an improved ability to detect evidence of a molecular clock at a rate similar to other bacteria that do not undergo frequent homologous recombination 19 , 25 , the length distribution of putative events is similar to that detected experimentally 13 , and that recombination events that can be inferred from phenotypic data ( e . g . , serotype switches ) are predicted at the correct locus on the expected branch of the tree 12 , 19 ., However , we note that there is an inherent bias in the method described by Croucher et al . , shared with other methods that use SNP density to detect recombination ( e . g . , maximum Chi-square method , ClonalFrame 21 ) , in that it is prone to missing short recombination events that happen to bring in few SNPs into the genome ., Nonetheless , such events have a relatively small effect on estimates of branch length , and therefore estimates of the molecular clock rate ., However , such bias means that we have likely under-estimated the rate of micro-recombination ., This is best illustrated by comparing SNP density to the observed size of the recombination ( Figure 6 ) ., The observed negative correlation between SNP density and recombination size ( Spearmans rank correlation: , for PMEN1 and , for CC180 ) is likely the result of the detection bias described above , and this suggests that we may lack the sensitivity to accurately quantify the rate of micro-recombination events ., Simulations of the heterogeneity suggest that the actual rate of micro-recombination is likely to be roughly three times the estimated rate ., Correspondingly , we found that the methods employed in this study were able to correctly identify the underlying model of evolution when simulations were performed under different models of diversification ., This suggests that our observations are unlikely to be an artefact of the method used to detect recombination ., The presented analysis provides a quantitative model that could potentially explain other observations of recombination in the pneumococcus using whole genome data ., The near-simultaneous import of multiple fragments through transformation has been observed previously in in vivo 14 , 15 and in vitro studies 13 ., We found that the micro/macro-recombination process could be consistent with size distributions of recombinations in some patient-derived sequences ( cf . Fig . 7 ) ., However , there is weak evidence that this happens in the case of transformation in vitro ., Therefore the observation of these two different types of recombination requires an explanation that can link the differences in properties and kinetics ., It could be that genetic transformation through the competence system is only responsible for recombination through one of the modes , like micro-recombination , while other forms of bacterial “sex” , like conjugation or transduction , would lead to the acquisition of long stretches of DNA associated with macro-recombination ., Conjugation has been observed to cause extensive sequence transfer in other streptococci , which would be consistent with this hypothesis conjugative transfer can result in multiple events if multiple conjugative origins are involved 26 ., However , these exchanges are associated with ori sequences from conjugative elements , and therefore result in more regular recombination boundaries than are observed for the macro recombination events in this analysis 27 ., Similarly , general transduction of sequence can import large DNA fragments of variable lengths , but typically only one can be packaged into a virion ., As such mispackaging events are rare , this does not provide a likely explanation for the near-simultaneous import of multiple fragments 28 ., Another potential explanation of the difference between micro- and macro-recombination may be how stretches of DNA are processed within the cell ., For example , the recently identified competence-specific DNA-binding protein SsbB has been found capable of storing about 1 . 15 Mb of DNA imported by the competence system 27 ., As the expression of this protein varies according to regulatory processes , it could play an important role in controlling the properties of recombination ., However , given the comparatively homogeneous length distribution of recombinations observed in experimental transformation of the pneumococcus , it seems likely that extracellular degradation or intracellular processing are not the best candidates to explain the observed heterogeneity ., Hence it seems more likely that the observed dynamics represent transformation behaving in two distinct modes ., One known threshold that could explain the variation is saturation of repair systems ., MMR inhibits the acquisition of polymorphisms through transformation , but in the pneumococcus becomes saturated upon the import of around 150 SNPs 23 , 24 ., Here we found moderate but significant evidence for this hypothesis , which would suggest that it is the extent and type of DNA imported that triggers the switch between the two types of exchange ., In the PMEN1 dataset , each homologous recombination imports a mean of 70 substitutions ( 116 substitutions for CC180 ) , and in vitro experiments have demonstrated that multiple fragments can be imported simultaneously ., Therefore the availability of high concentrations of divergent DNA , as observed in pneumococcal biofilms 29 , or a state of ‘hyper-competence’ , in which cells imported DNA more readily than normal , would seem likely to saturate the MMR system and potentially trigger the conditions required for macro-recombination ., The idea of the emergence of micro-recombination and macro-recombination via saturation of the MMR has the advantage that it is consistent with the observed positive correlation between frequency and size of recombinations ( cf . Fig . 2C and 2F ) ., Many macro-recombinations found in this study are considerably larger than any individual segment of donated sequence acquired by S . pneumoniae in vitro ., Thi

Introduction, Methods, Results, Discussion

The bacterium Streptococcus pneumoniae ( pneumococcus ) is one of the most important human bacterial pathogens , and a leading cause of morbidity and mortality worldwide ., The pneumococcus is also known for undergoing extensive homologous recombination via transformation with exogenous DNA ., It has been shown that recombination has a major impact on the evolution of the pathogen , including acquisition of antibiotic resistance and serotype-switching ., Nevertheless , the mechanism and the rates of recombination in an epidemiological context remain poorly understood ., Here , we proposed several mathematical models to describe the rate and size of recombination in the evolutionary history of two very distinct pneumococcal lineages , PMEN1 and CC180 ., We found that , in both lineages , the process of homologous recombination was best described by a heterogeneous model of recombination with single , short , frequent replacements , which we call micro-recombinations , and rarer , multi-fragment , saltational replacements , which we call macro-recombinations ., Macro-recombination was associated with major phenotypic changes , including serotype-switching events , and thus was a major driver of the diversification of the pathogen ., We critically evaluate biological and epidemiological processes that could give rise to the micro-recombination and macro-recombination processes .

Streptococcus pneumoniae , a bacterium commonly carried asymptomatically by children , is a major cause of diseases such as pneumonia and meningitis ., The species is genetically diverse and is known to frequently undergo the remarkable process of transformation via homologous recombination ., In this process , the bacterial cell incorporates DNA from other , closely related bacteria into its own genome , which can result in the development of antibiotic resistance or allow cells to evade vaccines ., Therefore it is important to quantify the impact of this process on the evolution of S . pneumoniae to understand how quickly the species can respond to the introduction of such clinical interventions ., In this study we followed the recombination process by studying the evolution of two important and very different lineages of S . pneumoniae , PMEN1 and CC180 , using newly available population genomic data ., We found that pneumococcus evolves via two distinct processes that we term micro- and macro-recombination ., Micro-recombination led to acquisition of single , short DNA fragments , while macro-recombination tended to incorporate multiple , long DNA fragments ., Interestingly , macro-recombination was associated with major phenotypic changes ., We argue that greater insight into the adaptive role of recombination in pneumococcus requires a good understanding of both rates of homologous recombination and population dynamics of the bacterium in natural populations .

medicine and health sciences, mathematics, genome evolution, statistics (mathematics), epidemiology, biology and life sciences, population biology, physical sciences, computational biology, evolutionary biology, evolutionary genetics

journal.pgen.1000978

A Genome-Wide Association Study of Optic Disc Parameters

The optic nerve head , or optic disc , is the place where the axons of the retinal ganglion cells leave the eye and form the optic nerve ., Its morphology , visible by ophthalmoscopy , is important in the diagnosis and follow-up of patients with ( neuro- ) ophthalmologic diseases , such as ischemic and hereditary optic neuropathies , optic neuritis , papilledema and primary open-angle glaucoma ( OAG ) ., Optic disc parameters of interest are the surface of the optic nerve head referred to as the optic disc area ( measured in units of mm2 ) , and the vertical cup-disc ratio ( VCDR ) ., The optic disc area is associated with general characteristics ( such as body height ) as well as ocular ones ( such as axial length ) 1 , 2 ., The relation to axial length makes the optic disc size directly relevant for nearsightedness ( myopia ) , one of the most common ophthalmic disorders ., Furthermore , it has been suggested that larger optic discs may suffer more from intraocular pressure-related stress , a strong risk factor for OAG 3 ., However , the association of the size of the optic disc to OAG is not clear since it has been argued that larger optic discs may have a larger anatomical reserve for various optic neuropathies such as OAG due to higher number of nerve fibers 4 ., Effects may even partially counteract each other 4 ., The VCDR is a parameter commonly used in the clinical glaucoma management 5 ., The VCDR is determined by comparing ( in a vertical direction ) the size of the cup , a region without axons , to the size of the optic disc ., An increase in VCDR may indicate the occurrence of glaucomatous changes of the optic nerve head , referred to as glaucomatous optic neuropathy 6 ., In addition , an unusual large VCDR at a single observation is a significant determinant of glaucoma 7 , 8 ., The heritability of the optic disc area and VCDR are estimated to be around 52–59% and 48–80% , respectively , 9–12 suggesting a major role for genetic factors ., This prompted us to study the genes determining the optic disc area and VCDR as endophenotypes for myopia and OAG ., To identify genetic determinants of optic disc area and VCDR , we performed a genome-wide association study ( GWAS ) of optic disc area and VCDR using data from Caucasian participants of the Rotterdam Study RS ( cohort I and II , in which participants have an identical age distribution and eye assessment ) and replicated our findings in three independent cohorts of Caucasian ethnicity: the Rotterdam Study III RS-III , a younger cohort , the Erasmus Rucphen Family ERF study and the TwinsUK cohort ( see Materials and Methods for details of all cohorts ) ., Next , we examined whether the genome-wide significant Single Nucleotide Polymorphisms ( SNPs ) were related to myopia and OAG using data from patients with ( one of ) these diseases from the Rotterdam Study I ., The discovery cohorts included 5 , 312 ( RS-I ) and 2 , 048 ( RS-II ) participants who were genotyped and had reliable optic disc data , resulting in a total of 7 , 360 participants included in the primary GWAS discovery set ., A small fraction ( 205 from RS-I and 90 from RS-II ) , had missing or unreliable baseline data; for these we used the data available at follow-up ., From RS-III , 1 , 966 participants were included , and from ERF 1 , 646 , resulting in a total of 10 , 972 when combining the discovery and replication cohorts from the Netherlands , and 11 , 815 when also including the 843 participants of TwinsUK ., Table 1 summarizes the general characteristics of the discovery and replication cohorts ., There are significant differences between the cohorts in terms of age ( discovery cohort is older ) , gender ( TwinsUK includes only women ) and optic disc parameters ( due to different disc-assessment techniques see Materials and Methods; the analyses were adjusted for this difference ) ., Figure S1 and Figure S2 show the Q-Q plots for the observed versus expected p-values for each individual study and for the meta-analysis of the discovery and replication cohorts for optic disc area and VCDR , respectively ., Genomic control for all four cohorts showed low dispersion for optic disc area as well as for VCDR with inflation factors in the range of 1 . 024 and 1 . 061 ., Figure 1A presents the −10log p-plot for the primary discovery cohort for optic disc area and shows two loci on chromosomes 1 and 10 , including 192 SNPs that are beyond the genome-wide significance threshold of 5×10−8 ., Exclusion of OAG ( N\u200a=\u200a188 ) and myopia ( N\u200a=\u200a115 ) cases did not alter the results ., Replication analyses in two independent cohorts of Dutch origin ( RS-III and ERF study ) showed that the findings from all cohorts were consistent in the direction of the effect with p-values ranging from 1 . 69×10−3 to 2 . 39×10−10 ( Table 2 ) ., The combined analysis of the discovery and Dutch replication cohorts yielded an overall p-value 1 . 82×10−27 for rs1192415 ( optic disc area increased by 0 . 064±0 . 006 mm2 beta ± standard error when comparing those heterozygous with homozygous persons for the reference allele ) , and p-value 2 . 05×10−32 for rs1900004 ( optic disc area decreased by 0 . 068±0 . 006 mm2 ) ., Table 2 shows the results of the top SNPs of all loci with p-values <10−6 observed in the meta-analysis ., The meta-analysis of the four Dutch cohorts revealed a cluster of 10 SNPs on chromosome 16q12 . 1 showing borderline genome-wide significant evidence for association with the optic disc area ( p\u200a=\u200a6 . 48×10−8 ) ., When joining the Dutch data with the TwinsUK series ( Table 3 ) , this region became genome-wide significant ( p\u200a=\u200a5 . 07×10−9 ) ., Table 3 shows that also the chromosome 1 and 10 regions were also replicated consistently in the TwinsUK cohort ., The regions of interest for optic disc area are shown in Figure, 2 . The first region on chromosome 1p22 is located between the cell division cycle 7 ( CDC7 ) and the transforming growth factor beta receptor 3 ( TGFBR3 ) gene , but the SNPs in the intergenic region were most significant ., The genome-wide significant region on chromosome 10q21 . 3-q22 . 1 was quite large and included several genes ., The region includes the Myopalladin ( MYPN ) gene , the heterogeneous nuclear ribonucleoprotein H3 ( 2H9 ) ( HNRNPH3 ) gene , RUN and FYVE domain containing ( RUFY2 ) gene , DNA replication helicase 2 homolog ( yeast ) ( DNA2 ) gene , solute carrier family 25 ( mitochondrial carrier; Graves disease autoantigen ) , member 16 ( SLC25A16 ) gene ., However , the most significant evidence was found in the region between the atonal homolog 7 ( ATOH7 ) gene and the phenazine biosynthesis-like protein domain containing ( PBLD ) gene ., The nearest gene in the third region on chromosome 16q12 . 1 was the sal-like 1 ( SALL1 ) gene ., Together , the three SNPs associated with optic disc area explained up to 2 . 7% of the variation in optic disc area ., Next , we evaluated the association of these loci with clinically relevant ophthalmic outcomes ( myopia and OAG; Table S1 ) ., None of the optic disc area loci were associated with myopia-related outcomes ( p-values ranging from 0 . 09 to 0 . 80 ) ., Of the three loci associated with optic disc area we found only the 10q21 . 3-q22 . 1 locus to be marginally associated with OAG ( p\u200a=\u200a0 . 04 for rs1900004 ) ., All analyses for VCDR were adjusted for optic disc area ., Figure 1B presents the −10log p-plot for the discovery cohorts ( meta-analyzed RS-I/RS-II GWAS ) for VCDR and shows two loci reaching genome-wide significance at a threshold of 5×10−8 ., Adjustment for the intraocular pressure did not alter the results nor did exclusion of the OAG cases ., The combined analysis of the discovery and two Dutch replication cohorts yielded an overall p-value of 1 . 96×10−14 for rs1063192 and 9 . 30×10−11 for rs10483727 ( Table 4 ) ., The regions of interest for VCDR are shown in Figure, 3 . The genome-wide significant region on chromosome 9 included two genes from the same gene family ( cyclin-dependent kinase inhibitor 2A CDKN2A and CDKN2B ) ., For chromosome 14 , several genes were included in the region of interest ., The strongest association was found for rs10483727 close to the sin oculis homeobox homolog 1 ( SIX1 ) gene , but also several SNPs flanking SIX6 were genome-wide significant as well as one SNP between RNA-binding motif 8B ( RBM8B ) and the protein phosphatase 1A ( PPM1A ) gene ., Furthermore , there were four other loci that showed consistent evidence for association and reached genome-wide significance in the combined analysis of all Dutch cohorts ( Table 4 ) ., This included the chromosome 10q21 . 3-q22 . 1 region identified for the optic disc area ( Table 2 ) ., For chromosome 11q13 , the most significant SNPs were found in between the FERM domain containing 8 ( FRMD8 ) and the SCY1-like ( SCYL1 ) gene ., The region of interest also harboured latent transforming growth factor beta binding protein 3 ( LTBP3 ) ., The genome-wide significant SNPs of these three regions were all in the same linkage disequilibrium block , hampering determination of the most important variant ( Figure 3 ) ., Of the other two genome-wide significant loci , the SNPs point to the doublecortin–like kinase 1 ( DCLK1 ) for chromosome 13q13 , and CHK2 checkpoint homolog ( CHEK2 ) for chromosome 22q12 . 1 ( Figure 3 ) ., Finally , when combining all top SNPs from the joint analysis of the four Dutch cohorts with the TwinsUK , one additional borderline genome-wide significant region emerged as genome-wide significant ., The region comprises 2 SNPs on chromosome 17q23 ( p\u200a=\u200a2 . 81×10−8; Table 5 ) ., The combined effect of the six loci associated with VCDR explained 2 . 2% of the variation in the VCDR ., Also for the VCDR none of the loci were associated to myopia at p<0 . 05 ., When we evaluated the association to OAG , four of the loci associated with VCDR were also found to be marginally associated with OAG , 9q21 ( p\u200a=\u200a0 . 017 ) , 14q22-23 ( p\u200a=\u200a0 . 021 ) , 11q13 ( p\u200a=\u200a0 . 049 ) , and the overlapping gene ATOH7 discussed earlier ., In the present study we identified three genetic loci ( 10q21 . 3-q22 . 1 , 1p22 and 16q12 . 1 ) associated with optic disc area , and six genetic loci ( 9q21 , 14q22-23 , 10q21 . 3-q22 . 1 , 11q13 , 13q13 , and 22q12 . 1 ) associated with VCDR ., Of these , one ( 10q21 . 3-q22 . 1 ) was associated with both quantitative traits ., For these regions , the evidence for the association was genome-wide significant and our findings were consistently replicated in the independent replication cohorts ., The SNPs in these loci were common variants with minor allele frequencies ranging from 0 . 21 to 0 . 46 ., The genome-wide significant SNPs of the present study were not in linkage disequilibrium with known missense mutations ., The combined effect of the three SNPs involved in the optic disc area explained 2 . 7% , while the six loci associated with VCDR explained 2 . 2% of the variation ., The region with the strongest statistical evidence for association was a locus on chromosome 10q21 . 3-q22 . 1 , which was associated with both optic disc area and VCDR , and included multiple genes ., Although the genome-wide significant region is very large for the optic disc area analysis , the ATOH7 gene ( also known as Math5 ) showed the most significant evidence for association with VCDR ., This gene is expressed in the retina where it controls photoreceptor development 13 ., In animal studies with mice , ATOH7 expression has been found in the developing optic nerve during embryogenesis 14 ., During retinogenesis , seven different major classes of cells develop out of the progenitor cells in the eye: photoreceptors ( rods and cones ) , bipolar cells , horizontal cells , amacrine cells , retinal ganglion cells ( RGC; these are the cells involved in OAG ) and Müller cells ., Degeneration of these cells may lead to blindness 15 ., In mutant mice and zebrafish without ATOH7 , optic nerves and RGC are not further developed , while amacrine cells and cones are formed in excess 16 , 17 ., Overexpression of ATOH7 and interaction with the neuroD gene in chickens increases the amount of RGC and photoreceptors 18 ., The duration of expression of ATOH7 is regulated by several proteins , including Growth and Differentiation Factor 11 ( GDF11 ) 19 ., Another factor involved in this genetic pathway is Sonic hedgehog ( SHH ) , which mediates the direction of growth as the eye develops from the central part towards the periphery ( including the optic nerve ) 20 ., Thus the SHH and GDF11 regulate ATOH7 , which in turn regulates Brn3b ., This gene may play a role in further differentiation of RGC and is expressed in post-mitotic RGC precursors ., First , RGC differentiate into the lower retinal epithelium ( later becoming the RGC layer ) ., At the same time , the dendrites reach the bipolar , horizontal , and amacrine cells in the inner retinal plexiform layer , while their axons form the optic nerve , optic chiasm , superior colliculus and lateral geniculate nucleus 20 ., Although ATOH7 has been implicated in retinal development in animals , this gene has not been linked to the development of the optic nerve pathology in humans ., The analysis of VCDR showed that the ATOH7 ( rs1900004 ) was also significantly associated with VCDR , independent of optic disc area ., This suggests that this gene is involved in both the optic disc area as in VCDR ., The 1p22 region is second in terms of strength of association based on the p-values ., This region includes the genes CDC7 and TGFBR3 associated with optic disc area ., CDC7 encodes for a cell division cycle protein with kinase activity ., Overexpression of this gene has been found in neoplastic transformations in some tumors ., Although this region is associated with the optic disc area , the protein that CDC7 encodes for interacts with the CDKN2A protein associated with VCDR ., However , also the TGFBR3 is of interest because of the interaction of ATOH7 with GDF11 , a member of the bone morphogenetic protein ( BMP ) and the TGFbeta superfamily ., The genes therefore point to the same signaling pathway ., GDF11 interacts with the latent transforming growth factor beta binding protein 3 ( LTBP3 ) ., In our analyses targeting VCDR , we found genome-wide significant evidence for a relation of LTBP3 to VCDR ( see below ) ., While CDKN2A is not known to be involved in TGFbeta signaling , CDKN2B has been implicated in this pathway ., As in the VCDR analysis , the most significant SNPs on chromosome 9p21 were located within the CDKN2B gene ., This gene ( also known as p15Ink4b ) lies adjacent to the tumor suppressor gene CDKN2A and encodes a cyclin-dependent kinase ., The protein encoded by CDKN2B is thought to play a role in cell growth regulation and is induced by transforming growth factor beta ( TGFB ) 21 ., The p15ink4b protein phosphorylates and inactivates the retinoblastoma tumor suppressor ( pRb ) protein 22 ., Deletions of this gene and of the retinoblastoma 1 gene are often found in malignant gliomas and melanomas 23 ., A recent study in mice found that p15Ink4b was ectopically expressed in both zinc finger E-box binding homeobox 1 ( Zeb1 ) mutant cells and neuroectodermally derived cells , including the developing retina , optic nerve and muscles surrounding the eye 24 ., Taken together , our findings point to a central role of TGFbeta in the development of the optic disc and VCDR ., TGFbeta is a multifunctional cytokine that modulates developmental and repair processes in several tissues ., TGFbeta signaling has been implicated in a wide variety of diseases including inflammation , autoimmune disorders , fibrosis , cancer and cataracts ., The region has recently also been associated with myocardial infarction and type 2 diabetes mellitus 25 ., The CDKN2B/CDKN2A and CDC7/TGFBR3 loci influence the VCDR independently of optic disc area as these genes were not significantly associated with the optic disc area ( p>0 . 05 ) ., However , TGFBR3 appears to be involved in VCDR through its role in optic disc area , as the effect of this gene on VCDR increased two fold when we did not adjust for optic disc area ( RS-I: unadjusted beta\u200a=\u200a0 . 015 , standard error\u200a=\u200a0 . 004 , p\u200a=\u200a2 . 45×10−5 compared to beta\u200a=\u200a0 . 007 , standard error\u200a=\u200a0 . 003 in the adjusted analysis ) ., Regarding the optic disc area , we found one additional region genome-wide significantly associated when pooling the data of the Dutch and TwinUK ., Although the chromosome 16q12 . 1 region concerns a gene desert , the closest gene in the third locus associated with optic disc area is SALL1 ., Defects in this gene are a cause of Townes-Brocks syndrome and the bronchio-oto-renal syndrome , two autosomal dominant disorders 26 ., Only rare variants have been implicated in Townes-Brocks syndrome and bronchio-oto-renal syndrome , while the association we report here is with common variants ., One of the traits involved in the latter syndrome is myopia 27 ., However , in our analyses we could not find evidence for an association of the common SNPs in the SALL1 region to myopia ( rs1362756; p\u200a=\u200a0 . 802 ) ., SALL1 encodes a zinc finger transcriptional repressor ., When considering the protein pathway , SALL1 interacts with SIX1 28 ., Rare variants in SIX1 are involved in the bronchio-oto-renal syndrome 29 ., We found that common variants in SIX1 were genome-wide significantly associated with VCDR ., Regarding VCDR , chromosome 14q22-23 was genome-wide significant in the discovery cohorts and was replicated consistently in the other cohorts ., The region includes two genes which are obvious candidates SIX1 and SIX6 ( the latter also known as Optx2 and about 94kb distance from rs10483727 ) ., This gene is involved in eye development and has been related to congenital glaucoma ., Defects in this gene have been associated with anophtalmia in mice 30 and in humans 31 , 32 ., Embryological studies have shown expression in the ventral optic stalk , which later becomes the optic nerve 33 ., In the adult mouse retina , Optx2 mRNA has been found in cells within the ganglion cell layer and inner nuclear layer 34 ., This gene is expressed in the developing retina , optic nerve and other brain structures 31 ., There were three more genome-wide significant loci on chromosomes 11q13 , 13q13 and 22q12 . 1 associated with VCDR ( Table 2 ) ., On 11q13 most SNPs were found close to SCYL1 , which has been associated with optic nerve atrophy in mice 35 ., However , also the presence LTBP3 in this region is of interest , as this protein binds to TGFB1 , TGFB2 , and TGFB3 , and is thus involved in the same signalling pathway as CDKN2B ., LTBP3 is further of interest because of its homology to LTBP2 , which has been implicated in primary congenital glaucoma 36 , 37 ., The DCLK1 gene on 13q13 is expressed in the optic tectum 38 ., This is a probable kinase that may be involved in a calcium signaling pathway controlling neuronal migration in the developing and mature brain ., Finally , the CHEK2 gene has been associated with several types of cancer , including breast cancer 39 ., A literature search did not show a direct link between CHEK2 and the eye , however one study reported mapping of a locus on chromosome 22q12 . 1–q13 . 1 ( OPA5 ) to autosomal dominant optic atrophy 40 and one case-report described an association of chromosome 22q11 . 2 deletion syndrome with optic disc swelling , which is probably caused by the resulting hypocalcaemia 41 ., Regarding the association of CHEK2 with breast cancer , it is of interest that also one borderline significant SNP is located in a gene breast carcinoma amplified sequence 3 ( BCAS3 ) involved in this pathway ., Although our study has convincingly identified SNPs involved in optic disc area and VCDR , there are also a number of limitations ., At this point , we cannot pinpoint the two endophenotypes to a single clinical outcome ., There was some marginal evidence suggesting that four of the genes involved in the development of the optic disc area and VCDR are relevant for OAG ., However , the findings were far from genome-wide significance and remain to be confirmed ., Another limitation concerns the differences in methodology ., Two of the four replication cohorts , RS-III and ERF , used confocal scanning laser ophthalmoscopy to determine the optic disc area , while the other studies , RS-I , RS-II and TwinsUK , used digitized stereoscopic images ., Although this may be considered a drawback , we do not think this distorted our results , since , several studies compared both methods and found high correlations for all stereometric parameters 42–44 ., Moreover , since our findings replicated in all cohorts differences across measurements are probably small and unlikely to influence our results , beyond that the estimation of the effects ( beta-coefficients ) may differ across studies ., Finally , the TwinsUK study served as a replication cohort in this study , but is also involved as a replication cohort for a GWAS based on a discovery cohort from Australia ( Macgregor , et al . unpublished data ) ., Both , Dutch and Australian cohorts independently implicated ATOH7 as playing a role in optic disc phenotypes and both utilize the TwinsUK data to replicate their findings ., Although the association of ATOH7 was genome-wide significant in the Dutch validation cohorts , this overlap in replication samples should be taken into account ., In conclusion , by conducting GWA analyses , we found genome-wide significant evidence for the association of three genetic loci associated with optic disc area , and another six with VCDR ., Although multiple genes were included in the regions of interest , the most interesting ones for optic disc area were TGFBR3 on chromosome 1p22 , ATOH7 on chromosome 10q21 . 3-22 . 1 ( also for VCDR ) and SALL1 on chromosome 16q12 ., Regions of interest for VCDR were CDKN2B on chromosome 9p21 , SIX1 on chromosome 14q22-23 , SCYL1 on chromosome 11q13 , CHEK2 on chromosome 22q12 . 1 , DCLK1 on chromosome 13q13 , and BCAS3 on chromosome 17q23 ., There are several pathways implicated but the most interesting is the TGFbeta signaling pathway that appears to play a key role ., Further research is needed to implicate these finding to pathology of the eye ., The Rotterdam Study I ( RS-I ) is a prospective population-based cohort study of 7 , 983 residents aged 55 years and older living in Ommoord , a suburb of Rotterdam , the Netherlands 45 ., Baseline examinations for the ophthalmic part took place between 1991 and 1993; follow-up examinations were performed from 1997 to 1999 and from 2002 to 2006 ., The RS-II and RS-III are two other prospective population-based cohort studies of 3 , 011 residents aged 55 years and older and 3 , 392 residents aged 45 years and older respectively ., The rationale and study design are similar to those of the RS-I 45 ., The baseline examinations of RS-II took place between 2000 and 2002; follow-up examinations were performed from 2004 to 2005 ., Baseline examinations of RS-III took place between 2006 and 2009 ., The Erasmus Rucphen Family ( ERF ) Study is a family-based cohort in a genetically isolated population in the southwest of the Netherlands with over 3 , 000 participants aged between 18 and 86 years ., Cross-sectional examination took place between 2002 and 2005 ., The rationale and study design of this study have been described elsewhere 46 , 47 ., All measurements in these studies were conducted after the Medical Ethics Committee of the Erasmus University had approved the study protocols and all participants had given a written informed consent in accordance with the Declaration of Helsinki ., Finally , the TwinsUK adult twin registry is a volunteer cohort of over 10 , 000 healthy twins based at St Thomas Hospital in London ., Participants were recruited and examined between 1998 and 2008 ., A total of 843 had complete data , all of whom were Caucasian ., This cohort is predominantly female , as only 3% of included participants were male ., The ophthalmic assessment in RS-I and RS-II , both for baseline and follow-up , included a medical history , autorefraction , keratometry , visual field testing and optic nerve head imaging with Topcon ImageNet System of both eyes after mydriasis with topical tropicamide 0 . 5% and phenylephrine 2 . 5% ., RS-III was similar to RS-I except for optic nerve head imaging with confocal scanning laser ophthalmoscopy ( Heidelberg Retina Tomograph 2 HRT ) ., The ophthalmic assessment in ERF included a medical history , autorefraction , keratometry and optic nerve head imaging with HRT of both eyes after pharmacologic mydriasis ., In the TwinsUK optic disc parameters were measured from stereo disc photographs using the Nidek-3DX stereo camera , with digitized images scanned from Polaroid images and StereoDx stereoscopic planimetric software ( StereoDx ) using a Z-screen ( StereoGraphics Corp ) and software obtained from James Morgan from Cardiff University software , Wales , UK 48 ., ImageNet , which was used in RS-I and RS-II , takes simultaneous stereoscopic images of the optic disc at a fixed angle of 20° , using a simultaneous stereoscopic fundus camera ( Topcon TRC-SS2; Tokyo Optical Co . , Tokyo , Japan ) ., Images were analyzed using the ImageNet retinal nerve fiber layer height module ., On each stereoscopic pair of optic disc images four points were marked on the disc margin , defined as the inner border of the peripapillary ring or the outer border of the neural rim , if a scleral ring was visible ., Next , the software drew an ellipse using these points to outline the disc margin and to determine the cup ., The amount of correspondence between the marked points on the two images of the stereoscopic pair is expressed as a “bad points” percentage , which indicates the percentage of points lacking correspondence ., This percentage can be used as an indicator of image quality ., Images with 25% or more bad points were excluded 49 ., HRT 2 , used in RS-III and ERF , uses a focused 670-nm diode laser light beam to acquire scans of the optic nerve head region , using the confocal principle ., The HRT obtains , during one scan , three series of 16 to 64 confocal frontal slices ., From each of these series , a 3-dimensional image of the optic nerve head is reconstructed , from which the software calculates several optic disc parameters ., To define the cup , the HRT places a reference plane 50 mm below the peripapillary retinal surface in the region of the papillomacular bundle ., Imaging was performed after entering the participants keratometry data into the software and after adjusting the settings in accordance with the refractive error ., In RS-III all HRT 2 data was converted to HRT 3 ., As an indicator of image quality we used the topographic standard deviation of the scan , which is a measure of the variability among the three series of a single HRT scan ., Scans with a topographic standard deviation exceeding 50 mm were excluded ., The inter-observer variability and agreement for both systems have been described elsewhere 44 ., Details of the optic disc measurements in TwinsUK are described elsewhere 50 ., Myopia was defined as a spherical equivalent of −6 . 00D or lower ., For each eye the spherical equivalent was calculated using the standard formula: spherical equivalent\u200a=\u200aspherical component+ ( cylindrical value/2 ) ., The mean spherical equivalent of both eyes was included ., Those eyes with a history of cataract surgery were excluded from this analysis ., OAG diagnosis was primarily based on glaucomatous visual field loss ( VFL ) ., The visual field of each eye was screened with a Humphrey Field Analyzer ( HFA II 740; Zeiss , Oberkochen , Germany ) using a 52-point threshold-related supra-threshold test that covered the central field with a radius of 24° ., This test was modified from a standard 76-point screening test 51 , 52 ., VFL was defined as non-response in at least three contiguous test points ( or four including the blind spot ) ., If the first test was unreliable ( >33% false-positive or false-negative catch trials ) or a reliable test showed VFL in at least one eye , a second supra-threshold test was performed on that eye ., If the second supra-threshold test was reliable and showed VFL , a full-threshold HFA 24-2 test ( second follow-up ) or Goldmann perimetry ( Haag Streit , Bern , Switzerland; baseline and first follow-up ) was performed on both eyes ., The classification process of the Goldmann perimetry test results 51 and the full-threshold HFA 24-2 test results Czudowska , et al . unpublished data have been described before ., In short , VFL was considered to be glaucomatous VFL only if reproducible and after excluding all other possible causes ., For the present study , participants were considered as having glaucomatous VFL if they had glaucomatous VFL in at least one eye during either follow-up round ., Cases had to have an open anterior chamber angle and no history or signs of angle closure or secondary glaucoma were allowed 52 ., Criteria for glaucomatous optic neuropathy , such as VCDR , were not included in the criteria for OAG ., In the RS-I , RS-II and RS-III cohorts , DNA was genotyped by using the Illumina Infinium II HumanHap550chip v3 . 0 array according to the manufacturers protocols ., Details are described elsewhere 53 ., After exclusion of participants for reasons of low-quality DNA , a total of 5 , 974 participants were available with genotyping data from RS-I , 2 , 157 participants from RS-II and 2 , 082 from RS-III ., In ERF , DNA was genotyped on four different platforms ( Illumina 6k , Illumina 318K , Illumina 370K and Affymetrix 250K ) , which were then merged ., After exclusion of participants for whom genotyping data were unavailable , 2 , 385 had genotyping data ., As we did not use the same microarray for the various study populations we imputed our genotype data using HapMap CEU as reference population , resulting in over 2 . 5 million SNPs ., Extensive quality control analyses have been performed in each cohort ., Finally , the genotyping of the TwinsUK cohort took place in stages; in the first stage participants were genotyped by using Illuminas HumanHap 300K duo chip , whereas in the second stage participants were genotyped with Illuminas HumanHap610 Quad .

Introduction, Results, Discussion, Materials and Methods

The optic nerve head is involved in many ophthalmic disorders , including common diseases such as myopia and open-angle glaucoma ., Two of the most important parameters are the size of the optic disc area and the vertical cup-disc ratio ( VCDR ) ., Both are highly heritable but genetically largely undetermined ., We performed a meta-analysis of genome-wide association ( GWA ) data to identify genetic variants associated with optic disc area and VCDR ., The gene discovery included 7 , 360 unrelated individuals from the population-based Rotterdam Study I and Rotterdam Study II cohorts ., These cohorts revealed two genome-wide significant loci for optic disc area , rs1192415 on chromosome 1p22 ( p\u200a=\u200a6 . 72×10−19 ) within 117 kb of the CDC7 gene and rs1900004 on chromosome 10q21 . 3-q22 . 1 ( p\u200a=\u200a2 . 67×10−33 ) within 10 kb of the ATOH7 gene ., They revealed two genome-wide significant loci for VCDR , rs1063192 on chromosome 9p21 ( p\u200a=\u200a6 . 15×10−11 ) in the CDKN2B gene and rs10483727 on chromosome 14q22 . 3-q23 ( p\u200a=\u200a2 . 93×10−10 ) within 40 kbp of the SIX1 gene ., Findings were replicated in two independent Dutch cohorts ( Rotterdam Study III and Erasmus Rucphen Family study; N\u200a=\u200a3 , 612 ) , and the TwinsUK cohort ( N\u200a=\u200a843 ) ., Meta-analysis with the replication cohorts confirmed the four loci and revealed a third locus at 16q12 . 1 associated with optic disc area , and four other loci at 11q13 , 13q13 , 17q23 ( borderline significant ) , and 22q12 . 1 for VCDR ., ATOH7 was also associated with VCDR independent of optic disc area ., Three of the loci were marginally associated with open-angle glaucoma ., The protein pathways in which the loci of optic disc area are involved overlap with those identified for VCDR , suggesting a common genetic origin .

Morphologic characteristics of the optic nerve head are involved in many ophthalmic diseases ., Its size , called the optic disc area , is an important measure and has been associated with e . g . myopia and open-angle glaucoma ( OAG ) ., Another important and clinical parameter of the optic disc is the vertical cup-disc ratio ( VCDR ) ., Although studies have shown a high heritability of optic disc area and VCDR , its genetic determinants are still undetermined ., We therefore conducted a genome-wide association ( GWA ) study on these quantitative traits , using data of over 11 , 000 Caucasian participants , and related the findings to myopia and OAG ., We found evidence for association of three loci with optic disc area: CDC7/TGFBR3 region , ATOH7 , and SALL1; and six with VCDR: CDKN2B , SIX1 , SCYL1 , CHEK2 , ATOH7 , and DCLK1; and additionally one borderline significant locus: BCAS3 ., None of the loci could be related to myopia ., There was marginal evidence for association of ATOH7 , CDKN2B , and SIX1 with OAG , which remains to be confirmed ., The present study reveals new insights into the physiological development of the optic nerve and may shed light on the pathophysiological protein pathways leading to ( neuro- ) ophthalmologic diseases such as OAG .

genetics and genomics/gene discovery, public health and epidemiology, public health and epidemiology/epidemiology, ophthalmology, genetics and genomics, ophthalmology/glaucoma, genetics and genomics/population genetics

journal.pgen.1000858

Cdk1 Targets Srs2 to Complete Synthesis-Dependent Strand Annealing and to Promote Recombinational Repair

Homologous recombination ( HR ) is a fundamental DNA repair pathway and its deregulation is responsible for a variety of genomic rearrangements , including chromosome loss , DNA translocations and inversions , which are typical of the genetic alterations seen in tumor cells ( reviewed in 1 ) ., The mechanisms and proteins involved in HR have been well conserved throughout evolution and much of our knowledge on HR comes from studies conducted in the yeast Saccharomyces cerevisiae ( reviewed in 2–5 ) ., HR targets multiple DNA lesions , including single-stranded DNA ( sDNA ) breaks and DSBs , promoting their repair using a region of homology as a template ., Diverse pathways can seal sDNA breaks , but the role of HR in DSB repair is essential ., Different HR sub-pathways compete for DSB repair and some are less accurate than others 6 , 7 ., The position of DNA sequences involved in recombination and the extent of their homology influence the kinetics of DSB repair ., Irreparable DNA breaks 8 , or even those repaired slowly 9 , appear to be sequestered to the nuclear periphery , through a mechanism resembling that used to tether telomeres at the nuclear membrane 10 ., When a region of homology is found on both sides of a DSB , the preferred pathway of repair is gene conversion ( GC ) ., Among the initial steps in GC is the formation of Rad51 presynaptic nucleofilaments assisted by accessory factors ., While Rad51 nucleation can occur directly at sDNA breaks , the ends of DSBs must be first processed to produce sDNA tails in order to recruit Rad51 ., Multiple factors with nuclease and/or helicase activities , including the Mre11/Rad50/Xrs2 complex , Sae2 , Exo1 , Dna2 and Sgs1 cooperate in 5′ to 3′ DSB resection ( reviewed in 11 ) ., Assembled Rad51 nucleofilaments invade and displace a duplex donor homologue DNA template leading to the formation of a D-loop structure ., The D-loop is the site of DNA synthesis , which is promoted by extension of the 3′ invading strand ., According to the canonical DSB repair model 12 , the capture of the second end of the DSB generates a double Holliday junction ( dHJ ) whose resolution , by cutting or branch migration , influences the formation of crossover products associated with GC , that is , the extent of DNA exchanges associated with DSB repair ., If the second DSB end is not captured , it can anneal to the invading strand evicted from the D-loop soon after DNA synthesis ., In this process , called synthesis-dependent strand annealing ( SDSA ) , GC is limited to DNA synthesized from donor strand and crossovers are prevented 4 ., Another HR pathway , known as single strand annealing ( SSA ) , is used when DSB repair occurs between direct repeats 4 ., In this case resected homologous sequences anneal without DNA synthesis and DSB repair is associated with deletion of the sequence between the repeats ., Notably , during SSA , a D-loop is not formed and Rad51 is not required ., The formation of presynaptic Rad51 nucleofilaments is fundamental for HR commitment during GC ., However , Rad51 could nucleate improperly on DNA or even be engaged into damaged filaments when other recombination factors are inactivated: in both cases HR is not proficient , rather it becomes toxic for other DNA transactions ., Many in vivo studies suggest that Srs2 , a member of UvrD family of DNA helicases conserved from bacteria to human , is involved in the removal of toxic Rad51 filaments from sDNA 13–17 ., Further , the Srs2 protein disrupts presynaptic Rad51 filaments through its DNA translocase activity in vitro 18 , 19 ., This Srs2 anti-recombination activity requires a physical interaction with sumoylated PCNA , as it was evidenced in the absence of the post-replication repair ( PRR ) pathway , a context in which Srs2 prevents deadly the recombinational repair 20 , 21 ., Srs2 also exhibits 3′ to 5′ DNA helicase activity on duplex DNA 22 ., Recent in vitro studies in yeast and plants suggest that Srs2 unwinds DNA structures mimicking a D-loop 23 , 24 ., Genetic evidence , indeed , suggests that Srs2 favors the SDSA pathway , since the loss of Srs2 results in an increase in crossover products 25–27 ., Moreover , Srs2 is essential for DSB repair through either SSA or ectopic GC 25 , 28 , 29; in SSA repair , Srs2 is required to mediate recovery from checkpoint-mediated arrest 29 ., Since Srs2 affects HR in several ways , Srs2 functions in recombination are probably regulated ., Previous studies demonstrated that Srs2 is a target of the cell cycle-dependent kinase ( Cdk1 ) in vivo 30 and in vitro 31 ., Cdk1 has been implicated in the DNA damage response and in DSB repair 32; by monitoring repair of one HO-induced break , it was shown that Cdk1 is required both at the level of resection and at a step after Rad51-dependent strand invasion 33 , 34 ., It is known that Cdk1 triggers the resection of DSB ends by phosphorylating Sae2 35 , but other direct targets in DSB repair are unknown ., We found that srs2 mutants that are unable to undergo Cdk1-dependent phosphorylation can still remove toxic Rad51 nucleofilaments , but these srs2 mutants fail to promote homologous recombinational repair ., Analysis on repair of a single HO-induced break through ectopic GC shows that the proper turnover of Srs2 , at D-loop intermediates , is dependent on its modification by phosphorylation and this phosphorylation is essential for completion of the SDSA reaction that results in non-crossover products ., Moreover , the phosphorylation-dependent role of Srs2 does not require an interaction with PCNA and does not affect the turnover of Rad51 at invading filaments ., In the absence of Srs2 phosphorylation , the protein is sumoylated and this is the main cause of the recombinational repair defects seen in the nonphosphorylatable srs2 mutant ., Thus , coordination of the sumoyaltion and phosphorylation modifications on Srs2 is essential during homologous recombinational repair ., Saccharomyces cerevisiae Srs2 contains characteristic amino acid motifs important for ATP-binding and DNA-binding that are highly conserved among members of UvrD family 36 ., All these motifs are located in the N-terminal domain of the Srs2 protein ( grey box in Figure 1A ) and are sufficient for the helicase activity 22 , but not for translocase-dependent removal of Rad51 nucleofilaments , as tested in vitro 37 , 38 ., The C-terminal tail of Srs2 protein plays an important regulatory function , since it mediates protein-protein interactions , including interaction with Rad51 and PCNA 15 , 21 , 37–40 ., Moreover , a cluster of five consensus sites for Cdk1 kinase is present in the C-terminal region of Srs2 , while two additional sites are located in the helicase domain ( Figure 1A; 39 ) ., The last 138 amino acids ( aa ) of the Srs2 C-terminal tail are required for the interaction with PCNA 21 and also contain three consensus sites for sumoylation ( Figure 1A ) ., We previously showed that changing the seven serine/threonine Cdk1 consensus sites to the unphosphorylatable residues alanine/valine abolished DNA damage-induced phosphorylation of Srs2 , which can be monitored as an electrophoretic mobility shift on SDS-PAGE ( Figure 1B; 39 ) ., We then produced a new srs2 allele in which the same serine/threonine residues were changed to the negatively charged aspartic acid/glutamic acid residues , with the aim of producing a mutated version of Srs2 that mimics the constitutively phosphorylated protein isoform ., As shown in Figure 1B , the levels of wt Srs2 and the two mutated Srs2 isoforms are similar , both in normal conditions and in response to DNA damage by methyl methanesulfonate ( MMS ) -treatment ( data not shown ) ., Henceforth , we will refer to the unphosphorylated and phosphorylated srs2 mutants , respectively , as srs2-7AV and srs2-7DE ., To investigate whether Cdk1-dependent phosphorylation of Srs2 is important for its roles in HR , we first evaluated cell survival of the two srs2 phospho-mutants following UV-light and zeocin treatments ., Wild type ( SRS2 ) and srs2Δ strains were used as controls ., Previous studies have shown that the UV-sensitivity of srs2Δ strains is suppressed by mutations in RAD51 , indicating that cell lethality is due to accumulation of toxic Rad51 nucleofilaments at gaps whose repair can occur in the absence of HR 16 ., We found that srs2Δ and rad51Δ mutants are also sensitive to zeocin , a radiomimetic chemical that induces DSBs ( Figure 2A and data not shown ) ., Thus , zeocin-treatment induces DNA lesions whose repair is strictly HR-dependent and prevented in the absence of Srs2 ., As shown in Figure 2A , we found that both srs2-7AV and srs2-7DE mutants , as SRS2 strains , survive UV-light doses that kill srs2Δ mutants ., Conversely , the srs2-7AV mutant , but not the srs2-7DE mutant , is sensitive to zeocin and , indeed , is even more sensitive than the srs2Δ strain ., Previous reports showed that srs2Δ mutations are synthetically lethal with either sgs1Δ or rad27Δ mutations 13 , 14 , 41 ., While the synthetic lethality of srs2Δ sgs1Δ double mutants is suppressed by rad51Δ 13 , single rad27Δ mutants are themselves lethal in combination with rad51Δ 42 ., Thus , the types of spontaneous DNA damage accumulating in sgs1Δ and rad27Δ mutants mirror those induced by UV and zeocin treatments: only in rad27Δ mutants and under zeocin treatment , HR is essential for DNA repair ., We crossed the srs2-7AV and srs2-7DE phospho-mutants and srs2Δ as control with sgs1Δ or rad27Δ mutants ., Heterozygous diploid mutants were sporulated and tetrad analysis was performed ., As shown in Figure 2B , neither srs2Δ sgs1Δ nor srs2Δ rad27Δ double mutants form viable spores; the srs2-7AV mutation , but not the srs2-7DE mutation , is synthetically lethal with the rad27Δ mutation , while both srs2-7AV sgs1Δ mutants and srs2-7DE sgs1Δ mutant spores form colonies ., Hence , the phenotypes of srs2-7AV mutants suggest that Srs2 phosphorylation is dispensable for the reversal of toxic Rad51-dependent recombination intermediates induced at sDNA by UV or by the absence of Sgs1 , but phosphorylation is required to promote recombinational repair in zeocin and in the absence of Rad27 ., Previous data suggested that the Srs2 protein sensitizes postreplication repair ( PRR ) mutants , because it prevents HR 43 , 44 ., Accordingly , as show in Figure 2C , the sensitivity of rad5Δ mutants to MMS is alleviated by deleting SRS2 ., srs2 mutants encoding a protein that displays attenuated translocase activity also suppress the DNA damage sensitivity of PRR mutants , although they are not sensitive to DNA damaging agents by themselves 40 ., Hence , we analyzed the srs2 phospho-mutants in a PRR mutant context , in which the importance of having an intact DNA translocation activity should be revealed ., We constructed srs2 phospho-mutants in rad5Δ or rad18Δ backgrounds and then tested viability on medium containing MMS ., We found that srs2-7AV mutation hypersensitizes rad5Δ and rad18Δ mutants to DNA damage , but , conversely , the srs2-7DE mutation partially suppresses the lethality of rad5Δ or rad18Δ mutation ( Figure 2C and data not shown ) ., Notably , srs2-7AV and srs2-7DE mutants are not sensitive to MMS , even at a higher MMS dose than those employed in Figure 2C ( data not shown ) ., Thus , we conclude that , even in a PRR context , unphosphorylatable Srs2 can remove Rad51 at DNA gaps ., On the other hand , the phosphorylated Srs2 protein isoform appears to be less proficient in the anti-recombinational role ., The observation that srs2-7AV mutants are sensitive to treatment with zeocin suggests that phosphorylation of Srs2 is important in DSB repair ., To directly examine this , we tested the behavior of srs2 phospho-mutants in response to a single DSB created by a galactose-inducible HO endonuclease ., Previous studies have shown that srs2Δ mutants can not survive a single HO-induced DSB when repair of this break occurs either by ectopic GC or by SSA 25 , 28 , 29 ., While the GC pathway strictly depends on RAD51 , SSA can occur in the absence of Rad51 ., There are important differences in the requirement for Srs2 in the two pathways: Srs2 is not required to complete DSB repair during SSA , but it is required for recovery from the DNA damage-induced cell cycle arrest 29 ., RAD51 deletion rescues the checkpoint recovery defect in srs2Δ mutants 29; thus , one hypothesis is that Rad51 accumulates on DNA contributing to the lethal checkpoint-induced arrest , since it can not be removed in absence of Srs2 29 , 45 ., Conversely , during ectopic GC , srs2Δ mutants are unable to complete DSB repair , with a specific reduction in non-crossover products formation 25 ., Since the region of DNA homology involved is limited in ectopic DSB repair , the formation of crossovers might be prevented because the formation of the dHJ intermediate is reduced 46 ., Thus , the failure to carry out SDSA results in loss of non-crossover products and there is a marked reduction in DSB repair efficiency 25 ., To analyze the requirement of Srs2 phosphorylation in the DSB repair response , we assayed cell survival of srs2 phospho-mutants in a SSA system in which DSB repair occurs between repeated sequences , one of which is located 25kb from the DSB and results in a chromosomal deletion 29 or in an ectopic GC system in which DSB repair occurs between chromosomes V and III 25 ., In agreement with previous findings , the rate of cell survival of srs2Δ mutants is 2% in both the SSA and GC systems ( Figure 3A and 3B ) ., This high cell lethality in srs2Δ mutants correlates with inability to dephosphorylate the checkpoint kinase Rad53 , which is activated in response to DSB induction ( Figure 3A and 3B ) ., Cell survival of srs2-7AV mutants is 25% in the GC system where they also fail to fully dephosphorylate Rad53 24 hours after DSB induction ( Figure 3A and 3B ) ., Survival of the srs2-7AV mutant is normal in the SSA system and survival of the srs2-7DE mutant is normal in both systems ., Thus , Srs2 phosphorylation is necessary for cell survival when DSB repair proceeds through the Rad51-dependent GC pathway , but is dispensable in the SSA pathway , which does not require Rad51 ., As mentioned above , although SSA is Rad51-independent pathway , in absence of Srs2 , Rad51 might improperly accumulate on DNA and interfere with checkpoint recovery 29 , 45 ., Since srs2-7AV survive DSB repair via SSA , this further strengthens the conclusion that Srs2 phosphorylation is not required for reversal of toxic Rad51-dependent intermediates ., We used Southern blotting with a probe that recognizes the MAT locus , to physically observe DSB repair in the GC system ( Figure 3B ) ., As mentioned above , in this system a DSB is induced at a MAT locus inserted into Chromosome V and is repaired using a unique uncleavable MAT-inc cassette on chromosome III ( Figure 3B ) ., Notably , crossovers that are associated with the GC event can be evaluated by restriction analysis , since crossovers give rise to chromosomal bands that differ in size from the parental chromosomes and the non-crossover GC products ( Figure 3B ) ., As shown in Figure 3C , DNA of SRS2 and srs2 mutants were analyzed by Southern blotting ., DSB repair efficiency is about 30% in srs2Δ strain , in agreement with previous findings 25 and in srs2-7AV it is reduced to 70% compared to SRS2 or srs2-7DE ( Figure 3C ) ., Moreover , the percentage of crossovers associated with GC increases three-fold in srs2Δ and two-fold in srs2-7AV compared to SRS2 or srs2-7DE ( Figure 3C ) ., Similar to the srs2Δ mutants , the increase in crossovers is associated with a reduction in non-crossover repair efficiency in the srs2-7AV mutant ( Figure 3C ) ; thus , DSB repair defects in the absence of Srs2 phosphorylation likely indicate a specific failure to carry out repair via the SDSA pathway that results in non-crossover products ., Our analysis indicates that Srs2 phosphorylation is required for Rad51-dependent DSB repair ., Although we found that Srs2 phosphorylation is not essential for the removal of toxic Rad51 nucleofilaments at DNA gaps or during DSB repair by SSA , it might be specifically required to remove Rad51-dependent recombination intermediates initiated at D-loop intermediate ., To investigate this possibility , we analyzed Rad51 binding to DSBs by ChIP and Q-PCR in SRS2 and srs2 phospho-mutants ., We used DNA primers that amplified the region of homology located on donor chromosome III ., Using this strategy , proteins localizing either at broken or recipient chromosomes will be immunoprecipitated at the DSB when the invading strand is in duplex DNA , which most likely represents the D-loop ., As shown in Figure 4A , Rad51 protein is undetectable at the donor MAT locus before HO induction , while it is loaded at the DSB with similar kinetics in SRS2 and all srs2 mutated strains ., Thus , we conclude that Rad51-mediated strand invasion occurs with similar kinetics in SRS2 and srs2 mutants ., We also conclude that Rad51 is removed from the DSB with similar kinetics in all contexts and strains analyzed ., Thus , DSB repair defects in srs2Δ or srs2-7AV mutants are unrelated to an abnormal persistence of Rad51 after strand invasion ., Previous findings have indicated that Srs2 is loaded at DSBs 47 ., We asked whether Srs2 phosphorylation could influence its ability to be recruited to DSBs in our GC system ., Using the same ChIP strategy employed above , we found that Srs2 is sited at the invading strand with a three-fold enrichment ( Figure 4B ) ., The Srs2 and Srs2-7DE proteins are loaded and dislodged from DNA with kinetics resembling that of Rad51 , but the Srs2-7AV protein accumulates only at later times and abnormally persists on DNA for at least 24 hours after DSB induction; notably , Rad51 has been displaced from DNA , when Srs2-7AV protein accumulates ( Figure 4 ) ., In summary , the data in Figure 4 suggest that Srs2 is loaded at the D-loop during GC and its proper recruitment is governed by Cdk1-dependent phosphorylation ., However , the DSB repair defects in srs2-7AV or srs2Δ mutants appear not be related to inefficient metabolism of Rad51 nucleofilaments at donor DNA sequences ., In the course of our studies on Srs2 phosphorylation , we noticed that in response to massive DNA damage , such as treatment with 0 . 3% MMS , Srs2 accumulates as additional modified isoforms , which can be visualized as a ladder on SDS-PAGE analysis ( Figure 5A ) ., These Srs2 protein isoforms are recognized by SUMO-specific antibodies ( Figure 5A ) ., Preliminary characterization of Srs2 sumoylation indicates that none of the well-characterized SUMO ligases , including Siz1 and Siz2 , are involved in this modification ( Figure S1A ) ., Three putative sumoylation sites have been mapped to the C-terminus tail of Srs2 ( Figure 1A ) ., Our data indicated that DNA damage-induced sumoylation of Srs2 was abolished in srs2-3KR mutants , in which the three lysine residues in the motifs identified as modified by SUMO were mutated to arginine ( Figure 5A ) ., Notably , the Srs2-3KR protein can be fully phosphorylated ( Figure S1B ) ., Intriguingly , while sumoylation of native Srs2 is induced at 0 . 3% MMS , the unphosphorylatable Srs2-7AV protein can be detected as SUMO-modified isoforms at ten-fold lower MMS doses ( Figure 5A ) ., Thus , while sumoylation and phosphorylation can occur independently , Srs2 accumulates in a sumoylated form in the absence of phosphorylation ., The biological relevance of Srs2 sumoylation is still obscure , as extensive studies of the phenotypes of the srs2-3KR mutant were inconclusive ( D . Callahan and H . Klein , unpublished results ) ., However , the finding that unphosphorylatable Srs2 is hyper-sumoylated prompted us to ask if the srs2-7AV mutant defects in recombinational repair might be related to Srs2 sumoylation ., To test this , we mutagenized the sumoylation consensus sites in the srs2-7AV mutant to create the srs2-7AV3KR allele , which is simultaneously impaired for phosphorylation and sumoylation ., We then tested the behavior of the srs2-7AV3KR mutant in the DSB repair GC system in which srs2-7AV mutant was highly sensitive ( see Figure 3 ) ., We found that srs2-7AV3KR mutant survived DNA damage ( Figure 5B ) ; DSB repair is accomplished efficiently and a normal level of crossovers is seen in srs2-7AV3KR ( Figure 5B ) ., In addition , the srs2-7AV3KR mutant correctly reversed the checkpoint response after DSB induction and repair , as seen by Rad53 kinase dephosphorylation ( data not shown ) ., Furthermore , the srs2-3KR mutant , which is only impaired in sumoylation , can accomplish DSB repair ( Figure 5B ) ., To see if ablation of Srs2 sumoylation rescues the phosphorylation defects in recombinational repair in other contexts , we crossed the srs2-7AV3KR mutant with the rad27Δ mutant to generate rad27Δ srs2-7AV3KR double mutants ., While the rad27Δ srs2-7AV double mutants never form viable spores ( see Figure 2A ) , we found that rad27Δ srs2-7AV3KR double mutants developed into visible colonies ( 17/25 of total cases analyzed ) , although the double mutant grew very slowly ( Figure 5C ) ., This partial suppression highlights the importance of Srs2 protein modifications when it is likely that more than one lesion is formed ., Taken together , the data in Figure 5 indicate that Srs2 is sumoylated in vivo ., Sumoylation of Srs2 is not required for DSB repair , but the recombinational repair defects in unphosphorylatable srs2-7AV mutants are largely related to the unscheduled sumoylation of the protein ., The sumoylation consensus sites are located in the last 138 residues of the C-terminus tail of Srs2 ( Figure 1A ) , which also mediates the interaction with PCNA 21 ., Hence , we asked if this tail is important for the Cdk1-dependent role of Srs2 ., As shown in Figure 6 , we found that the srs2−ΔC138 mutant is viable after induction of a HO-mediated DSB and also when combined with a rad27Δ ., Conversely , unphosphorylatable srs2-7AV mutants lacking the PCNA-interaction domain ( srs2-7AVΔC138 ) are lethal in both contexts ., These data suggest that Cdk1 targets Srs2 to promote recombinational repair independently of the interaction with PCNA and sumoylation ., Moreover , elimination of sumoylation sites , but not deletion of the Srs2 tail containing the same sites , suppresses the recombination defects in the srs2-7AV mutant ., Recombination can be both prevented and stimulated in srs2 mutants , suggesting a dual role for Srs2 in HR ., The finding that Srs2 is a DNA translocase that antagonizes the formation of unscheduled Rad51 filaments explains certain srs2 phenotypes in HR that are suppressed by ablating RAD51; these include the synthetic lethality with sgs1 mutants or high sensitivity to UV-light 13 , 16 ., Nevertheless , srs2 mutants are defective in Rad51-dependent DSB repair 25 , 28 or lethal when combined with rad27Δ mutants 14 , 41 ., These are contexts in which HR is essential to restore DNA lesions and the activity of Srs2 is required to promote homologous recombinational repair ., In this study we analyzed the recombination phenotypes of two srs2 mutants that mimicked either the constitutive unphosphorylated ( srs2-7AV ) or Cdk1-dependent phosphorylated ( srs2-7DE ) protein isoforms ., We found that srs2-7AV unphosphorylatable mutants display only a subset of srs2Δ phenotypes and , in particular , they do not display those phenotypes that are suppressed by RAD51 deletion ., In fact , srs2-7AV mutants are not UV-sensitive or synthetically lethal with sgs1Δ , but are non-viable when combined with rad27 mutants or treated with the DSB-inducing drug zeocin ., Thus , functions of Srs2 in preventing unscheduled recombination or in allowing efficient recombinational repair are genetically separable ., The phosphorylation of Srs2 is dispensable for the removal of toxic Rad51 nucleofilaments assembled at gaps , while it is essential to promote recombinational repair ., In accordance with the finding that Srs2 phosphorylation is essential to promote recombination , we found that it is also required for Rad51-mediated DSB repair ., In particular , we have been able to show that Srs2 phosphorylation is necessary to complete SDSA in DSB repair ., ChIP data on Rad51 are consistent with the idea that strand invasion is not affected and that Rad51 protein does not persist on the D-loops in srs2Δ or srs2-7AV mutants , although we cannot rule out that presynaptic filament assembly may somehow be affected in the absence of Srs2 or its phosphorylation ., ChIP analysis conducted on Srs2 suggests that the protein is found at DSBs upon strand invasion , thus it is likely loaded at D-loops ., Taken together , these data are consistent with a role of phosphorylated Srs2 in SDSA pathway , but another helicase/translocase may be implicated in removing Rad51 at the D-loops ., We favour the idea that Cdk1 targets Srs2 to dismantle the D-loop intermediate in SDSA ( Figure 7 ) perhaps after DNA synthesis has extended the invading strand ., Srs2 helicase activity might be stimulated by binding to the D-loop structure and/or by interaction with other recombination factors ., ChIP data conducted on unphosphorylatable Srs2-7AV at the invading strand suggest that the mutated protein accumulates at later times and is not rapidly dislodged from DNA as the wild-type protein ., The fact that unphosphorylatable Srs2 appears glued at the D-loops is evocative of a protein working very inefficiently and whose turnover is largely prevented ., It is likely that the unscheduled accumulation of the protein on the DNA might contribute to impaired cell viability and , consistent with this idea , the lethal phenotype of srs2-7AV mutant in response to DSBs is dominant ( Figure S2 ) ., Our data indicate that the proportion of srs2-7AV cells that do not survive DSB repair via GC is higher than the one , which fails to repair DNA lesion ( Figure 3 ) ., This suggests that a fraction of srs2-7AV cells might die because of checkpoint-mediated arrest , as in srs2Δ mutants 25 ., However , Srs2 phosphorylation is not required for recovery during DSB repair by SSA , that is , when Srs2 is probably engaged to remove toxic Rad51-depedent DNA structures , rather than working at the D-loop intermediate 45 ., Thus , the checkpoint recovery defect in srs2-7AV mutants might have different causes during DSB repair by GC or SSA; as described below , perhaps some aspects of recovery defect in srs2 mutants in GC could be explained considering that Cdk1-dependent phosphorylation was no longer required for Srs2 recombination activity , if sumoylation is also prevented ., We found that Srs2 sumoylation can be detected in vivo in response to heavy DNA damage ., Protein modification is prevented ablating three lysine residues located in the extreme C-terminus tail of Srs2 ., Sumoylation and Cdk1-dependent phosphorylation modifications of Srs2 are independent events , but when phosphorylation fails , sumoylated Srs2 accumulates ., There is a functional relationship between these two DNA damage induced modifications , since ablation of sumoylation residues largely rescues the recombinational repair phenotypes of srs2-7AV mutants ., What may be the mechanism for the toxicity of sumoylation in the absence of phosphorylation ?, Sumoylation of Srs2 alone appears unnecessary for many of its recombination functions ( D . Callahan and H . Klein , unpublished results ) ; here we show that it is not essential in DSB repair ( see Figure 5B and Figure 6 ) ., While the biological significance of Srs2 sumoylation waits to be elucidated , we speculate that it might be important for degradation of Srs2 protein ., Srs2 can interact physically with Slx5 39 , that in complex with Slx8 , has been implicated in degradation of sumoylated proteins bound to irreparable DNA breaks at the nuclear periphery 8 , 9 ., Our data suggest that Cdk1-dependent phosphorylation of Srs2 counteracts its sumoylation , which takes over only in response to massive DNA damage ., Thus , in a possible scenario , unphosphorylated and sumoylated Srs2 is trapped at DSB and becomes channeled via the Slx5/Slx8 pathway to the nuclear periphery ( Figure 7 ) ., Since this emergency nuclear periphery pathway intervenes to degrade proteins in response to irreparable DSBs , it might normally act on phosphorylated and sumoylated Srs2 and , therefore , Srs2-7AV cannot be eliminated ., Conversely , after successful DSB repair , phosphorylated Srs2 could be recycled by other routes , and independent of sumoylation ., Intriguingly , the unscheduled Srs2-dependent sequestration of DSBs to the periphery might explain the checkpoint recovery defects in srs2-7AV and perhaps also that of srs2Δ , if we imagine that another unregulated DNA helicase takes over in the absence of Srs2 ., Our studies did not show any obvious alterations in Srs2 protein levels in srs2 phospho-mutants and/or SUMO-mutants ( M . Saponaro and G . Liberi , unpublished results ) , but local protein degradation events at damaged DNA could be relevant ., Elimination of sumoylation compensates for the absence of phosphorylation of Srs2 in DSB repair , but paradoxically this rescue requires the last 138 residues of Srs2 that are not normally necessary for DSB repair ., Hence , this suppression might require interaction with other factors ., Preventing sumoylation in the unphosphorylatable Srs2 rescues recombination defects that ensue after a single DSB , but the importance of these Srs2 modifications become evident when many breaks occur , as in the rad27Δ mutants ., We found that Srs2 phosphorylation is essential for recombinational repair of spontaneous damage occurring during S-phase in rad27Δ mutants ., Similar to the response to DSBs , sumoylation of Srs2 is a main cause of death in srs2-7AV phospho-mutants ., It is more difficult to predict the kind of damage which requires phosphorylated Srs2 in rad27Δ mutants ., Rad27 is required for Okazaki DNA fragment processing 48 and in its absence , Srs2 might dismantle DNA and/or RNA structures that block HR ., In any case , based on our conclusion that Srs2 phosphorylation is not essential for the processing of toxic Rad51 filaments , we think it more probable that the helicase activity , rather than translocase activity , is crucial for the survival in rad27Δ mutants ., This proposed role of phosphorylated Srs2 in replication might seem at odds with the role suggested for Srs2 in preventing recombinational repair during S-phase through recruitment by sumoylated PCNA 20 , 21 ., However , in PRR mutants , Srs2 is proposed to be recruited by PCNA to disrupt Rad51 filaments at DNA gaps , while in the absence of Rad27 , we are considering that Srs2 acts in a PCNA-independent and phosphorylation-dependent role as a helicase , rather than as a translocase ., Importantly , srs2-7DE mutants slightly suppress the MMS sensitivity of PRR mutants , suggesting that the phosphorylated Srs2 is less efficient as a DNA translocase than the non-phosphorylated isoform ., This is unmasked in PRR mutants , where it is likely that many sDNA breaks occur ., Srs2 phosphorylation might modulate its interaction with PCNA , a hypothesis that will be interesting to test in the future ., Our data indicate that Srs2 is a new target of Cdk1 kinase in DSB repair , acting at the level of strand invasion , rather than during DNA end resection ., Srs2 phosphorylation is required to limit the extent of DNA exchanges during DSB repair with a function that is genetically separable from its role in processing toxic Rad51 filaments ., We suggest that Cdk1-mediated phosphorylation might control , throughout the interaction with PCNA and/or other factors , the ability of Srs2 to function as a translocase or a helicase that inhibits or allows HR depending on the context ., Furthermore , our data unravel a novel aspect of Cdk1-dependent regulation in counteracting untimely sumoylation events , which might become toxic for recombination if not properly scheduled .

Introduction, Results, Discussion, Materials and Methods

Cdk1 kinase phosphorylates budding yeast Srs2 , a member of UvrD protein family , displays both DNA translocation and DNA unwinding activities in vitro ., Srs2 prevents homologous recombination by dismantling Rad51 filaments and is also required for double-strand break ( DSB ) repair ., Here we examine the biological significance of Cdk1-dependent phosphorylation of Srs2 , using mutants that constitutively express the phosphorylated or unphosphorylated protein isoforms ., We found that Cdk1 targets Srs2 to repair DSB and , in particular , to complete synthesis-dependent strand annealing , likely controlling the disassembly of a D-loop intermediate ., Cdk1-dependent phosphorylation controls turnover of Srs2 at the invading strand; and , in absence of this modification , the turnover of Rad51 is not affected ., Further analysis of the recombination phenotypes of the srs2 phospho-mutants showed that Srs2 phosphorylation is not required for the removal of toxic Rad51 nucleofilaments , although it is essential for cell survival , when DNA breaks are channeled into homologous recombinational repair ., Cdk1-targeted Srs2 displays a PCNA–independent role and appears to have an attenuated ability to inhibit recombination ., Finally , the recombination defects of unphosphorylatable Srs2 are primarily due to unscheduled accumulation of the Srs2 protein in a sumoylated form ., Thus , the Srs2 anti-recombination function in removing toxic Rad51 filaments is genetically separable from its role in promoting recombinational repair , which depends exclusively on Cdk1-dependent phosphorylation ., We suggest that Cdk1 kinase counteracts unscheduled sumoylation of Srs2 and targets Srs2 to dismantle specific DNA structures , such as the D-loops , in a helicase-dependent manner during homologous recombinational repair .

Broken DNA molecules can be repaired by copying a homologous DNA sequence located elsewhere in the genome ., This process , called homologous recombination , needs to be carefully regulated , because unwanted DNA exchanges can lead to genome rearrangements and cell death ., Cdk1 kinase is required for cell cycle progression and phosphorylates DNA repair factors , such as Srs2 , a protein that can both translocate on single-stranded DNA and open the two strands of DNA double helix ., DNA translocation activity of Srs2 is crucial to prevent unwanted recombination , while DNA unwinding activity might be important to promote recombination ., In this study , we used two srs2 mutants that constitutively express the unphosphorylated or Cdk1-dependent phosphorylated Srs2 protein isoforms ., We found that Srs2 performs genetically distinct functions in preventing or promoting homologous recombination ., Cdk1 targets Srs2 to promote accurate repair of double-stranded DNA breaks , but is not essential for the removal of toxic recombination intermediates assembled at single-stranded DNA breaks ., Further , Cdk1 counteracts sumoylation of Srs2 , which is responsible for recombination defects due to the lack of Srs2 phosphorylation ., In summary , Cdk1-dependent Srs2 phosphorylation prevents its unscheduled sumoylation and targets the helicase to promote accurate homologous recombinational repair .

molecular biology/recombination, genetics and genomics/gene function, molecular biology/dna repair

journal.pcbi.1004605

Computer Simulations Imply Forelimb-Dominated Underwater Flight in Plesiosaurs

Plesiosaurians ( = plesiosaurs ) are an extinct group of highly derived predatory marine reptiles with a global distribution that spans 135 million years from the Early Jurassic to the Late Cretaceous ., During their long evolutionary history 1 , plesiosaurs maintained a unique body plan with two pairs of large wing-like flippers—a unique adaptation in the animal Kingdom 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 ., Although plesiosaurs were a key component of Mesozoic marine ecosystems , there are no extant ‘four-winged’ analogues to provide insights into their behavior or ecology , and their locomotion has remained a topic of debate since the first complete plesiosaur skeleton was described in 1824 10 ., Key areas of controversy have concerned the most efficient biologically possible limb stroke , and how the four limbs moved in relation to each other ., Previous studies of plesiosaur locomotion have endorsed a variety of strokes and gaits ., Stroke hypotheses include a rowing stroke akin to the oar of a boat 4 , a flight stroke 5 , 3 akin to penguins and turtles 11 , 12 , and a modified flight stroke 6 akin to sea lions 13 ., Study of plesiosaurian musculature does not rule out either rowing or flight strokes 14 ., Gait hypotheses include synchronous motion with all four limbs moving in phase 15 , asynchronous motion with the forelimbs and hindlimbs out of phase 12 , 7 , and semi-synchronous motion 3 ., Some authors have proposed that the hindlimbs provided most of the propulsion 16 , whereas others suggest that the forelimbs provided the majority of thrust 8 ., The question has been approached experimentally using robotics 17 and human swimmers with fabricated paddles 3 ., These studies , although informative , are limited because they do not deal with accurate representations of the plesiosaur form ., There is therefore still no consensus on how plesiosaurs swam , especially how they moved all four limbs relative to each other ., Our approach uses computer simulation to address the question of how plesiosaurs swam using a three-dimensional plesiosaur model in a simulated fluid ., The computational model explores a given range of joint motion to discover the swimming stroke and gait that propels the creature forward the greatest distance ., There are two main advantages to using computer simulation to investigate swimming motions of plesiosaurs ., First , we can construct a digital representation of the body that accurately matches the known body and limb shapes of a particular species ., Second , we can run thousands of trials with different strokes and gaits to explore the space of possible swimming motions ., We use an optimization method to search for the highest quality motion for a given range of joint angles ., Computer simulation has been used to investigate the motions of several types of modern-day swimmers , including fish 18–21 , tadpoles 22 , and copepods 23 ., To our knowledge , our work is the first use of computer simulation to study the swimming of plesiosaurs ., We constructed a life-sized plesiosaur model based on Meyerasaurus victor , a small ( 3 . 35 meters long ) taxon from the Lower Jurassic of Germany , because it is known from an almost complete articulated skeleton ( SMNS 12478 ) with all four limbs preserved in their entirety 24 ( Fig 1A ) ., In addition , Meyerasaurus possesses a generalized morphotype among plesiosaurs , with a moderately long neck , so it can be considered representative of the clade Plesiosauria as a whole , which contains long- and short-necked morphotypes 2 ., The shape and proportions of the model were based directly on the skeleton , with three-dimensional data and soft tissues ( e . g . muscles , cartilage , integument ) reconstructed based on evidence from other taxa ., Two dimensional cross-sectional data for the body was estimated by tracing around the skeleton in the horizontal plane ., Since the holotype of Meyerasaurus is dorsoventrally compressed , information from other plesiosaur specimens was used to estimate cross sections in the vertical plane 25 ( Fig 1B ) ., Transverse cross-sections of the limbs were derived from the three-dimensionally preserved propodials of the closely related Rhomaleosaurus thorntoni 26 ., The restored limbs are cambered hydrofoils in section with a narrower postaxial trailing edge ., The postaxial edge of the limb was extended beyond the osteological anatomy to reflect fossil evidence for a soft tissue trailing edge in this region in Seeleysaurus guilelmiimperatoris 27 and Hydrorion brachypterygius 28 ., The tail of the model was reconstructed with a short dorsally expanded mediolaterally compressed fin based on evidence from several plesiosaurian taxa , including the sister taxon to Meyerasaurus: Rhomaleosaurus 29 ., The virtual model was constructed in Maya , a widely used CAD tool ., First , the two-dimensional cross sections were aligned in the horizontal and vertical planes ( Fig 1B ) ., Second , the model was constructed using these sections as a reference ., Since our grid-based fluid simulator cannot detect features under 67 mm thick , the thinnest parts of the anatomy , such as the trailing edge of the limbs , were artificially dorsoventrally thickened ., For simplicity , the density of the plesiosaur model is identical to that of the fluid ( i . e . it is neutrally buoyant ) , and therefore our simulations do not take into account possible variation in buoyancy along the body of the animal due to air-filled lungs , or gastroliths 30 ., The life size final constructed plesiosaur model is 3 . 35 meters long from head to tail ( Fig 1C ) ., An alternative bulkier Meyerasaurus model with 50% greater soft-tissue mass around the base of the limbs was also created to test the effect of a bulkier body outline ., Our physics simulator ( based on previous methods 31 ) represents the plesiosaur as a collection of rigid body parts that meet at points of articulation ., Specifically , we model the body ( torso , neck , and tail ) as one rigid component and the four limbs as additional rigid parts ., In life , the plesiosaur torso was a rigid structure , since a sturdy trunk is a prerequisite for purely paraxial underwater locomotion ., The neck and tail in plesiosaurs were flexible in life to variable degrees 32 , so they may have had a relatively minor role in propulsive locomotion ., However , since our focus is on the question of limb-based propulsion in a four-winged paraxial swimmer , we kept the neck and tail immobile in the simulation to allow us to focus solely on the movement of the limbs ., Each limb is joined to the body by a three-degree-of-freedom joint ( Fig 2 ) ., These four joints can be actuated internally to generate motion ., Each swimming motion is represented as a sinusoidal function at each joint degree of freedom , and the actuators track these desired motions by applying torques at the joints ., In turn , the motion of the body and limbs affects the simulated fluid that surrounds the animal ., Our simulator resolves the motion of the animal body and the fluid simultaneously , so that the final motion is due to the interaction between the body and the fluid ., This is in contrast to studio-created computer animation , where the motion of the animal through the fluid is scripted by an artist , and may not obey the governing laws of physics ., To study the forward swimming motion across a wide range of periodic swimming strokes of the limbs , we require a motion representation that is expressive but that is also biologically plausible ., We decouple the degrees of freedom at a joint into a dorsal/ventral component , an anterior/posterior component , and a pronate/supinate component ( the rotational angle of the limb ) ., We specify the limb motions of the plesiosaur by describing sinusoidal patterns for each of the three degrees of freedom at a given joint ., The limb motion for a given component is specified by three values: the minimum and maximum value of the sinusoid , and the phase of the sinusoid ., We use the same frequency ( 0 . 5 Hz ) for all of the sinusoids across all of the limbs ., Since the motion for each degree of freedom is given by three values ( maximum and minimum range , and phase ) , nine numbers fully describe the motion of a single limb ., We assume that the left and right limbs move in synchrony while the animal is swimming straight , as is the case for penguins , sea lions , marine turtles , and nothosaurs 33 ., However , we specifically allow the front and back limbs to follow different patterns of motion: the minimum and maximum angles , and the phase of the sinusoid for the front and back limbs can be set differently ., This allows us to test , for instance , the possibility that the front and back limbs move together or with offset phases ., To specify both front and back limb sinusoidal motion , we require a total of 18 parameter values ., Table 1 shows the optimized minimum/maximum ranges for all limbs , as well as the average travelling velocity and distance traveled for all of our experiments ., Although using sinusoidal motions of various angle ranges and phases gives a wide range of possible swimming strokes , the motions of modern-day swimming animals depart from pure sinusoidal motion in at least two ways ., Animals such as penguins that use an underwater flight stroke 12 hold the angle of rotation steady during the downstroke , quickly rotate the limb at the bottom of the stroke , and then hold the angle of rotation steady again during the upstroke ., We allow for this possibility by using one additional degree of freedom for the rotation of a limb that specifies the duration of a motionless interval during which the limb maintains a zero rotational velocity ., This interval can be set to zero , which indicates pure sinusoidal motion , or it can be non-zero to hold the angle of rotation steady through a portion of the stroke ( Fig 3A ) ., This gives us two additional motion parameters , one for the forelimbs and one for the hindlimbs ., We also allow for the possibility that the animal’s downstroke and upstroke take different amounts of time ( Fig 3B ) ., This is in recognition of the observation that plesiosaurs may have had stronger musculature governing the downstroke of their limbs 7 , 9 , 14 ., We add one more degree of freedom for each sinusoid to specify its degree of time asymmetry ., This gives us six additional motion degrees of freedom , bringing the total number of parameters that describe a periodic swimming motion to 26 ., Table 2 shows the time asymmetry as well as motionless portion in pronate/supinate direction optimized in all of our experiments ., Plesiosaur limbs contain a single mobile joint located between the propodial and the girdle: the glenohumeral joint in the forelimb and the acetabulum-femoral joint in the hindlimb ., The articulation points for these joints in the model are located in the anatomically-correct positions ( Fig 2 ) ., Neutral limb positions were derived from existing estimates for Plesiosaurus sp ., 3 ., In the forelimb the neutral position is -15 degrees from the horizontal and -16 degrees from a line drawn perpendicular to the long axis of the body ., In the hindlimb the neutral position is -30 degrees from horizontal and -27 degrees from a line drawn perpendicular to the long axis of the body ( Fig 2 ) ., In forelimb only and hindlimb only optimizations , the static limbs are locked into these neutral positions and the active limbs are initiated in the fully abducted positions ., While the available range of rotation along the long axis of the limb was identical in all limbs and optimizations: up to 30 degrees supination and 45 degrees pronation ( Fig 2C ) , we tested three different ranges of motions in the dorsal/ventral and anterior/posterior directions ( Fig 2A–B , D ) ., The degree of freedom in the joints of living plesiosaurs was dependent on the extent and thickness of the cartilage that covered the head of the propodials and lined the glenoid and acetabulum ., To account for possible differences in cartilage thickness and to investigate stroke efficiency and gait under different specified parameters , we performed optimizations under three different ranges of joint freedom: ‘narrow’ , ‘medium’ , and ‘wide’ ( Fig 2 ) ., The narrow range was taken directly from conservative estimates of degrees of freedom in Plesiosaurus sp ., 3 ., The wide range provides an expanded degree of freedom that possibly exceeds the biologically possible range in the living animal , and the medium range represents a realistic compromise between the conservative narrow range and generous wide range ., In life , rotation of these joints was complicated , but for simplicity , they pivot around a single point in the model ., Although there are no additional mobile joints in plesiosaur limbs , cartilage and tendons would have allowed dorsoventral flexibility and twisting along the long axis of the limb 5 ., This could have resulted in an increased range of motion at the tip of the limb compared to the range of motion at the joint , and may have affected water flow and minimized drag ., One limitation of our method is that it does not currently replicate flexibility of this kind—the limbs are rigid elements in our simulations ., We accounted for this , in part , by providing simulations with wider ranges ., To fully address limb flexibility would require an entirely different simulator that uses the finite element method to allow limb deformations ., This would also require a different approach to solid/fluid coupling in the simulator ., Since a single swimming motion requires the specification of many different sinusoidal parameters ( 26 , as described above ) , we use numerical optimization to explore the range of possible plesiosaur swimming motions ., Specifically , we use the sample-based method called Covariance Matrix Adaptation 34 which has been used to investigate eel swimming 19 and animal walking gaits 35 ., Our optimization process runs several thousand different simulations with different joint motions , narrowing in on the set of motions that produces the fastest swimming motion ., Note that for such a large parameter space , CMA is not guaranteed to find the global optimum ., However , we observed only small variations in the final results of different CMA runs with the same parameter settings ., Fig 4 shows the resulting optimal swimming motions for each of the three joint ranges , where both the front and back limbs move in a manner that best propels the plesiosaur forward ., The white paths in the figure show the distal tip traces ., ( See the two accompanying S1 and S2 Videos for the detailed motions . ), The best strokes for the forelimbs in both the narrow and medium range is an underwater flight motion , in which the limbs move primarily in the dorsoventral direction , and only rotate at the top and bottom of the stroke ( Fig 4A and 4B ) ., This pure flight stroke has been suggested as the most likely swimming stroke for plesiosaurs based on several anatomical lines of evidence 5 ., In contrast , our optimization determined that the best forelimb stroke for the large range of joint motion is a modified U-shaped flying stroke ( Fig 4C ) ., This is similar to a flight stroke , but with more posterior motion during the power stroke and with a partially feathered recovery , and has also been proposed for plesiosaurs 6 , 15 , 8 ., Note , however , that our optimizations suggest that this modified flight stroke is only plausible under the most liberal of joint range assumptions , and such a wide range of motion is considered biologically impossible 3 ., None of our optimizations produced a substantial rowing stroke , as had been suggested by early researchers 4 ., S1 Video shows the highest quality swimming motions from each of our optimization runs ., To produce each of these video segments , we re-computed the simulation that corresponded to the highest quality motion sample that was found during the given optimization run ., There are nine motion clips in this video , corresponding to the wide , medium and narrow ranges of limb joints , with motion from both pairs of limbs , just the forelimbs , and just the hindlimbs ., As in the optimization runs , the model plesiosaur is initially at rest and then begins to move ., During the first stroke , the model plesiosaur sometimes turns upwards , but then moves straight during subsequent strokes ., For this reason we do not include the motion of the first stroke in our quality assessment ( described in detail later ) ., The simulations used to make these videos not only provide the motion of the plesiosaur model , but also give us the velocity field for the simulated fluid at each simulation time step ., This allows us to show not only the plesiosaur motion , but also the accompanying motion of the water ., In each video clip , we show the motion of the fluid using particle traces ( particle trajectories over the last few time steps ) from randomly positioned particles in the virtual fluid ., These massless particles are passively advected through the fluid ., The particle traces are drawn only at locations with large vorticity ( greater than 1 . 5 s-1 ) to concentrate them at regions of interest ., S2 Video shows the plesiosaur limb stroke motions from these same nine simulations , and provide traces of the tips of each moving limb ., These motion clips allow a clearer picture of the limb motions relative to the body ., In these video clips , the camera moves together with the body so that the body appears to be stationary ., Two views are provided for each motion: a lateral view and a posterolateral view ., For clarity , particle traces have been omitted in these video clips ., Fig 5 shows the best swimming speeds for each joint range ., These swimming speeds are similar to prior estimates of an optimum swimming speed of 0 . 48 m/s for Meyerasaurus 36 ., Because all of the strokes have a two second period , it is to be expected that the larger ranges of motion result in a faster speed , so no conclusions should be drawn from the relative speeds between the three different ranges of joint motion ., There are several potential relative motions between the forelimbs and hindlimbs during swimming ., One proposal is that asynchronous motion is the most likely way to produce continuous forward motion 9 ., Synchronous ( and semi-synchronous ) motion has been deemed more likely by other researchers 15 , 8 , 3 ., Our optimizations provided no clear answer to this question , which is significant in itself ., In the narrow range the limbs move asynchronously , in the medium range they move semi-synchronously , and in the wide range they move synchronously ., In order to deduce the separate contributions of the forelimbs and hindlimbs to propulsion we performed optimizations with forelimb-only motion and hindlimb-only motion ., The results were strikingly consistent across all three joint ranges ., The forelimb-only strokes from optimization were roughly as fast as the best gaits resulting from optimizations using all four limbs ( Fig 5 ) ., In contrast , the hindlimb-only strokes from optimization provide a much slower forward motion , even for the widest joint range ., This inability of the hindlimbs to generate thrust explains the lack of consistency in the relative motions of the forelimbs and hindlimbs in our optimization results ., It does not matter whether the plesiosaur moves its hindlimbs in or out of phase with the forelimbs , since neither strategy will contribute substantially to forward motion ., Our optimization results imply that plesiosaurs were forelimb-dominated swimmers , and that the hindlimbs contributed little to their forward motion ., This is consistent with trace-fossil evidence for forelimb-dominated locomotion in nothosaurs 33 ., It also corroborates other studies that concluded that the hindlimbs were used primarily for steering and stabilization during swimming 8 , and that two-flippered gaits serve well for low-cost cruising 17 ., The plesiosaur hindlimbs , despite their wing-like shape and large size , played a diverse role in locomotion , but a relatively minor role in propulsion ., They supplemented the forelimbs , which were the primary propulsive organs , by enhancing maneuverability and stability , possibly in conjunction with the tail 29 ., To investigate the effect of small body changes to swimming speed , we constructed an alternative version of the Meyerasaurus model with 50% greater ‘muscle mass’ around the bases of the limbs ., Fig 6 shows the original and modified body shapes ., We ran three simulations with the bulkier model using the limb motion parameters taken from the results of the optimizations with the ‘slimmer’ model ( medium—all limbs , medium—forelimbs only , medium—hindlimbs only ) ., Because it is unlikely that substantially different limb motion parameters would give a more effective swimming motion , we did not use optimization to search for new motion parameters ., Table 3 shows a comparison of the original ( slim ) and modified ( muscle bulk ) model , and for each simulation case gives the swimming distance and vertical deviation from the horizontal ., The results showed that the bulkier model was marginally slower in each case , probably due to the increased drag ., With both models , however , the contribution of the hindlimbs to locomotion is small ., This test shows that manipulation of the fine details of the model does not have a major impact on the swimming speed ., Larger modifications to the body shape could have a more substantial effect , and is an area for future work ., With larger changes to the body shape , it would be necessary to use optimization to search for the most effective limb motions ., There is great variation in head and neck proportions , flipper aspect ratios and relative limb proportions within the clade 37 ., For example , in Meyerasaurus and other Lower Jurassic plesiosaurs the fore- and hindlimbs are subequal in size , whereas in derived pliosaurids the hindlimbs are larger than the forelimbs , and in derived plesiosauroids the forelimbs are largest ., Furthermore , Meyerasaurus has high aspect ratio flippers , whereas some genera ( e . g . Cryptoclidus ) have low aspect ratio flippers 37 ., Head and neck proportions also vary considerably within Plesiosauria ., ‘Plesiosauromorph’ taxa possess a long neck and small head , while ‘pliosauromorph’ taxa possess a short neck and large head 2 ., The Meyerasaurus model used in our experiments possesses an intermediate morphology with a moderately long neck and moderately large skull , so the results represent a generalized plesiosaur morphotype ., Variations in head and neck size could shift the center of mass relative to that of our model , possibly affecting the relative contributions of the forelimbs and hindlimbs during swimming ., A longer neck and/or a larger head would increase drag and slow the forward motion in a manner similar to our test with increased limb muscle bulk ., It is also possible that the tail contributed to forward thrust during swimming ., Although we conservatively extend our general conclusions for Meyerasaurus to all plesiosaurs , our method could be sensitive to substantial changes in bodily proportions , and different plesiosaurs may have swam in different ways ., Further experimentation is therefore required to assess how bodily variation might affect locomotion in other types of plesiosaurs ., We simulate the fluid dynamics based on Euler equation ( Navier-Stokes equation without the viscosity term ) , since the viscosity of water is negligibly small ., ∂u∂t+u∙∇u+1ρ∇p\xa0=\xa0g, ∇∙u\xa0=\xa00, where u is the velocity of fluids , ρ is the density , p is pressure and g is gravity ., We simulate the fluid using a staggered MAC grid 38 based solver ., We use BFECC 39 to integrate the advection term and use explicit Euler scheme to integrate the gravity force ., We solve the incompressibility term along with the two-way coupling between fluids and solids ( See Section 2 . 3 for details ) ., The dynamics of articulated rigid body systems ( the plesiosaur ) is described by the equations of motion in the generalized coordinate:, M ( q ) q¨+C ( q , q˙ ) =\xa0τint+τext, where q , q˙ and q¨ are positions , velocities and accelerations in the generalized coordinates respectively , M ( q ) is the mass matrix , C is the Coriolis and Centrifugal force , τext are the external generalized forces , including the fluid pressure and gravity , and τint are internal torques exerted by the actuated joints ., Given a reference swimming stroke , we use Stable Proportional-Derivative controllers 40 to compute the internal joint torques τint to closely track the stroke ., We build our two-way coupling solver based on Tan’s two-step procedure 31 ., In the first step , both the fluid and the solid are simulated independently ., In the second step , a linear system of the pressure field is formulated , taking into account both the incompressibility of the fluid and the dynamics of the solid due to the fluid pressure ., Our simulator needs to voxelize the plesiosaur onto the grid at each time step , which marks the grid cells that are inside the animal as solid cells and the remaining as fluid cells ., Due to the large computational requirement of the two-way coupling simulation , we use relatively coarse grid resolution ( 100x80x60 to represent a 6 . 6 by 5 . 28 by 3 . 96 m3 region of water ) , which will cause stair-step boundary artifacts at the interface between the fluid and the animal ., This can result in seemingly higher viscosity near the animal’s body ., For this reason , we incorporate the variational approach 41 into our two-way coupling simulation ., This enables us to perform simulation with sub-grid resolution and gain smoother results at the solid-fluid interface with almost negligible additional cost ( Fig 7 ) ., The traditional fluid simulation enforces the incompressibility by solving the following Poisson equation , with Neumann boundary condition at fluid-solid faces and Dirichlet boundary condition at free surface ., Following Tan’s 31 derivation , at coupled faces ( solid-fluid interface ) , the total generalized force exerted by the fluid pressure to articulated rigid-bodies is an integral of pressure over the surface of the plesiosaur:, τtotal\xa0=\xa0∬Sτ\xa0=\xa0∬SJTpn\xa0=\xa0∑i\xa0=\xa01kJiT ( Δx ) 2pini, where Ji is the Jacobian matrix at the ith fluid-surface interface , Δx is the length of a single grid cell , p is the pressure inside the fluid cell and n is the solid surface normal ., Similar to Batty’s variational approach 41 , we apply divergence theorem to convert the surface integral to the volume integral ., We dropped the second term in the integral because the divergence of J is always zero for rigid body motions , and since p is zero everywhere inside solids and equals to the pressure in fluid cell at fluid-solid interface , we can discretize the above integral to be, τtotal\xa0=\xa0∭VJT ( ∇p ) =\xa0∑i\xa0=\xa01kvi ( Δx ) 3JiT ( piΔx ) ni\xa0=\xa0∑i\xa0=\xa01kvi ( Δx ) 2JiTpini, where vi is fluid volume fraction at the coupled interface ., The acceleration in the generalized coordinate of the coupled interface is, q¨\xa0=\xa0M-1τtotal, where M is the mass matrix of the articulated rigid-body ., Therefore the acceleration for the coupled interface in Cartesian space would be, a\xa0=\xa0nT ( J˙q˙ ) =\xa0nT ( Jq¨+J˙q˙ ) =\xa0nT ( JM−1∭vJT ( ∇p ) +J˙q˙ ) =\xa0nT ( ∑ ki\u2009=\u20091vi ( Δx ) 2JM−1JiTpini+J˙q˙ ), Intuitively , the variational approach adds fluid volume fraction information vi to the original coupled linear equation 31 , making use of volume fraction weighted stencils , instead of being discrete values of zero or one ., To compute the fluid volume fraction vi , we perform inside-outside tests of the plesiosaur model at the center of every cell of a higher resolution grid ., We shoot a ray at the cell center and count the number of intersections ( parity ) between the ray and the model ., Since our model is watertight , if the intersection count is even , the cell is outside of the body of the plesiosaur , and thus occupied by the fluid ., Otherwise , the cell is inside the creature , and thus it is solid ., We implemented a Surface Area Heuristic Kd-Tree 42 to accelerate ray-model intersection performance ., In our implementation , we use a sub-grid resolution of 3x3x3 to compute the fluid volume fraction , achieving 1/27 volume fraction precision in the pressure solve ., Finally , we construct a sparse symmetric positive linear system with the following equations at the fluid-solid faces ,, D ( ∑i\xa0=\xa01kvi ( Δx ) 2JM-1JiTpini ) =\xa0D ( u*Δt-J˙q˙ ), where D is discretized volume-weighted divergence operator and u* is the intermediate velocity field before enforcing incompressibility ., We use normal stencils of Laplacian and Divergence operators at fluid-fluid faces ., We apply Preconditioned Conjugate Gradient solver to solve the system , after which we project the fluid velocity with the following equation, u\xa0=\xa0u*-Δt∇pρ, where u* and u represents the fluid velocity before and after projection ., As a result , we simultaneously enforce incompressibility of the fluids , compute dynamics of the solids , satisfy boundary velocity constraints and achieve smoother fluid flow at the solids-fluids interface ., Note that our simulated water does not behave exactly the same as real water ., Two common issues in numerical fluid simulations , numerical viscosity and voxelization artifacts , affect the accuracy of our two-way coupled simulator ., The numerical errors in the simulation can make the simulated fluid more viscous than its real counterpart ., This is called numerical viscosity 43 ., Even though we dropped the viscosity term from the Navier-Stokes equations and used a higher-order integration scheme , BFECC 39 , the numerical viscosity cannot be eliminated entirely ., Voxelization artifacts are caused by converting the rigid bodies ( the plesiosaur ) into a regular grid of cells for fluid simulation ( Fig 8 ) ., The streamlined body shape of the animal is lost in this voxelization process ., As a result , the animal may swim slower than in real life due to the increased form drag 44 ., The variational framework 41 that we used allows us to simulate at sub-grid resolutions , and thus ameliorates this problem ., However , this issue cannot be completely eliminated ., Given a plausible range of motions for the front and back limbs , we wish to find the swimming motion that propels the animal the farthest distance in a given amount of time ., We also want to favor straight swimming ., Since a single swimming motion requires the specification of 26 parameters we formulate our question in terms of optimization ., Let us call a specific set of swimming motion parameters a motion sample , and we can think each such motion sample s as a single point in a 26 dimensional space ., Given a motion sample , the plesiosaur/fluid simulator generates a swimming motion through the hydrodynamic interaction ., We formulate the quality function q ( s ) that favors a swimming stroke that leads to a longer swimming distance , and we penalize deviation from swimming straight ., We calculate the swimming distance as the displacement of the plesiosaur’s center of mass ( COM ) in one swimming cycle projected onto its initial heading direction ., The deviation from straight swimming is decomposed into two components: Directional deviation is measured by the displacement of COM that is perpendicular to the initial heading direction ., Orientation deviation is measured simply as the orientation change in one swimming cycle ., Since the simulation starts with a static plesiosaur submerged in motionless water , we evaluate the objective function only after the first stroke cycle is completed , when the plesiosaur has reached a steady speed ., Forelimb-only and all limb optimizat

Introduction, Results/Discussion, Methods

Plesiosaurians are an extinct group of highly derived Mesozoic marine reptiles with a global distribution that spans 135 million years from the Early Jurassic to the Late Cretaceous ., During their long evolutionary history they maintained a unique body plan with two pairs of large wing-like flippers , but their locomotion has been a topic of debate for almost 200 years ., Key areas of controversy have concerned the most efficient biologically possible limb stroke , e . g . whether it consisted of rowing , underwater flight , or modified underwater flight , and how the four limbs moved in relation to each other: did they move in or out of phase ?, Previous studies have investigated plesiosaur swimming using a variety of methods , including skeletal analysis , human swimmers , and robotics ., We adopt a novel approach using a digital , three-dimensional , articulated , free-swimming plesiosaur in a simulated fluid ., We generated a large number of simulations under various joint degrees of freedom to investigate how the locomotory repertoire changes under different parameters ., Within the biologically possible range of limb motion , the simulated plesiosaur swims primarily with its forelimbs using an unmodified underwater flight stroke , essentially the same as turtles and penguins ., In contrast , the hindlimbs provide relatively weak thrust in all simulations ., We conclude that plesiosaurs were forelimb-dominated swimmers that used their hind limbs mainly for maneuverability and stability .

Plesiosaurs are an extinct group of Mesozoic marine reptiles with a global distribution that spans 135 million years ., They maintained a unique body plan with two pairs of large wing-like flippers throughout their long evolutionary history , but how plesiosaurs swam has remained a topic of debate for almost 200 years ., We address the question of how plesiosaurs swam using a digital , three-dimensional , free-swimming model of a plesiosaur in a simulated fluid ., We performed thousands of simulations under different parameters to investigate possible plesiosaur swimming patterns ., Our simulations show that the forelimbs provide the majority of thrust , and that the thrust from the hindlimbs is weak ., The plesiosaur swims primarily with its forelimbs using an underwater flight stroke , essentially the same as turtles and penguins .

journal.pgen.1006003

Survival of the Curviest: Noise-Driven Selection for Synergistic Epistasis

Patterns of inheritance suggest that genetic variation plays a central role in the etiology of common , serious diseases ( e . g . , autism , schizophrenia , multiple sclerosis , diabetes ) ., Yet genetic variants found by Genome Wide Association Studies ( GWAS ) typically explain only a small proportion of the heritability of such disorders 1 ., Some have proposed that causative alleles elude discovery because their effect sizes are too small , or because they are not included among the variants that are normally ascertained ., Another proposed reason is epistasis , i . e . the combinatorial action of variants at different genetic loci , such that they contribute significantly to disease only when they appear together in an individual 2–4 ., Epistasis , also referred to as gene-gene interaction , can be exceedingly difficult to detect in GWAS—due to computational as well as statistical challenges 5 , 6—and much debate exists about the extent to which efforts should be made to search for it ., Epistasis is common in the experimental genetics of model organisms , but one cannot simply extrapolate the behaviors of the large-effect alleles favored by experimental geneticists to the small-effect alleles thought to dominate the landscape of standing genetic variation ( the fact that a continuous function can be approximated as linear over a small enough interval implies that , as selection drives the effect sizes of alleles toward zero , non-linear interactions—i . e . epistasis—should become insignificant ) ., Although quantitative-trait-locus mapping in populations of model organisms has verified the existence of substantial epistasis among two 7–10 and even three or more genes at a time 11 , the number of replicated examples of strong epistasis underlying human traits—particularly disease traits—is modest 12 , 13 ., Whether this paucity of examples reflects the methodological difficulties ( e . g . lower statistical power ) inherent in working with human populations , or the fact that the major genetic causes of disease lie elsewhere ( e . g . among rare , large-effect variants ) is one of the most hotly debated questions in human genetics ., Progress toward addressing this question could be enhanced if investigators knew when and where to look for epistatic interactions in the human genome ., One tactic is to use existing mechanistic knowledge—e . g . known biochemical or genetic pathways associated with the physiology disrupted by disease—to narrow the search , thereby improving statistical power 6 ., The drawback to this approach is that , by working only outward from what we know , we sacrifice much of the ability to detect anything radically new ., An alternative approach is suggested by the fact that epistasis , when present , must have evolved , i . e . come into being via the usual forces of mutation , drift , and natural selection ., If we could determine the circumstances under which diseases with a strong epistatic genetic component do and do not evolve , we might be able to use that information as additional prior knowledge in genetic association studies ., The evolution of epistasis has previously been investigated by several groups , who have reached a variety of conclusions about it , not all of which are mutually compatible 14–18 ., The approach in the present study differs from that prior work in that it is focused not on the typical or average behaviors of populations , nor on traits in general , but rather on genetic diseases—by which we mean traits that should be under strong negative selection—and populations in which disease incidence is in the range of many common , serious human diseases , i . e . between 0 . 1% and 10% ., Using both theoretical analysis and evolutionary simulation , we find that the extent to which epistasis explains a disease phenotype depends on the nature of the interaction between gene products ( i . e . the biochemistry ) , the strength of selection on the trait , and especially on the nature and magnitude of stochastic and environmental fluctuations that influence gene function ., To quantify epistasis is to capture the degree to which phenotypes cannot be accounted for by summing the phenotypic effects of variation at multiple genes , but one may approach this task in multiple ways ., Among population geneticists , it is traditional to calculate the proportion of phenotypic variance , σ2 , that can be explained by an additive model , i . e . hpop2=σA2/σ2 , where σA2 represents the maximum variance that a linear model can produce , as determined by linear regression ( Section 1 in S1 Text ) ., The quantity hpop2 is termed the additive , or narrow-sense , heritability , and if non-genetic contributions to variance can be neglected or corrected for ( a situation that we denote by an asterisk in what follows ) , 1−hpop*2 captures what is traditionally termed “statistical epistasis” ., Alternatively , if one assumes that a particular phenotype or set of phenotypes may be treated as reference ( i . e . “wildtype” ) , one may measure epistasis from the standpoint of an individual ., For example , if Z ( g ) represents the phenotype associated with a given genotype , g , then for a phenotype controlled by two loci , we may define, hind* ( x , y|x0 , y0 ) =Z ( x , y0 ) −Z ( x0 , y0 ) +Z ( x0 , y ) −Z ( x0 , y0 ) Z ( x , y ) −Z ( x0 , y0 ) ,, ( 1 ), where hind* ( x , y|x0 , y0 ) represents the proportion of phenotypic difference between genotype x , y and reference genotype x0 , y0 that is due to additive effects ( extension to a greater number of loci is straightforward ) ., The asterisk in ( Eq 1 ) reminds us that this definition neglects non-genetic contributions to the phenotype , generalization to which will be introduced later ., In this case , 1 − hind* measures what is often termed “functional epistasis” ., Unlike statistical epistasis , functional epistasis may be further classified as synergistic ( hind* < 1 ) or antagonistic ( hind* > 1 ) ., Although ( Eq 1 ) is defined in terms of the genotype of a single individual , relative to a single reference , it is straightforward to generalize it to collections of “cases” and “controls” , as we in fact do later ., Thus , the primary difference between functional and statistical epistasis is not that one is population-centered and the other is individual-centered , but rather that functional epistasis focuses on the causes of variation from a pre-specified “wild” or “normal” state , whereas statistical epistasis focuses on the structure of variation within a population overall ., From the standpoint of understanding the molecular causes of human disease , or predicting who will develop disease , functional epistasis is the more relevant quantity , as foreknowledge of a “normal” phenotype , or range of phenotypes , is implicit in the notion of disease ., It is thus the evolution of functional epistasis that we are most concerned with here ., On the other hand , statistical epistasis is closely tied to the mechanics of evolution , because it is hpop*2 that effectively determines how natural selection acts on phenotypic variation 19 ., This makes it relatively straightforward to develop insights about how statistical epistasis evolves , but such insights may be unhelpful with regard to functional epistasis , because the latter can exist in the absence , or near absence , of the former , e . g . 14 , 20–22 ., How this can happen is shown in Fig 1 ., Fig 1A depicts an arbitrary two-dimensional phenotypic landscape ( i . e . allele values for each of two genes are on the independent axes , and phenotype value is on the dependent axis ) and illustrates how hind* ( x , y|x0 , y0 ) simply reflects the nearness of Z ( x , y ) ( top-most point of colored landscape ) to the plane formed by Z ( x0 , y0 ) , Z ( x0 , y ) and Z ( x , y0 ) ( black surface visible directly below Z ( x , y ) ) ., On the other hand , hpop*2 reflects how well the landscape is fit by a plane ( translucent surface in Fig 1B ) , where the fit is weighted by the population distribution ( a sample from one possible population distribution is shown in Fig 1C ) ., In order for any sort of epistasis to occur , the landscape must be non-linear , but a large discrepancy between hind* and hpop*2 can occur at genotypes where the non-linearity is strong while the population density is low ., For example , it is relatively easy to construct scenarios in which nearly all of the phenotypic variation displayed by “diseased” individuals is explained by functional epistasis , even though there is little statistical epistasis in the population as a whole ( Fig 1D ) ., Even if diseased individuals represent as much as 50% of a study population ( as in the artificially selected populations of “cases” and “controls” that are assembled for GWAS ) , statistical epistasis can still be well below the level of functional epistasis of the cases , relative to the controls ( S1A and S1B Fig; section 2 in S1 Text ) ., In our evolutionary simulations , described below , we typically choose a phenotypic threshold between health and disease that assigns ~1% of the population to the latter category , consistent with the ascertainment level of many “common” human disorders ., We thus focus on the questions of when , in the course of evolution , situations should arise in which functional epistasis is large among a subset of individuals of this magnitude ., We began by considering a particularly simple model of non-additive gene-gene interaction—the Limiting Pathway ( LP ) model—that can be applied to a variety of biological processes 23 ., In this model , a trait with value z depends on the rate-limiting value of a number of genetically controlled inputs , e . g . , in the case when the number of inputs is two , z = Minx , y , where x and y represent the inputs from two genes ( Fig 2A ) ., These might represent the expression levels of two different polypeptides that combine to form a multi-subunit protein , the abundance of which is given by z ., Alternatively , x and y could represent rates of synthesis of small molecules , such as enzyme cofactors , that catalyze a process occurring at rate z ., For systems involving more than two inputs , we have z = Minx1 , x2 , x3… , but for simplicity we will focus here on the two-component model ., We assume there is an optimal phenotype value , zopt , and that the trait quantified by z is under stabilizing selection , so that fitness is reduced when z falls above or below zopt ., Thus , at evolutionary equilibrium , we expect the wild-type levels of the gene products , which we will call x0 and y0 , to be such that Minx0 , y0 = zopt ., The representation of this condition on the phenotypic landscape , i . e . the intersection of the plane z = zopt with the function z = Minx , y , defines two perpendicular “arms” , meeting at a “corner” ( green dashed lines in Fig 2A ) ., Only when populations reside near this corner , i . e . , x0 ≈ y0 , can significant epistasis occur ., For example , under such conditions mutations that increase x or y may have little phenotypic effect individually , but a large effect in combination ( synergistic epistasis ) ., In contrast , for a population residing on one of the arms—e . g . , x0>>y0 , corresponding to the horizontal arm in Fig 2A—mutations that increase or modestly decrease x will have no phenotypic effect; mutations that affect y will alter z proportionately; and the combined effect of both mutations will be no different from the sum of their individual effects ., For any pair of wild-type input values ( x0 , y0 ) and mutation effect sizes ( Δx , Δy ) , we may quantify the magnitude of functional epistasis as 1 –hind* ( x0 + Δx0 , y0 + Δy0 | x0 , y0 ) , according to ( Eq 1 ) , as described above ., For a given mutation effect ( Δx = Δy > 0 ) , we repeat this calculation for different combinations of wild-type values x0 and y0 , to produce an epistasis map ( Fig 2B ) ., Asking whether epistasis will tend to evolve in the LP model thus amounts to asking whether input values in natural populations will evolve towards and remain at the corner in Fig 2A ., Re-plotting Fig 2A in terms of fitness , which we take to be the following generic function of phenotype value, w ( z ) =exp ( −s ( lnzzopt ) 2 ) ,, ( 2 ), ( where the strength of stabilizing selection is quantified by s ) yields an L-shaped ridge on which any point along the ridge is equally and maximally fit ( Fig 2C ) ., Given this fitness landscape , our naïve expectation might be that , under selection and drift , populations should follow a random walk along the ridge , with individual input values equally likely to be clustered around any ridge location ., However , evolutionary simulations ( see Methods ) show that selection drives populations away from the corner and out onto the arms of the ridge ( Fig 2D ) ., To understand why , note that at “arm” locations where y << x , mutations affecting x will be selectively neutral ., But if the population is close to the corner , such that y < x < y + σμ , where σμ is the typical mutational step size , mutations that reduce x will be deleterious ., Because of the increased frequency of deleterious mutations near the corner , natural selection disfavors genotypes with input values within σμ of the corner ( compare shaded regions at long times with horizontal dashed lines in Fig 2D ) ., It follows that the population is least likely to be found precisely where epistasis is strongest ., Is the evolutionary instability of epistasis in the LP model something common to all genotype-phenotype relationships in which strong synergistic epistasis can arise ?, Or is the LP model , as currently formulated , insufficiently general ?, To approach this question , we explored two modifications of the LP model ., First , we generalized the phenotype function with the aid of parameter k , which allows us to vary the strength of interaction between pathways:, z=12 ( k+x+y− ( k+x+y ) 2−4xy ), ( 3 ), When k = 0 , ( Eq 3 ) reduces to the simplified LP model , i . e . z = Minx , y ., If x and y were to represent the concentrations of two reactants , and z the concentration of their product , then k would represent the corresponding biochemical dissociation constant ., In this interpretation , if we choose our units so that zopt = 1 , k >> 1 would correspond to weak binding while k << 1 would be tight binding ., We may also interpret x and y in ( Eq 3 ) as rates of synthesis of gene products , though the physiological meaning of k would then be different , incorporating binding and rates of degradation ., As shown in Fig 3A and 3B , the effect of increasing k is to make the corner in the fitness landscape more round ., This increases ( for any given mutational step size ) the range of values of x and y over which significant epistasis will occur , but also decreases the maximum potential magnitude of that epistasis ., Second , we added phenotypic noise into the model ., By noise we mean processes that add randomness into the genotype-phenotype relationship ., For purposes of this analysis , we divide noise into two categories: developmental and environmental ., By developmental noise , we mean processes that affect each individual in the population independently and are not substantially correlated between parents and offspring ( Fig 3C ) ., An example of developmental noise would be idiosyncratic variation in the biochemical processes underlying development , such that individuals with identical genotypes do not necessarily display identical phenotypes ., Other factors that could contribute to developmental noise by this definition would include micro-environmental effects that vary from individual to individual and genetic variation at loci not explicitly incorporated into the model ( assuming high recombination ) ., By environmental noise , we mean processes that have a coordinated effect on all individuals in a population , and may persist from one generation to the next ( Fig 3D ) ., Processes like climate change , dietary change , and cultural change , which have the potential to fluctuate on time scales of many generations , would fall into this category ., Although it is common to model such processes as perturbations to the fitness ( i . e . , “fluctuating selection” , e . g . , 24 ) , we model these processes as perturbations to phenotype , in order to facilitate comparison between the two classes of noise ., Accordingly , we take noise generally to be a process that transforms an individual’s nominal input values ( the values of x and y dictated by the individual’s genotype; white circles in Fig 3C and 3D; S2A and S2C Fig ) into an effective set of input values ( red circles in Fig 3C and 3D; S2B and S2D Fig ) that produce , via ( Eq 3 ) , an altered phenotype ., The effective input values can be thought of as the set of values that would produce the observed phenotype in the absence of any noise ., Motivated by the fact that cell-to-cell variation in the levels of gene products is often found to fit log-normal distributions 25 , 26 , we implement developmental noise in x , say , via the transformation, lnxe=lnx+Δx, ( 4 ), where Δx is drawn from a normal distribution with mean zero and variance σdev2 ( Fig 3C; inset ) ., Perturbations to y are performed independently using the analogous transformation ., With environmental noise , the same perturbation affects the x- and y-values of all individuals in the population ( Fig 3D; inset ) ., Although that perturbation is also drawn randomly ( this time from a zero-mean Gaussian with standard deviation σenv ) , it is not re-drawn at random every generation , as environmental effects may vary on slow time scales ., Details of how this is modeled are provided in the Methods section , but for all results presented in the main text the Δx and Δy values for environmental noise have an autocorrelation time of 27 generations ., How should we expect phenotypic noise to affect population dynamics on the generalized LP landscape ?, So long as the autocorrelation time of phenotypic noise is short relative to the timescale on which natural selection drives genetic change , we can approximate the effect of natural selection on a particular combination of input values by averaging over their different fitness realizations due to noise ., The resulting “effective fitness” measures how robust the phenotype ( and nominal fitness ) of that particular set of input values is to random variation of their values ., When we compute this effective fitness landscape , we find a narrow peak at the corner for k = 0 ( Fig 4A ) that develops into a broad peak at k = 0 . 1 ( Fig 4B ) ., Though a peak at the corner is intuitively expected to localize the population there , it is less clear how the shape of the peak affects localization ., Since the motion of the population is driven by fitness differences , a narrow peak ( and a steep fitness gradient ) should exert a strong force driving the population towards the corner ., Yet that force can only be felt in the immediate vicinity of the corner , potentially allowing populations located elsewhere to drift away from the corner ., On the other hand , though a broad peak exerts a weaker force ( due to its shallower fitness gradient ) , its range of influence is greater , potentially drawing populations to the corner that would have otherwise escaped it ., To find out which landscape best stabilizes the corner at steady state , we mapped the evolution of a population on the 2D effective fitness landscape to a 1D random walk problem ( Section 3 of S1 Text ) ., The random-walk theory predicts that localization at the corner depends monotonically on k , with the greatest degree of corner localization expected at the smallest values of k ( Fig 4D ) —precisely the conditions under which epistasis is strongest ( Fig 3A ) ., Assuming that k is low enough , we next asked how corner localization depends on the noise level , by which we mean the width of the Gaussian from which the phenotypic perturbation is drawn ( σdev or σenv; Fig 3C and 3D ) ., When the noise level drops below 0 . 1 , the random walk is effectively unbiased , with the effective fitness profile being approximately flat ( Fig 4E ) and the probability of finding the walker at least a given distance from the corner falling linearly with distance ( Fig 4F ) ., The random-walk theory , though providing significant insight , is an approximation that is expected to hold only when the product of the population size and the mutation rate is small , i . e . Nμ << 1 ., To address more general conditions , we examined the population dynamics using evolutionary simulations ., Time courses plotted from individual simulations show that both developmental and environmental noise are effective at localizing the population in proximity of the corner ( where epistasis can arise ) when the interaction parameter , k , is small enough ( Fig 5A and 5B; see also S2E and S2G Fig ) ., To summarize the degree of corner localization under a given set of parameter values , we sampled simulations at intervals of 2N generations ( where N is population size ) and computed the fraction of simulations where the population is at least a given distance from the corner ( Fig 5C and 5D ) ., For both developmental and environmental noise , simulated data confirm the general trend predicted by the random-walk theory: the sharper the corner ( and the greater the potential for strong epistasis ) , the more effective phenotypic noise is at localizing the population there ( see also S4 Fig ) ., We repeated the evolutionary simulations for a range of noise levels , either pure developmental or pure environmental ., Fig 5E shows that the dependence of corner localization upon developmental noise is remarkably similar to that predicted by the random-walk theory ( Fig 4F ) ., However , when the phenotypic noise is environmental , the results of theory and simulation diverge ., Evolutionary simulations show that corner localization depends non-monotonically on environmental noise level such that only intermediate noise levels ( σenv ≈ 0 . 1 ) hold the population near the corner ( Fig 5F ) ., In contrast , the theoretical dependence on noise level was always monotonic ( Fig 4F ) ., The accuracy of random-walk theory in predicting the role of developmental but not environmental noise in simulations reflects the fact that the theory takes as input an effective fitness landscape representing the average effect of phenotypic noise ., While such averaging is expected to faithfully capture pure developmental noise , which is independently realized for each individual and at each generation ( Fig 3C ) , this approximation does not account for the correlations ( among individuals and across generations ) intrinsic to our implementation of environmental noise ( Fig 3D ) ., In summary , both theory and simulation show that phenotypic noise drives populations to the corner of the generalized LP fitness landscape , with the greatest degree of corner localization expected at the smallest values of k—the same condition under which the genotype-phenotype map is most nonlinear ( Fig 3A ) ., Though such nonlinearity is clearly a necessary condition for generating epistasis in individuals of the population , it is certainly not sufficient ., To shed light on human disease biology , we next engaged the full power of our simulations to determine the conditions under which individuals affected by disease exhibit epistasis relative to healthy controls , and the frequencies at which such individuals may arise ., We modeled case-control studies of common disease ( prevalence ≈ 1% ) by sampling individuals with extreme and typical phenotype values ( Fig 6A; S2 Fig ) , which we denoted “cases” and “controls” , respectively ., For each case individual , we partitioned the difference between its phenotype value , zcase , and the phenotype associated with the median input values of the control samples , zcontrols , into a fraction corresponding to the difference in nominal input values ,, Hind=zH−zcontrolszcase−zcontrols ,, ( 5 ), which we refer to as the “heritability” , and a non-heritable fraction , 1-Hind ., The quantity zH represents the heritable phenotype of the case individual , which we approximated by its phenotype value before developmental noise was added ( Fig 6B ) ., In contrast , environmental noise is not expected to affect phenotype heritability assessed either within a generation ( since genetically identical individuals have the same phenotype even after the addition of environmental noise ) or between two consecutive generations ( since environmental noise persists over many generations ) ., Note that Hind , closely resembles the concept of broad-sense heritability as used by human geneticists , but is defined here at the level of a single individual , rather than the population as a whole ., We can naturally partition Hind into an additive fraction ,, hind= ( zx−zcontrols ) + ( zy−zcontrols ) zcase−zcontrols ,, ( 6 ), and an “epistatic fraction” , Hind—hind , where zx and zy represent the phenotype values of the single mutants defined by the heritable input values of the case individual ( Fig 6B ) ., Epistatic fraction is a generalized measure of functional epistasis that is valid in the presence of phenotypic noise ., To see this , note that , in the absence of noise , Hind reduces to 1 and hind reduces to hind* ( x , y | x0 , y0 ) , where hind* is defined by ( Eq 1 ) and the input values ( x , y , etc ) are determined by the relations Z ( x , y ) = zH = zcase and Z ( x0 , y0 ) = zcontrols ., Thus Hind—hind reduces to 1—hind* , which is the definition of functional epistasis presented earlier ., The no-noise limit is instructive for another reason: it turns out that hind* evaluated at one population standard deviation relative to the median control sample directly determines the narrow-sense heritability used in human genetics , hpop*2 ( S1C Fig; see also Section 1 in S1 Text ) ., We measured the epistatic fraction of case-control phenotype difference , Hind—hind , and the fitness ( given by ( Eq 2 ) , normalized by the mean fitness of the control samples ) for each case individual and for each time point sampled during the course of our evolutionary simulations ., Fig 6C ( data aggregated over all time points ) and 6D ( data stratified by time point ) both show that , though the vast majority of individual case phenotypes are largely additive ( Hind ≈ hind ) , significantly unfit case individuals ( i . e . fitness < 0 . 9 ) are always predominantly epistatic ( Hind—hind > 0 . 5 ) with respect to controls ., Put another way , although severe disease may arise infrequently in these simulations , when it does , the explanation for it is generally epistatic ., Because these evolutionary simulations keep track of individual mutations across generations , we can also calculate the frequency , in simulated populations , of the individual alleles that account for the epistatic interactions that give rise to unfit cases ., As S5 Fig . shows , alleles involved in producing exceedingly unfit individuals ( fitness < 0 . 75 ) tend to be relatively rare ( frequency between 0 . 2% and 0 . 5% ) , whereas in somewhat less severely unfit cases ( fitness between 0 . 75 and 0 . 9 ) , causal alleles may display frequencies >1% , in the range of frequencies associated with minor alleles that can be followed in GWAS ., Fig 6E shows how the epistatic fraction , Hind-hind , averaged over all substantially unfit cases ( those with relative fitness < 0 . 9 ) , varies with changing levels of phenotypic noise ., In the absence of developmental noise ( vertical axis of Fig 6E ) , strong , synergistic epistasis ( <Hind-hind> > 0 . 5 ) is observed only when environmental noise is of intermediate strength , because only then is the population localized to the corner ( Fig 5F ) ., More surprising , at first sight , is the fact that , in the absence of environmental noise ( horizontal axis of Fig 6E ) , developmental noise must also be of intermediate strength , and is even more constrained than environmental noise ( compare the extent of the red region along the two cardinal axes in Fig 6E ) ., This constraint on developmental noise level originates in a trade-off between its effects on corner localization and ( total ) heritability , Hind ., In the absence of environmental noise , modest developmental noise localizes the population to the corner ( Fig 5E ) , favoring epistasis ( σdev<0 . 075; Fig 6E ) ; on the other hand , large-amplitude developmental noise kills heritability ( Fig 6F; S2A Fig ) , and with it , epistasis ( σdev>0 . 075; Fig 6E ) , which is bounded above by heritability ( recall that heritability , Hind , is partitioned into an additive portion , hind , and an epistatic portion , Hind-hind ) ., Thus , there is a range of developmental noise levels ( σdev≈0 . 05–0 . 10; Fig 6E ) capable of localizing the population near the corner while preserving heritability ., How does the degree of epistasis among unfit ( fitness <0 . 9 ) cases depend upon the other key ingredient of the extended LP model—the interaction parameter , k ?, While decreasing k below 0 . 01 does not affect corner stability ( S4 Fig ) , it does increase the mean epistatic fraction , <Hind-hind> significantly ( S6 Fig; also compare Fig 3A with 3B ) ., Thus , by making the LP landscape as nonlinear as possible , by sending k to zero , one maximizes the strength of epistasis one expects to observe in unfit case individuals ., The fact that the relative fitness of a case individual is such a strong predictor of its probability of being epistatic ( Fig 6C and 6D and S7 Fig ) prompted us to ask how epistasis is affected by the degree to which a trait is under selection , as measured by the selection factor , s , in ( Eq 2 ) ., Fig 6G shows that the trait of interest must be under a certain amount of selection before we may expect strong epistasis among severely affected cases ., That level of selection varies inversely with the level of phenotypic noise , σ=σdev2+\u200bσenv2 , as indicated by the red band in Fig 6G ., In fact , this figure shows that epistasis depends primarily on the product sσ2 ( see also Section 4 in SI ) , so that increasing the strength of selection is equivalent to increasing the noise variance ., Epistasis is greatest for intermediate values of sσ2 ( in the vicinity of 0 . 005 ) , as previously seen in Fig 6E , which varies σ while holding the selection factor fixed at s = 1 ., In summary , evolutionary simulations indicate that strong synergistic epistasis can underlie disease that arises in populations evolving on a generalized LP landscape ., Neither disease nor epistasis tend to arise frequently , but when they do , they correlate; moreover , the strength of that correlation is a function of the severity of the disease ., As a result , epistatic , rather than additive , effects tend to provide the most likely explanation for sufficiently severe disease ., The level of severity ( fitness loss ) required for this is sufficiently small that causal alleles may even rise to relatively high frequencies ( e . g . >1% ) ., As we found when exploring the dynamics of populations as a whole , the degree of phenotypic noise and the interaction parameter , k , have important effects on the degree to which epistasis and disease correlate ., Interestingly , when k is small , it has a greater effect on the prevalence and magnitude of epistasis within a population ( S6 Fig ) than it does on localizing the population to the corner ( S4 Fig ) ., This effect arises because the local shape of the small portion of the landscape’s corner actually occupied by the population , which depends sensitively on k , matters more for the epistasis realized in individuals of the population than for the population dynamics as a whole ., Put another way , moderate nonlinearities are sufficient to localize the population to the most nonlinear part of the landscape , but large nonlinearities are needed to generate strong epistasis in severely affected cases ., We have seen how strong epistasis can arise in severely affected cases ( those with low fitness ) , but how does the dis

Introduction, Results, Discussion, Methods

A major goal of human genetics is to elucidate the genetic architecture of human disease , with the goal of fueling improvements in diagnosis and the understanding of disease pathogenesis ., The degree to which epistasis , or non-additive effects of risk alleles at different loci , accounts for common disease traits is hotly debated , in part because the conditions under which epistasis evolves are not well understood ., Using both theory and evolutionary simulation , we show that the occurrence of common diseases ( i . e . unfit phenotypes with frequencies on the order of 1% ) can , under the right circumstances , be expected to be driven primarily by synergistic epistatic interactions ., Conditions that are necessary , collectively , for this outcome include a strongly non-linear phenotypic landscape , strong ( but not too strong ) selection against the disease phenotype , and “noise” in the genotype-phenotype map that is both environmental ( extrinsic , time-correlated ) and developmental ( intrinsic , uncorrelated ) and , in both cases , neither too little nor too great ., These results suggest ways in which geneticists might identify , a priori , those disease traits for which an “epistatic explanation” should be sought , and in the process better focus ongoing searches for risk alleles .

The contribution of epistasis , or non-additive effects of risk alleles at different loci , to complex traits is much debated among human geneticists ., In this study we use modeling and simulation to identify when evolutionary forces should drive epistasis to become a major part of the explanation for such traits ., We simulate populations evolving in the presence of “phenotypic noise” , i . e . intrinsic and environmental variability in the relationship between genotype and phenotype , and focus specifically on traits that are substantially deleterious ( e . g . fitness loss of at least 10% ) and moderately common ( population frequency between 0 . 1 and 10% ) ., These criteria describe much “common” , heritable human disease , and knowing the expected contribution of epistasis to such diseases could greatly assist efforts to use genome-wide genetic associations to identify their mechanistic underpinnings ., Here we show that , provided that appropriate levels and kinds of phenotypic noise are present during evolution , genetic contributions to disease traits can be expected to be strongly and synergistically epistatic , particularly among the most severely affected individuals in a population ., By identifying scenarios most likely to select for synergistic epistasis , this study may help biologists identify when the extra effort is merited to search for combinatorial genetic interactions in human disease .

epistasis, evolutionary genetics, developmental biology, fitness epistasis, phenotypes, natural selection, heredity, genetics, biology and life sciences, population genetics, population biology, evolutionary biology, genetics of disease, evolutionary processes, genetic loci, evolutionary developmental biology

journal.ppat.0030010

Reconstitution of an Infectious Human Endogenous Retrovirus

A characteristic that is unique to retroviruses is their propensity to integrate their genomes into host-cell DNA as an essential part of their replication cycle ., Thus , if the target cell population of a given retrovirus includes germ cells or their progenitors , retroviral genomes can be inherited in a Mendelian manner as so-called “endogenous” forms ( see 1 for review ) ., Indeed , endogenous retroviruses have accumulated over time in the genomes of many organisms and are extraordinarily common in mammals , comprising approximately 8% of human DNA 2 ., Nonetheless , while some avian , murine , and primate species harbor replication-competent retroviruses within their genomes , intact retroviruses are relatively infrequent and almost all endogenous retroviruses are obviously defective due to the presence of stop codons and frameshifts in one or more genes ., Among the numerous families of defective human endogenous retroviruses ( HERVs ) found in modern human DNA , the human mouse mammary tumor virus–like 2 ( HML-2 ) subfamily of HERV-K proviruses is of special interest ., Even though replication-competent forms of HERV-K ( HML-2 ) have not been found , some proviruses were deposited in the human genome after speciation and represent some of the youngest HERVs known 3–6 ., Also , occasional reports link their expression with human disease 7 ., The age of an endogenous provirus can be roughly estimated by comparing sequence of the two long terminal repeats ( LTRs ) ., At integration , the two proviral LTRs should be identical , but during host DNA replication , each LTR independently accumulates mutations as a function of age , and it is estimated that one difference between two LTRs should occur every approximately 200 , 000 to 450 , 000 y ., Several HERV-K ( HML-2 ) proviruses have been identified in human DNA that have less than five differences between the two LTRs , suggesting deposition perhaps less than 1 million y ago 3–6 ., HERV-K ( HML-2 ) –related proviruses are found only in Old World primates genomes , and many are unique to humans , with nonhuman primate genomes containing empty preintegration sites at orthologous loci ., Compellingly , polymorphism exists in humans with respect to the presence or absence of proviruses at some HERV-K integration sites , indicating insertion relatively recently in human evolution 3–6 ., Furthermore , many of the younger HERV-K ( HML-2 ) proviruses contain a subset of open reading frames ( ORFs ) with a few or no mutations 3 , 6 , 8 ., However , all known HERV-K proviruses are replication defective ., There are several ways in which a defective provirus can proliferate in a hosts genome , including via exogenous infection events following complementation in trans , where functional proteins are supplied by other endogenous or exogenous viruses ., Alternatively , for some retroelements , envelope-independent retrotransposition can occur in cis , where an element copies itself and inserts into a new genomic locus within the same cell , forgoing the normal extracellular phase of the retroviral life cycle ., Defective proviruses can also be proliferated as a result of long interspersed element retrotransposition 9 ., However , most HERV-K ( HML-2 ) replication appears to have been a consequence of autonomous infection by extracellular virions 10 , 11 ., This conclusion is based on the comparatively low number of stop codons and ratio of nonsynonymous to synonymous changes ( dN/dS ) in HERV-K ORFs , indicating a purifying selection on all proteins ., Notably , this finding holds for HERV-K Env 10 , which should be required for replication that includes an extracellular step but not for any other mode of provirus proliferation ., Ancient retroviruses are of interest , in part because they likely imposed selective pressure on host defenses in human ancestors ., Indeed , the tripartite motif ( TRIM ) 5α and apolipoprotein B mRNA-editing enzyme , catalytic polypeptide-like ( APOBEC ) 3 proteins that provide part of the host defense against modern retroviruses have been under positive selection for much of primate evolution 12–16 ., As a retrovirus that appears to have replicated in the ancestors of modern Old World monkeys , apes , and humans , HERV-K may be partly responsible for this pressure ., Moreover , it is conceivable that HERV-K exists today in an undetected replication-competent form in rare humans 4 ., However , no studies of the virology or pathogenic potential of this ancient human virus have been possible because a contemporary , replication-competent HERV-K strain has not been identified and may not exist at all ., Despite some functional degradation due to mutation during deposition or during human DNA replication , HERV-K ( HML-2 ) proviruses that have been deposited in human DNA in the past few million years should be reasonably well preserved and have relatively few inactivating mutations ., Indeed , various studies have shown that individual proteins from certain HERV-K proviruses can function in vitro 17–25 ., We reasoned that it might be possible to resurrect HERV-K ( HML-2 ) in replication-competent form using proviruses that are thought to have most recently entered the human genome as a template ., Therefore , we constructed a HERV-K strain whose genome sequence is a consensus of a subset of HERV-K ( HML-2 ) proviruses ., Importantly , we demonstrate that all viral proteins necessary for viral replication encoded by this provirus are functional and that proteins and genomes based on the reconstructed HERV-K ( HML-2 ) viral genome can be used to generate infectious exogenous retrovirus particles ., Initially , our attempts to construct an infectious HERV-K provirus were based on the sole HERV-K provirus ( HERV-K113 ) that has apparently intact ORFs for all viral proteins 6 ., This provirus , believed to be among the youngest human-specific HERV-K proviruses , is present in the genomes of a minority of humans ., Unfortunately , construction of Gag , Gag-protease ( PR ) , and Gag-PR-Pol expression plasmids based on HERV-K113 resulted in proteins that were poorly expressed and were inefficiently processed and released as virus-like particles ( VLPs ) ( unpublished data ) ., Thus , because presence of intact ORFs did not necessarily imply intact function , we took an alternative approach ., Specifically , we adopted a strategy that was based on the assumption that any inherent replication defects that are encoded within HERV-K ( HML-2 ) proviruses present in contemporary human DNA are either unique to each provirus or shared only by a minority of recently integrated proviruses ., If this assumption was correct , then each individual defect should be absent from a sequence representing the consensus of a collection of proviruses , even if each provirus that contributes to the consensus is defective ., We selected a group of ten full-length HERV-K ( HML-2 ) proviruses to derive a consensus HERV-K sequence ., Specifically , the ten proviruses with the best scores following BLAST searching of human DNA with full-length HERV-K113 sequence were chosen ., As well as HERV-K113 itself , this search yielded HERV-K101 , HERV-K102 , HERV-K104 , HERV-K107 , HERV-K108 , HERV-K109 , HERV-K115 , HERV-K11p22 , and HERV-K12q13 ., All of these proviruses are known to be unique to humans , indicating integration into the germ-line within the last 6 million y , when the human lineage is believed to have diverged from the chimpanzee lineage 3–6 ., Moreover , several show insertional polymorphism in humans , with intact preintegration sites present in a fraction of the human population , suggesting even more recent replication ., While all except HERV-K113 encoded an obvious defect in at least one ORF , all of the selected proviruses also had an intact ORF for at least one of the putative HERV-K proteins ( Figure 1A ) ., The nucleotide encoded by the majority of each of the ten proviruses was deduced for each of 9 , 472 nucleotide positions to derive HERV-KCON ., Thereafter , using a set of synthetic , approximately 60 base oligonucleotides spanning the entire HERV-KCON sequence and a PCR- based strategy to progressively link them together , we first constructed a plasmid containing the HERV-KCON proviral genome ., The complete proviral consensus sequence and the viral proteins that it encodes are shown in Figure S1 ., As expected , the HERV-KCON sequence was positioned close to the root of a phylogenetic tree constructed using HERV-KCON itself and each of the ten proviruses used to derive it ( Figure 1B ) ., Thus , we reasoned that HERV-KCON represented a reasonable approximation to the ancestor of HERV-K sequences that integrated into the human genome within the past few million years and might , therefore , be capable of replication ., Interestingly , pairwise comparisons indicated that the majority of nucleotide differences between HERV-KCON and each of the ten contributing proviruses were either G-to-A or C-to-T changes , or vice versa ( Figure 1C ) ., This finding hints at a possible role of cytidine deaminases in driving HERV-K evolution in humans , and perhaps contributing to the inactivation of contemporary proviruses ., To determine whether the major HERV-KCON structural proteins and enzymes were capable of assembling into retrovirus-like particles , we constructed plasmids expressing the consensus Gag , Gag-PR and Gag-PR-Pol ORFs ., The HERV-K genome has an unusual nucleotide composition in that it is relatively A-rich ., This feature , which is also characteristic of lentiviruses such as HIV-1 , is partly responsible for the nuclear retention of HIV-1 mRNAs and contributes to the requirement for Rev in mediating export of incompletely spliced HIV-1 transcripts ., Indeed , HERV-K encodes a functional ortholog of the Rev protein , termed K-Rev or Rec , that mediates nuclear export of HERV-K RNAs 18–20 , 26 ., Therefore , because of the likely requirement for a Rev-like post-transcriptional activator for efficient HERV-K mRNA export , cDNAs encoding the HERV-KCON structural proteins were inserted into a previously described expression vector , termed pCRV1 27 , that provides an HIV-1 Rev response element to the expressed mRNA in cis and the HIV-1 Rev protein in trans ., Transfection of pCRV1-based plasmids expressing HERV-KCON Gag , Gag-PR , or Gag-PR-Pol resulted in the expression of a protein of approximately 70 to 80 kDa , detected by Western blotting using a commercially available antibody raised against HERV-K Gag ( Figure 2A ) ., This approximated to the size expected ( 74 kDa ) of the intact HERV-K Gag precursor ., A concurrent analysis of proteins pelleted from culture supernatant through 20% sucrose revealed that Gag expression alone could efficiently generate extracellular particles ( Figure 2A ) ., In addition to the 74-kDa Gag precursor , a protein of approximately 40 kDa that reacted with the HERV-K Gag antibody was detected in lysates of cells transfected with Gag-PR and Gag-PR-Pol expression plasmids ., While the precise identity of the 40-kDa protein is unknown , it likely represents a proteolytically processed form of Gag and , therefore , this finding suggested that the HERV-KCON protease was active ., Consistent with this notion , Western blot analysis of extracellular particles generated following Gag-PR and Gag-PR-Pol expression did not contain detectable Gag precursor but did contain the 40-kDa apparently processed Gag species ( Figure 2A ) ., Analysis of total protein in extracellular particles by SDS-PAGE and silver staining revealed that HERV-KCON Gag expression alone generated particles composed of a single dominant protein of the size predicted for the HERV-K Gag precursor , as expected ( Figure 2B ) ., Particles generated by Gag-PR contained a dominant protein of 30 kDa , which based on previous studies likely represents HERV-KCON capsid ( CA ) 24 , 28 , 29 ., A smaller protein or proteins of 20 kDa were also observed in Gag-PR particles , which presumably represents other mature Gag processing product or products such as matrix or nucleocapsid ( Figure 2B ) ., Additionally , a protein of 40 kDa that likely corresponded to the 40-kDa band detected by Western blotting was also observed on silver-stained gels ., However , the 40-kDa protein was a minor species in Gag-PR particles , and it is therefore possible that this protein represents a partly processed intermediate ., HERV-K Gag-PR-Pol expression also yielded particles containing the same apparently processed Gag proteins as those generated by Gag-PR but at slightly lower levels ( Figure 2B ) ., The appearance of the 30-kDa putative CA protein on silver-stained gels ( Figure 2C ) was abolished when three predicted active site residues ( Asp-Thr-Gly ) in the HERV-KCON protease ORF were mutated to Ala-Ala-Ala ., Additionally , a higher-molecular-weight protein , possibly representing the Gag-PR precursor , was observed in particles harvested from cells expressing the mutant Gag-PR protein ( Figure 2C ) ., Although their low abundance relative to contaminating extraneous cellular proteins and the lack of available antibodies precluded unambiguous identification of Pol proteins in SDS-PAGE analyses of HERV-KCON VLPs , supernatants of 293T cell cultures transfected with the HERV-KCON Gag-PR-Pol expression plasmid contained quite high levels of reverse transcriptase activity , as detected by an ELISA-based assay designed for the detection of HIV-1 reverse transcriptase ( Figure 2D ) ., As controls , no reverse transcriptase activity was detected in cultures transfected with HERV-KCON Gag or Gag-PR expression plasmids ., Coexpression of HERV-KCON Gag and Gag–green fluorescent protein ( GFP ) fusion proteins in 293T cells revealed that HERV-KCON Gag localized predominantly to the plasma membrane , where numerous fluorescent puncta were observed ( Figure 2E ) ., Moreover , electron microscopic examination of 293T cells expressing HERV-KCON Gag-PR revealed the presence of cell-associated retrovirus-like particles and structures that appeared to represent assembly intermediates ( Figure 2F ) ., Most particles appeared as 100 to 150 nm , apparently spherical immature virions , with a minority assembled as aberrant particles that appeared as two or more connected , partly assembled , virions ., While we did not observe unambiguously mature virions associated with the surface of Gag-PR–expressing cells , it is possible that full maturation , which was clearly indicated by the biochemical analysis of extracellular VLPs ( Figure 2B and 2C ) , occurred only after the completion of particle release from cells ., Completely or incompletely assembled particles appeared exclusively at the plasma membrane with a morphology resembling partly assembled alpharetroviruses or gammaretroviruses ., Even though betaretroviruses represent HERV-Ks closest exogenous retrovirus relatives , no cytoplasmic , nonenveloped particles , typically observed in betaretroviruses , were found ., To determine whether particles containing the HERV-KCON genome , Gag , PR , and Pol proteins were capable of infectious transfer of the HERV-KCON genome to target cells , we inserted a reporter gene cassette ( cytomegalovirus CMV-GFP ) into the env gene of the HERV-KCON proviral plasmid ., Additionally , because the HERV-K LTR promoter is extremely weak in 293T cells ( unpublished data ) , we replaced U3 sequences 5′ to the TATA box with corresponding sequences from the promoter/enhancer of CMV ., This construct was named CHKCG ( Figure 3A ) ., As expected , transfection of this Env-defective CHKCG construct in 293T cells resulted in GFP expression in transfected 293T cells , but inoculation of target cells with 0 . 2-μm filtered supernatant harvested from these cells did not result in infectious transfer of the reporter gene ., However , when an envelope protein from vesicular stomatitis virus ( VSV-G ) was expressed in trans , clear GFP expression was observed in rare foci of target cells inoculated with filtered supernatant from CHKCG-transfected cells ., Moreover , by boosting HERV-K protein expression , the yield of infectious virions was improved ( Figure 3B–3E ) ., Indeed , when K-Rev/Rec was expressed in trans with CHKCG and VSV-G , infectious particle yield was in excess of 102 IU/ml ( Figure 3B and 3E ) ., Similarly , when the HERV-KCON Gag-PR-Pol expression plasmid was provided in trans , the yield of infectious particles also increased to greater than 102 IU/ml ( Figure 3D ) ., The combined expression of CHKCG , VSV-G , HERV-K Gag-Pol , and K-Rev/Rec yielded the highest infectious titers ( up to 103 IU/ml , Figure 3C and 3E ) , and this combination of plasmids was used to generate infectious HERV-KCON ( VSV-G ) pseudotyped particles in subsequent studies ., While this infectious titer is low compared to that generated by many exogenous retroviruses ( e . g . , murine leukemia virus MLV and HIV-1 ) , the yield of infectious HERV-K particles was of the same order as that obtained with similarly constructed human T-cell lymphotropic virus-I–based vector systems ( 30 and unpublished data ) ., To verify that transfer of reporter gene expression by HERV-KCON particles was via bona fide retrovirus-based transduction , we inoculated 293T cells with HERV-KCON ( VSV-G ) particles containing the CHKCG genome in the presence of azidothymidine ( AZT ) , a reverse transcriptase inhibitor ., AZT is a thymidine analog chain terminator and is known to inhibit reverse transcriptases from a wide variety of retroviruses 31 ., As can be seen in Figure 3F , application of AZT to target cells inhibited HERV-K–mediated reporter gene transduction by approximately 30-fold ., Thus , reporter gene transfer by HERV-KCON was clearly dependent on reverse transcription ., In some cases , low levels of reporter gene expression mediated by retroviral gene transfer can be mediated by reverse-transcribed but nonintegrated retroviral DNA , which can exist as linear or circular forms in target cells 32–34 ., However , these retroviral DNA forms are diluted during cell division and eventually lost ., Stable retrovirus-mediated gene transfer that is transferred to both daughter cells requires that retroviral DNA be integrated into the target cell genome ., While the formation of clear multicellular foci of GFP-positive cells suggested the reporter gene was maintained in daughter cells , integration events are most effectively assayed by daughter cell colony formation under antibiotic selection using retroviral genomes that carry resistance markers ., Therefore , we constructed a variant of the CHKCG genome ( Figure 3A ) in which the CMV-GFP cassette was replaced by one carrying a CMV-driven puromycin resistance gene , termed CHKCP ., We generated HERV-KCON ( VSV-G ) particles carrying CHKCP in the same way as previously ( Figure, 3 ) and found that puromycin-resistant colonies were formed following exposure of 293T target cells to these virions and antibiotic selection ( Figure 4A ) ., Indeed , the infectious titers of puromycin resistance transducing particles were similar to that of GFP-transducing particles ., To further demonstrate that HERV-KCON genomes were capable of integration , hamster ( CHO745 ) cells were infected with HERV-KCON ( VSV-G ) particles carrying the CHKCP genome and four single cell clones were derived from the resulting puromycin-resistant cell population ., Cellular genomic DNA was extracted following expansion of the clones for 2 wk in culture and analyzed for the presence of integrated HERV-K DNA using a PCR-based strategy ( Figure 4B ) ., Hamster CHO745 cells were used for these experiments , because they were found to be as sensitive as human cells to HERV-KCON ( VSV-G ) infection ( see below ) , but unlike human cells , they lack endogenous HERV-K proviruses that would complicate detection and analysis of de novo HERV-K integration events ., As can be seen in Figure 4C , PCR analysis using HERV-K gag specific PCR primers revealed that each of the CHKCP-transduced clones , but not parental CHO745 cells , carried HERV-K DNA ., Next , sequences flanking the integrated provirus were identified using a PCR-based strategy ( GenomeWalker kit; Clontech , http://www . clontech . com ) and in each case revealed the presence of a six-nucleotide duplicated sequence immediately flanking the provirus ( Figure 4D ) ., For three CHKCP-transduced CHO745 cell clones , PCR primers were designed that targeted hamster DNA sequences flanking the integrated HERV-KCON provirus ( Figure 4B and 4E ) , and these were used to authenticate the presence of the intact preintegration site in uninfected hamster cells ( e . g . , Figure 4E ) ., Moreover , PCRs using combinations of the hamster DNA-specific and HERV-K–specific PCR primers were used to authenticate the presence HERV-K provirus/hamster cellular DNA junctions in three of the CHKCG-transduced clones ( e . g . , Figure 4E ) ., Overall , these experiments demonstrate that HERV-K genomes can be replicated via exogenous infection in a reverse transcriptase–dependent manner , resulting in stable and authentic integration into the target cell genome ., Next , we determined whether VSV-G pseudotyped HERV-KCON particles could transduce reporter genes into cells other than 293T and CHO745 ., As can be seen in Figure 5A , several target cells of human , squirrel monkey , feline , and rodent origin could be infected by HERV-KCON ( VSV-G ) ., However , it was noticeable that murine NIH3T3 cells and squirrel monkey Pindak cells were somewhat less sensitive to HERV-KCON ( VSV-G ) , compared to the human and feline cells ., The human cells were each quite similar in their sensitivity to HERV-KCON ( VSV-G ) even though 293T cells display little or no TRIM5α-dependent resistance to retroviruses such as EIAV or N-tropic MLV , while TE671 and HT1080 exhibit strong TRIM5α-dependent resistance to N-tropic MLV and EIAV ., This finding suggested that HERV-KCON may not be sensitive to human TRIM5α ., Additionally , to test whether the HERV-KCON envelope sequence was functional , it was inserted into the HIV-1–based expression vector pCRV1 and expressed along with HIV-1 Gag-Pol proteins and the packageable GFP-expressing HIV-1 vector CSGW ., This transfection mixture should generate HIV-1 particles , putatively pseudotyped with the HERV-KCON envelope protein ., Notably , HIV ( HERV-KCON ) particles were capable of infecting 293T cells , with titers of around 3 × 102 IU/ml ( Figure 5B ) , while particles generated in the absence of HERV-KCON Env were noninfectious ., Inoculation of cells from a small panel of mammalian species revealed that several , including those of human , squirrel monkey , murine , and feline origin , could be infected with HIV-1 ( HERV-KCON ) pseudovirions ( Figure 5B ) ., While attempts were made to generate infectious particles that contained both HERV-KCON cores and Env proteins , we were not able to detect infection events using this combination ., Nevertheless , these experiments indicate that the HERV-KCON genome contains all functional components required to complete an exogenous retroviral replication cycle ., To test the sensitivity of HERV-K to retrovirus restriction factors that it might encounter in human cells and might be responsible for attenuation or extinction of replication therein , we first challenged unmodified , or human TRIM5α-expressing , hamster ( CHO ) -derived cell lines with HERV-KCON ( VSV-G ) ., Despite the fact that the human TRIM5α-expressing cell line was greater than 100-fold resistant to N-tropic MLV relative to the control cell line or B-tropic MLV ( Figure 6A ) , HERV-KCON infected unmanipulated and human TRIM5α-expressing cells with nearly identical efficiency ( Figure 6B ) ., Additionally , CHO cells expressing rhesus macaque TRIM5α or the unique owl monkey variant of TRIM5 ( TRIM-Cyp ) were also similarly sensitive to HERV-KCON ( VSV-G ) infection as unmanipulated control cells ( Figure 6A ) ., This was despite the fact that CHO cells expressing rhesus monkey TRIM5α and owl monkey TRIMCyp were about 30-fold and 100-fold , respectively , resistant to HIV-1 infection compared to HIV-1 carrying an SIVMAC CA ( Figure 6A ) ., Next , we tested whether APOBEC3G and APOBEC3F were capable of inhibiting HERV-KCON replication ., These cytidine deaminases are the major inhibitors of Vif-deficient HIV-1 infectivity , although APOBEC3G is a significantly more potent inhibitor of HIV-1 replication than is APOBEC3F ., Surprisingly , APOBEC3G expression during particle production only marginally inhibited HERV-KCON ( VSV-G ) infection , while APOBEC3F more potently reduced infectivity , reducing titers by about 50-fold ( Figure 6C ) ., This was despite approximately equivalent levels of HERV-KCON Gag expression and generation of viral particles in the presence or absence of APOBEC3G or APOBEC3F ( Figure 6D ) ., Overall , of the restriction factors tested that are likely to be encountered in human cells , HERV-KCON appeared to be resistant to human TRIM5α and APOBEC3G proteins but sensitive to APOBEC3F ., Here , we constructed a HERV-K provirus whose sequence resembles that of an ancestral human-specific HERV-K ( HML-2 ) ., We demonstrate that all viral proteins encoded by this provirus are capable of functioning in the context of a retroviral replication cycle ., While some recent studies have reconstituted “live” viruses from synthetic DNA 35 , 36 , this and a similar study of HERV-K which appeared online while this manuscript was in review 37 are the first examples in which the replication cycle of a virus has been reconstituted using a group of sequences that represent ancient fossils and are demonstrably defective ., The methods used here are conceptually similar to those applied to the reconstitution of the transposable element Sleeping Beauty , in which a functional Tc1/mariner-type transposon present only in defective forms in fish DNA was reconstituted 38 ., Successful reconstitution in that study was achieved using a majority consensus sequence to synthesize an active trasposase protein and selecting cis-acting sequences from a representative element that closely resembled those of the majority consensus sequence 38 ., HERV-K likely replicated in the ancestors of humans for approximately 30 million y but is not known to exist as a replication-competent virus today ., Indeed , it is possible , even likely , that HERV-K has not replicated as a retrovirus for hundreds of thousands of years ., It was not obvious what the optimal approach to reconstitute functional HERV-K sequences would be , since variation in HERV-K sequence could arise through natural variation via error-prone reverse transcription , mutational degradation after deposition in the primate germ-line , or cytidine deamination before , during , or after during initial germ-line deposition ( see below ) ., Moreover , it was possible that the population of proviruses accessible to us in modern DNA represented a highly biased sample of HERV-K genomes where defects might have been positively selected during primate evolution ., Thus , rather than attempt reconstruct the evolutionary history of HERV-K in primates , we adopted a conservative approach to reconstitute functional sequences , selecting ten proviruses that were most similar to a relatively young and comparatively intact HERV-K provirus ( HERV-K113 ) , reasoning that these were the least likely to have undergone substantial sequence degradation ., Moreover , all of the selected proviruses were unique to human DNA , and some were polymorphic in humans , suggesting comparatively recent replication ., While it was possible that all of the selected proviruses would have a common lethal defect , this appeared not to be the case ., Indeed , by compiling a simple majority consensus sequence , we successfully removed individual lethal defects represented in the group of proviruses that contribute to the consensus sequence , allowing replication of the consensus genome in a bona fide reverse transcription–dependent manner that resulted in the stable integration of HERV-KCON genomes into target cells ., Analysis of the replication cycle of HERV-KCON in human cells allowed preliminary characterization of aspects of HERV-K biology that have heretofore been refractory to investigation ., Assembly of HERV-K virions at the plasma membrane is notable , given that the exogenous retroviruses that are most closely related to HERV-K include mouse mammary tumor virus and Mason-Pfizer monkey virus , both of which are betaretroviruses that assemble complete capsids within the cytoplasm of infected cells ., Nonetheless , previous analyses have suggested that the small number of human cell lines that express HERV-K exhibit plasma membrane localized assembly intermediates 28 , 29 , as was observed here for HERV-KCON ., Moreover , previous work has shown that a single amino acid mutation in MPMV Gag protein can change its assembly characteristics from cytoplasmic to plasma membrane associated assembly 39 ., Thus , it should not be surprising that HERV-K assembly appears morphologically different to that of its betaretrovirus relatives ., Two major components of intrinsic defense against retrovirus and retroelement replication in primate cells are the TRIM5 and APOBEC3 gene products 40–42 ., Analysis of these genes in modern primates indicates that these genes have likely been under positive selection pressure for significant portions of primate evolution 12–16 ., As an endogenous retrovirus that has also apparently replicated exogenously and has been active for much of Old World primate evolutionary history , HERV-K is an excellent candidate for an agent that has imposed sustained evolutionary pressure on antiretroviral defenses present in modern primates ., Nonetheless , HERV-K infection was not inhibited by the TRIM5 proteins that were tested ., In the case of human TRIM5α , this was not unexpected , because HERV-KCON was derived from human-specific proviruses that must , by definition , have replicated in humans at some point in their evolution and may , therefore , have evolved resistance to human TRIM5α ., However , HERV-KCON was also resistant to rhesus monkey TRIM5α and also TRIM-Cyp , a form of TRIM5 that is unique to owl monkeys 43 , 44 , a New World monkey species that does not carry HERV-K ., At present , therefore , there is no evidence that TRIM5 proteins and HERV-K have exerted reciprocal evolutionary pressure during primate evolution ., However , analysis of CA sequences reconstructed from more ancient groups of HERV-K proviruses and inserted into HERV-KCON , as well as inclusion of more TRIM5α variants , may be illuminating ., The studies described herein suggest that such approaches to study interactions between ancient retroviruses and their hosts should be feasible ., It was notable that the process of generating a consensus HERV-K genome , in effect , primarily involved the replacement of A and T nucleotides in modern defective proviruses with G and C nucleotides , respectively ., The position of the consensus sequence near the root of the phylogenetic tree suggests that a G or C “ancestral” state at many nucleotide positions in human-specific HERV-K genomes has been replaced by A and T nucleotides in modern defective proviruses ., While these results do not necessarily lead to the conclusion that cytidine deaminases are responsible for the reduction or extinction of HERV-K replication in humans , they hint that this may have been the case ., At a minimum they suggest that cytidine deamination events have impacted HERV-K evolution in humans ., While G-to-A changes were the most frequently represented in comparisons of HERV-KCON with contemporary proviruses , plus-strand C-to-T changes also appeared to be overrepresented ., While most APOBEC-induced mutation is thought to result from deamination of minus strand cytidines during reverse transcription 45–48 , at least some APOBEC proteins are also are capable of inducing C-to-T changes on the plus strand of proviruses , by catalyzing deamination of viral RNA or perhaps dsDNA 49 ., HERV-K replication was sensitive to APOBEC3F but only marginally affected b

Introduction, Results, Discussion, Materials and Methods

The human genome represents a fossil record of ancient retroviruses that once replicated in the ancestors of contemporary humans ., Indeed , approximately 8% of human DNA is composed of sequences that are recognizably retroviral ., Despite occasional reports associating human endogenous retrovirus ( HERV ) expression with human disease , almost all HERV genomes contain obviously inactivating mutations , and none are thought to be capable of replication ., Nonetheless , one family of HERVs , namely HERV-K ( HML-2 ) , may have replicated in human ancestors less than 1 million years ago ., By deriving a consensus sequence , we reconstructed a proviral clone ( HERV-KCON ) that likely resembles the progenitor of HERV-K ( HML-2 ) variants that entered the human genome within the last few million years ., We show that HERV-KCON Gag and protease proteins mediate efficient assembly and processing into retrovirus-like particles ., Moreover , reporter genes inserted into the HERV-KCON genome and packaged into HERV-K particles are capable of infectious transfer and stable integration in a manner that requires reverse transcription ., Additionally , we show that HERV-KCON Env is capable of pseudotyping HIV-1 particles and mediating entry into human and nonhuman cell lines ., Furthermore , we show that HERV-KCON is resistant to inhibition by the human retrovirus restriction factors tripartite motif 5α and apolipoprotein B mRNA-editing enzyme , catalytic polypeptide-like ( APOBEC ) 3G but is inhibited by APOBEC 3F ., Overall , the resurrection of this extinct infectious agent in a functional form from molecular fossils should enable studies of the molecular virology and pathogenic potential of this ancient human retrovirus .

Retrovirus genomes integrate into the genomes of host cells ., If the target cells of a particular retrovirus include germ-line cells , e . g . , sperm or egg cells , then retroviral genomes can be inherited like cellular genes ., So-called “endogenous” retroviruses have accumulated throughout evolution in the genomes of many organisms , including humans ., While all known endogenous retroviruses of modern humans are unable to replicate as retroviruses , the human genome represents a fossil record of ancient retroviruses that once infected our ancestors ., In this study , a collection of “dead” endogenous retroviral genomes in modern human DNA was used to deduce the approximate sequence of an ancestral retrovirus , human endogenous retrovirus ( HERV ) -K , that is now thought to be extinct ., A pseudo-ancestral HERV-K DNA sequence was synthesized and used to produce viral proteins and RNA that could reconstitute the HERV-K replication cycle ., Thus , the replication and biology of a once-extinct retrovirus can now be studied in the laboratory ., Interestingly , reconstituted HERV-K replication experiments , and comparison of the reconstituted HERV-K DNA sequence with the dead HERV-Ks in modern human DNA , suggests that HERV-K may have been extinguished in humans in part by host defenses that induce mutation of retroviral DNA and that the reconstitution of the pseudo-ancestral HERV-K reversed these changes .

viruses, virology

journal.pcbi.1005785

The influence of astrocytes on the width of orientation hypercolumns in visual cortex: A computational perspective

The cortex is the outermost layer of cerebral tissue , composed of neuronal cell bodies and protoplasmic astroytes ., The neurons in the cortex are arranged in columns , and the neurons in each column usually respond to similar features ., In the macaque these columns , known as microcolumns or minincolumns have a density of 1270 minicolumns per mm2 , with each minicolumn having around 142 pyramidal cell bodies 1 ., Now the 3-d volume of cortical tissue could be locally approximated as a 2-d sheet of nodes , with a single node representative of all the neurons within a particular column ., With this approximation it becomes possible to describe a 2-d map in the neuronal space with each node responding to a particular feature in the stimulus space ., A number of such stimulus modality-specific feature maps are topographic in nature , meaning that features that are similar in the stimulus space are mapped onto neighboring locations in the cortical space ., A few examples include the tactile map in the primary somatosensory cortex 2 , the whisker map in the barrel cortex 3 , and the orientation , direction and retinotopic maps in the primary visual cortex 4 ., Understanding the mapping function allows prediction of what features a particular neuron will respond to ., A model which simulates the development of such maps , would aid in understanding which factors contribute to the development of such features maps ., These factors could include internal factors such as the connectivity between the nodes , or the available area of the cortex onto which the features are to be mapped ., Similarly features of the stimuli used for training the model themselves act as external factors ., Self-organizing maps ( SOMs ) have been used extensively to simulate the development of cortical maps 5–12 ., A SOM has two constraints: coverage and continuity ., Optimal coverage implies all input stimuli are mapped evenly on to the output space ., Continuity implies that neighboring neurons in the output space respond to similar stimuli ., The SOM uses local learning rules in order to optimize coverage and continuity ., A biologically realistic variant of SOM , namely the Gain Controlled Adaptive Lateral ( GCAL ) , has been used to investigate the factors involved in the development of a number of feature maps in the primary visual cortex ( V1 ) 13 ., The GCAL model consists of sheets of neurons ., Each neuron in each layer could have 3 kinds of connections , each of which is trained using a normalized Hebbian learning rule: A common feature in most SOMs is the presence of a mechanism by which neighboring neurons respond to similar features whereas those further away respond to dissimilar ones ., The GCAL model achieves this by having short range excitatory connections , and longer range inhibitory connections ., However in V1 inhibitory connections are local ( short range ) and may dominate responses 14 , whereas the long range connections are excitatory ., The effective long range inhibition is achieved by excitatory neurons synapsing onto inhibitory neurons which in turn synapse onto other neurons in its vicinity ., For high contrast stimuli , it is known that the long range connections are in effect ( multi-synapse ) inhibitory in nature 15 ., From a computational perspective , the radius of the effectively short range excitatory connections is important in determining the size of the orientation hypercolumn 6 , 16 ., In the absence of any excitatory connections , with Hebbian trained afferent connections and anti-Hebbian trained lateral connections , a sparse representation yielding independent components of the training set is realized 17 ., This implies that an OPM will give way to a salt-and-pepper organization , without a smooth shift in orientation preference among neighboring neurons , in the absence of lateral excitatory connections ., OPMs are present in carnivores ( such as cats and ferrets ) and primates but absent in rodents 18 ., The term ‘salt-and-pepper’ was originally used to describe the maps seen in rodents , since the orientation preference of neighboring neuronal columns appeared to be uncorrelated and resembled a random pattern ., However , recent experimental evidence suggests that the map is pseudo-random and exhibits some local similarities in orientation preference 19 ., We hypothesize a possible link between astrocytic arbors and presence of OPMs and try to show that larger astrocytic arbors are more conducive to the generation of OPMs ., We investigate the above hypothesis using computational modeling ., We propose a GCAL model having 2 V1 layers: one representative of neurons , whereas the other of astrocytes ., The synaptic activity of V1 neurons is given as input to an astrocyte layer ., The activity of the astrocyte layer is now used to modulate bidirectional plasticity of lateral excitatory connections in the V1 layer ., By simply varying the radius of astrocytes , the effective extent of lateral excitatory neuronal connections can be manipulated ., An increase in the effective radius of lateral excitatory connections subsequently increases the size of a single hypercolumn in the OPM ., When these effective lateral excitatory connections become small enough the OPM disappears and a salt-and-pepper organization emerges ., Hubel and Wiesel proposed that the emergence of orientation preference in principal ( layer 4 ) neurons in the primary visual cortex is primarily due to the spatial arrangement of LGN afferent connections 20 , though the effect of recurrent connections is now also clear ., This contribution of afferent feed-forward connections is also emphasized by Paik and Ringach , who attribute the development of orientation preference maps across species to the Moire interference patterns created due to the spatial arrangement of Retinal Ganglion Cells ( RGCs ) 21 ., While the contribution of feed forward connections to map formation is undeniable , as verified by a number of experiments , the contribution of recurrent lateral connections between cortical columns is also prominent ., At the level of columns , rather than at the level of single neuron , it is known that for high contrast inputs , due to the recruitment of local inhibitory inter-neurons , long range lateral connections are predominantly inhibitory in nature 15 , 22 , 23 ., This configuration of lateral connections is essential for map formation 7 ., What shapes the lateral circuitry in cortical networks ?, Are there mechanisms which could ensure that short range connections are excitatory , whereas as the long range connections are in effect ( considering the contribution of interneurons ) predominantly inhibitory ?, We hypothesize that protoplasmic astrocytes could play a key role in this regard ., Although there are a number of mechanisms by which astrocytes and neurons communicate with each other 24–27 , not all these mechanisms contribute to long term plasticity , crucial for the development of cortical maps ., It must however be noted that there are a number of ways in which astrocytes could influence long term plasticity ., These mechanisms could be summarized as follows: In each of these mechanisms the effective synaptic strength is influenced by astrocytic activity ., NMDA-dependent LTP/LTD is known to be a function of the postsynaptic calcium influx 34 ., The postsynaptic calcium influx is likely dependent on astrocytic activity as well ., The astrocytic influence could be abstracted using a plasticity or learning rule ( such as a BCM curve ) , where the threshold controlling LTP vs . LTD is dependent on the astrocytic activity ., The lateral excitatory connections in the modified GCAL model are modeled in such a manner ., The Gain Control , Adaptation , Laterally Connected ( GCAL ) model , has been used to develop stable and robust orientation maps 35 ., This model builds on the LISSOM model and has , as the name suggests , a mechanism which ensures gain control of input activations and homeostatic adaptation of weights ., The model has 3 layers: a photo-receptive input layer , an ON/OFF LGN layer and a V1 layer ., The activity of the ON/OFF LGN layer is given as L for a node i , j in the layer ., L i , j ( t + 1 ) = f ( γ o ∑ a , b x a , b ( t ) C i j , a b k + γ s ∑ a , b L i , j ( t ) C i j , a b s ) ( 1 ), where ( a , b ) denotes a neuron in the receptive field of the ( i , j ) th neuron in the output layer , with input given as xab; Cij , ab represents the weight from the ( a , b ) th neuron to the ( i , j ) th neuron ., A constant multiplier to the overall strength is given by γo; γs represents the gain-control ., The weights Cij , ab are defined as a difference of Gaussians ., C i j , a b = 1 Z c e x p ( - ( a - i ) 2 + ( b - j ) 2 2 σ c 2 ) - 1 Z s e x p ( - ( a - i ) 2 + ( b - j ) 2 2 σ s 2 ) ( 2 ), where Zc , and Zs denote the normalization factors , σc , and σs regulate the width of the gaussians ., The term C i j , a b s denotes the lateral inhibition received from other ON/OFF units ., C i j , a b s = 1 Z s e x p ( - ( a - i ) 2 + ( b - j ) 2 2 σ c 2 ) ( 3 ) The firing rate of a V1 neuron is dependent on only 3 kinds of inputs , namely: afferent inputs from the LGN ( Lab ( t − 1 ) ) , lateral effectively excitatory inputs , and lateral effectively inhibitory inputs ., Thus the firing rate ( yij ( t ) ) is given as:, y i j ( t ) = f ( p ∑ a , b A i j , a b L a b ( t - 1 ) + q ∑ k , l E i j , k l y k l ( t - 1 ) - r ∑ k , l I i j , k l y k l ( t - 1 ) ) ( 4 ), where p , q , r are scaling factors; Aij , ab is the afferent weight from neuron ( a , b ) to neuron ( i , j ) ; Eij , kl is the lateral excitatory weight from neuron ( k , l ) to neuron ( i , j ) and similarly Iij , kl is the lateral inhibitory weight from neuron ( k , l ) to neuron ( i , j ) ., The function f is a half wave rectifier in order to ensure that the activations are positive with a variable threshold point given as ρ ., The activations yij ( t ) are allowed to adapt in a homeostatic fashion ., The output activity yij and the threshold ρ are adapted as follows:, y ¯ i j ( t ) = ( 1 - β ) y i j ( t ) + β y ¯ i j ( t - 1 ) ( 5 ) ρ ( t ) = ρ ( t - 1 ) + λ y ¯ i j ( t ) - μ ( 6 ), where β is the smoothing parameter and λ is the homeostatic learning rate; y ¯ i j ( t ) is initialized to the average V1 activity ( μ ) ., In order to model astrocytic activation we simulate an additional layer whose input is the synaptic activity ( gs ) present at each node of the V1 layer ., Thus the activation of a single node in this astrocyte layer is given by Sij ., g s i j ( t ) = p ∑ a , b A i j , a b L a b ( t - 1 ) + q ∑ k , l E i j , k l y k l ( t - 1 ) - r ∑ k , l I i j , k l y k l ( t - 1 ) ( 7 ) S i j ( t ) = ∑ i , j ∈ R a s t r o g s i j ( t - 1 ) ( 8 ), where the radius of the astrocyte is given as RAstro ., There is some debate regarding the precise nature of GABA induced calcium oscillations in the astrocyte and the subsequent gliotranmission 36 ., Hence we run an additional simulation which does not consider the effect of GABA induced gliotransmitters ., Now the activation of a single node in the astrocyte layer is given by Sij ., g s i j ( t ) = p ∑ a , b A i j , a b L a b ( t - 1 ) + q ∑ k , l E i j , k l y k l ( t - 1 ) ( 9 ) S i j ( t ) = ∑ i , j ∈ R a s t r o g s i j ( t - 1 ) ( 10 ) The lateral inhibitory and afferent weights are trained using the same normalized Hebbian rule given by:, w i j , m n ( t + 1 ) = w i j , m n ( t ) + η y i j ( t ) P m n ( t ) ∑ m n ( w i j , m n ( t ) + η y i j ( t ) P m n ( t ) ) ( 11 ), where Pmn is the generalized notation for the pre-synaptic activity originating from the neuron ( m , n ) ; η is the learning rate ., These learning rates can be different for each of the connections: ηA , ηE and ηI are the learning rates for the afferent , excitatory and inhibitory connections respectively ., However the lateral excitatory connections adapt using a variant of the BCM rule with a threshold function θ being a function of the astrocytic activation at the corresponding node ., It has been previously proposed that astrocytes introduce metaplasticity by shifting the BCM curve 31 ., E i j , k l ( t + 1 ) = E i j , k l ( t ) + η E y i j ( t ) ( y i j ( t ) - θ i j ) y k l ( t ) ( 12 ) θ i j = ( 1 - S i j ) ( 13 ) Astrocytes communicate with each other via gap junctions; however only distal branches are connected , resulting in astrocytic microdomains with less than 10% overlap 37 ., The gap junctions could be modeled using Gaussian random lateral excitatory connections to the 8 nearest neighboring nodes ., A schematic of the model is shown in Fig 1 The parameters used for the GCAL model are a superset of those used in the standard LISSOM model ., The complete list of parameters are given in Table 1 ., The simulations are performed using the Topographica simulator 38 ., We vary the astrocytic radius and observe the changes in the orientation map developed ., The experimentally reported astrocytic radii are estimated using the Glial fibrillary acidic protein ( GFAP ) as the astrocytic marker ., However , the GFAP marked region accounts for only 15% of the actual astrocytic volume ., Hence we scale the astrocytic radii by a factor of 2 in the simulations ., The model is trained for 10000 iterations ., The training regime consists of elongated 2-dimensional Gaussians with centers and orientations drawn from a uniform random distribution ., The astrocytic radius is varied and the corresponding orientation maps developed are studied ( Figs 2 and 3 ) ., It is observed that on reducing the astrocytic radius , the periodicity of the map increases and the width of a single hypercolumn decreases ., Thus in a given area of cortical tissue 3 x 3 mm , the number of orientation hypercolumns would increase as we reduce the astrocytic radius ., The neuronal and astrocyte maps developed have similar orientation preferences which could be quantified by their stability index ., The stability index between the astrocytic and orientation maps is shown in Fig 4 ., These results demonstrate that the astrocyte radius has a profound effect on OPM formation ., The development of a few of these maps and their stability indices across iterations are shown in Figs 5 , 6 , and 7 ., The V1 orientation preference map is probed at 250 , 500 , 750 , 1000 , 2500 , 5000 , 7500 and 10000 iterations ., It is observed that the map developed becomes stable after a few initial iterations , as quantified by the corresponding stability indices ., These results demonstrate the model develops stable orientation maps ., We also simulate 2 additional conditions which could effect the development of the orientation maps: ( 1 ) Considering there is no GABA induced gliotransmission: Since the effect of GABA induced calcium oscillations is not well understood in literature , we also simulate the map development ignoring the corresponding term as described in Eq 9 ., ( 2 ) Considering the effect of gap junctions in the astrocyte layer: The basic simulation does not consider the effect of gap junctions among astrocytes ., As described in the methods section , we introduce gap junction by considering excitatory connections among the nearest neighbors in the astrocyte layer ., The maps formed for these 2 conditions are shown in Figs 8 and 9 respectively ., The maps developed using all 3 conditions ( basic , no GABA , Gap junctions ) appear visually similar and their features , which are further quantified ( See Fig 10 ) , show a similar trend ., These results indicate that the correlation between astrocyte radius and hypercolumn widths is robust for all the conditions considered ., The number of pinwheels observed in the neuronal ( V1 ) orientation map in the simulated region ( 3 x 3 mm ) is shown in Fig 10 ( A ) ., As expected , the number of pinwheels falls with increasing astrocytic radius ., The number of pinwheels per hypercolumn remain approximately constant , centered around π for the maps in which a clear orientation preference map ( OPM ) structure is present ( Fig 10 ( B ) ) ., However for smaller astrocytic radii the map begins to disintegrate ., These results strengthen the hypothesis that the astrocytic radii influence the formation of orientation maps ., For higher astrocytic radii the number of pinwheels per hypercolumn stabilizes to values around π ., This result is in keeping with experimental findings which show that the number of pinwheels per hypercolumn is a constant π across species 39 ., The trend observed in the simulated widths of the hypercolumn and the corresponding astrocytic radii are comparable with the scant experimental evidence available as shown in Fig 10 ( C ) ., The transition from a salt and pepper kind of map to a smooth orientation map could be quantified using 2 methods: ( 1 ) Change in the number of pinwheels/ hypercolumn: Experiment results indicate that the number of pinwheels/ hypercolumn remains a constant across species , even with differing hypercolumn widths 39 ., Thus , if such a ratio is no longer maintained , the map developed no longer resembles a smooth orientation preference map ., However , the map developed is also not truly random since there might be local patches with similar orientation preference ., A recent study has shown that in rodents the map only appears to be random , and has significant local orientation similarity 19 ., ( 2 ) Local similarity in orientation preference: This method quantifies the local smoothness of the map developed ., The mean angle of separation between the orientation preference of a node and all others within a predefined radius of interest is computed and compared for different astrocytic radii ., We then compare the results using the 2 methods and observe that a sharp transition between salt and pepper and a smooth orientation maps is absent ( Fig 11 ) ., Rather , an intermediate state which exhibits local patches of orientation similarity , but lacks the features of a true orientation map is seen ., A fascinating feature of orientation mapping is that not all species display a smooth transition in orientation preferences as we probe along the cortical surface ., Rodents , in particular have neuronal columns which are orientation specific but arranged in a seemingly randomized fashion across V1 ., This kind of organization is referred to as a salt and pepper configuration ., The presence or absence of OPMs and their potential consequences for information processing is a topic of current interest ., Another interesting fact in those species which do have OPMs is that the size of the hypercolumn varies from species to species ., However the number of pinwheels per hypercolumn appears to remain constant across species ., Self organizing mechanisms have been extensively utilized to model OPMs ., These models rely on a mechanism that ensures that neighbouring neuronal columns respond to similar features , whereas distant ones to different features ., This is invariably implemented by invoking local excitatory and larger inhibitory connections ., However cortical inhibitory connections are known to be short range , whereas excitatory ones are longer ranged laterally ., These long ranged inhibitory connections have been explained away as long ranged excitatory neurons recruiting local inhibitory neurons , such that the net effect is inhibitory ., However , a mechanism that ensures that short range connections are effectively more excitatory than inhibitory has proven elusive ., The arguments summarized above have been discussed in detail by Swindale 7 ., He postulates the possibility of extracellular diffusion of chemical messengers mediating this short range excitatory connectivity ., However there are a number of issues with the diffusion hypothesis ., Firstly diffusion , a passive process , would ensure roughly similar excitatory radii across species and would thus imply by extension similar widths of orientation hypercolumns across species ., In reality the widths of hypercolumns vary widely across species ., Rodents do not have a smooth topographical variation in orientation preference and hence do not have defined hypercolumns 18 ., Thus diffusion alone would not explain the variation in hypercolumn widths ., Secondly the pyramidal apical dendrites , which are used to define the width of a minicolumn ( also called microcolumn ) are roughly the same ( ≈ 30μm ) for the rhesus macaque and the rat 40 ., Thus the chemical messenger which diffuses should have similar effects at the level of cortical columns ., However as stated earlier this again does not hold true ., Astrocytes are known to regulate both excitatory and inhibitory cortical circuits , via a combination of glutamate and GABA re-uptake by transporters , gliotransmitter release , and regulation of neuronal excitability 27–29 , 31 , 41 , 42 ., Indeed , optogenetic astrocyte calcium activation modulates the excitatory-inhibitory balance and increases response selectivity of excitatory neurons within local cortical microcircuits 28 ., Thus the extent of astrocyte influence may directly influence the range of local influence in the cortex ., As mentioned earlier , protoplasmic astrocytes are thought to contribute to metaplasticity 30 ., As shown in Fig 12 , astrocytes release gliotransmitters which are known to shift a BCM like curve to the left , implying greater LTP for lower postsynaptic firing rates in excitatory neurons 31 ., This constitutes a greater increase in synaptic strength for those synapses in the vicinity of the gliotransmitter releasing astrocyte ., Now , astrocytes are understood to release gliotransmitters in correlation with their internal calcium levels 30 ., Glutamate in the synaptic cleft , either via receptors or transporters , mediates the calcium levels in the engulfing astrocyte ., Astrocytes associated with synapses corresponding to those layer 4 pyramidal neurons , which receive direct thalamic input , would have greater calcium levels as compared to other astrocytes ., Hence the metaplasticity induced in the neighboring excitatory synapses in the domain of these astrocytes would also be more pronounced ., Over time , this would lead to a greater excitatory drive for those neurons in the vicinity of the neuron receiving the direct thalamic input , as shown in Fig 13 ., We hypothesize that astrocytes influence local synapses and specify the radius of lateral excitatory connections ., This in turn influences the size of hypercolumns across species ., A comparative table specifying hypercolumn widths and astrocyte radii is specified in Table 2 ., In the standard GCAL model there is a constraint placed on the maximum radius of the lateral excitatory connections 35 ., Indeed , this constraint is necessary to ensure that the lateral excitatory connections are shorter in range than the lateral inhibitory ones ., Such a configuration is essential for the development of the self organized orientation map ., In our proposed model the lateral excitatory connections have no defined maximum radius ., The limit on the lateral excitatory connections is implicitly imposed due to the fact that the astrocytes control the BCM threshold of the excitatory neuronal synapses within their ( astrocyctic ) radius of influence and ensures LTP ., For synapses outside the astrocytic radius , the threshold is such that LTD occurs and these connections are pruned off automatically ., As a computational principle , any mechanism that can control the lateral excitatory radius with respect to the lateral inhibitory radius , could produce similar maps as those shown in this manuscript ., However , several lines of evidence indicate that astrocytes are strongly involved in this mechanism ., First , astrocytes have been shown to regulate the excitatory to inhibitory balance in local neuronal circuits 28 ., Importantly , astrocytes express transporters for both glutamate and GABA , and can thus regulate the strength of both excitatory and inhibitory synaptic transmission ., Second , they have a significant role in regulating local synaptic plasticity , in particular local neuronal excitation , via a range of mechanisms that include modulation of NMDA receptors as well as integrating other plasticity-mediating neuromodulators such as acetylcholine , noradrenaline 43 , 44 ., Together , these effects are well placed to implement the BCM rule ., Third , astrocytes are known to form microdomains with less than 10% overlap 37 ., Thus , the astrocytic organization automatically results in the formation of local domains , which are influenced by these transmitters/modulators ., Fourth , the radii of these local domains are known to be less than the effective lateral inhibitory radius , thus resulting in the required short range excitation and longer range inhibition ., We simulate the development of orientation maps with varying hypercolumn widths , by simply varying the radius of astrocytic connections using the LISSOM model with an additional layer simulating the astrocytic activation ., We observe that increasing the astrocytic radius , and thereby the effective radius of lateral excitatory connections in the V1 neuronal layer , the width of the hypercolumn developed shows a proportionate increase ., When the effective lateral excitatory radius is reduced so as to almost prevent any similarity in orientation preference of neighboring neurons , the OPM disappears and a salt and pepper configuration of neuronal arrangement of orientation preference is seen .

Introduction, Methods, Results, Discussion

Orientation preference maps ( OPMs ) are present in carnivores ( such as cats and ferrets ) and primates but are absent in rodents ., In this study we investigate the possible link between astrocyte arbors and presence of OPMs ., We simulate the development of orientation maps with varying hypercolumn widths using a variant of the Laterally Interconnected Synergetically Self-Organizing Map ( LISSOM ) model , the Gain Control Adaptive Laterally connected ( GCAL ) model , with an additional layer simulating astrocytic activation ., The synaptic activity of V1 neurons is given as input to the astrocyte layer ., The activity of this astrocyte layer is now used to modulate bidirectional plasticity of lateral excitatory connections in the V1 layer ., By simply varying the radius of the astrocytes , the extent of lateral excitatory neuronal connections can be manipulated ., An increase in the radius of lateral excitatory connections subsequently increases the size of a single hypercolumn in the OPM ., When these lateral excitatory connections become small enough the OPM disappears and a salt-and-pepper organization emerges .

Columns of neurons in the primary visual cortex ( V1 ) are known to be tuned to visual stimuli containing edges of a particular orientation ., The arrangement of these cortical columns varies across species ., In many species such as in ferrets , cats , and monkeys a topology preserving map is observed , wherein similarly tuned columns are observed in close proximity to each other , resulting in the formation of Orientation Preference Maps ( OPMs ) ., The width of the hypercolumns , the fundamental unit of an OPM , also varies across species ., However , such an arrangement is not observed in rodents , wherein a more random arrangement of these cortical columns is reported ., We explore the role of astrocytes in the arrangement of these cortical columns using a self-organizing computational model ., Invoking evidence that astrocytes could influence bidirectional plasticity among effective lateral excitatory connections in V1 , we introduce a mechanism by which astrocytic inputs can influence map formation in the neuronal network ., In the resulting model-generated OPMs the radius of the hypercolumns is found to be correlated with that of astrocytic arbors , a feature that finds support in experimental studies .

cell physiology, medicine and health sciences, neurochemistry, nervous system, astrocytes, junctional complexes, vertebrates, electrophysiology, neuroscience, macroglial cells, mammals, animals, gap junctions, synaptic plasticity, neurotransmitters, neuronal plasticity, developmental neuroscience, animal cells, glial cells, gamma-aminobutyric acid, biochemistry, cellular neuroscience, rodents, eukaryota, cell biology, anatomy, synapses, physiology, neurons, biology and life sciences, cellular types, afferent neurons, amniotes, neurophysiology, organisms

journal.pcbi.1000229

Behavioral Sequence Analysis Reveals a Novel Role for ß2* Nicotinic Receptors in Exploration

nAChRs are well-characterized transmembrane allosteric oligomers composed of five, identical ( homopentamers ) or different ( heteropentamers ) subunits 1 ., Nine, different subunits are widely expressed in the mammalian brain , modulating, neurotransmitter release , neuronal excitability and activity dependent plasticity in, most , if not all , mammalian brain structures 2 , 3 ., The elementary mechanisms, of nAChRs functions are investigated in great details , yet important issues relevant, for the role of nAChRs at the higher level , have received less attention ., The need, to fill this gap is reinforced by nAChR participation in a diverse array of, neuropathologies , including Alzheimers disease , Parkinsons, disease , schizophrenia , epilepsy and Attention-deficit hyperactivity disorder ., The, complex nature of all these disorders underlines the nicotinic influences over, neuronal circuits involved in attention , motivation and cognition 2 , 3 ., The issue then becomes how to tackle this problem in mouse models that allow, pharmacological and genetic manipulations , but for which, “psychological” processes must be inferred from observable, behaviors ., Mice deleted for ß2-subunit containing nAChR, ( ß2−/− ) have been the first nicotinic receptor mutant, to be characterized , and found to exhibit more rigid behavior and less behavioral, flexibility than wild-type ( WT ) animals 4 ., Overall , these, experiments suggest that ß2−/− mice reduce the time, allocated to explore a novel environment 4 , 5 ., Lentiviral, reexpression techniques indicate that this phenotype is linked to the expression of, ß2*-nAChRs in the ventral tegmental area 6 , 7 and in the, Substantia Nigra 8 ., ß2−/− mice were shown to be hyperactive in an, open-field paradigm , with a reduced movement at low speed , and consequently an, increased movement at high speed ., Hyperactivity in an open field is often used as a, general and non-specific term characterizing experimental conditions where animals, show either an increased amount of displacement and related locomotor behaviors , or, changes in the frequency of specific motor acts 9 ., Increased locomotor, activity in an open field can reflect different processing and alterations in the, organization of behavior 9 ., A complete description of hyperactivity then, requires to study duration and temporal patterning ( i . e . the sequence ) of behavioral, acts ., In this paper , we address the problem of tracing , by analyzing temporal, organization of movement , mouse cognitive and/or decision making behavior that can, account for mouse hyperactivity in the open-field ., Open-field behaviors have been used to study forced exploration of a new environment ., It has been shown that it involves both exploratory and stress/fear components 10–13 ., Furthermore , kinematic, features based on instantaneous speed and location have been used to demonstrate, that rat and mouse trajectories are far from random 14 , 15 , and that animals can, stop more frequently in specific locations of the field that structure their, trajectory 16 , 17 ., Here , we focus on the analysis of the behavioral, sequence , namely the time-organized sequence of patterns that composes the behavior ., Considering a sequence of acts , a question would be whether information contained in, the structure of this sequence and the presence of specific associations between, acts reflects decision-making behavior and can be used to assess alterations of this, process ., We developed further the method already successfully applied to detect modifications, of locomotor behavior caused by mutations in ß2−/−, mice 4 , 6 , or in goldfish 18 ., The principle of the, method is to decompose animal trajectories into a combination of discrete units, extracted by applying a threshold to continuous variables ., We show that the use of a, variable-length Markov model 19 to analyze the sequence of symbols allows to, unravel significant alterations in the way ß2 mutant mice organize their, behavior , and use “stops” to explore their environment ., Both WT and ß2−/− mice were active in the, open-field ., They exhibited movements along the wall , sequences of trajectories, in the middle of the field ( Figure, 1A ) , and alternation between locomotor progression and periods of, slow movements ., This allowed us to describe locomotor activity in terms of a, sequence of four states {PI , PA , CI , CA} ( Figure 1B and 1C ) ., ß2−/− mice have been shown to be hyperactive in the, open-field ( Granon et al 2003 , Avale et al , 2008 ) , with a distance traveled, during 30 min being 1 . 25 times longer in KO compared to WT mice ( Figure 2A ,, Δ\u200a=\u200a34 . 57 m ) ., This hyperactivity was, reflected in the time spent in an inactive or active state with a decreased time, in the inactive state in mutant mice ( Figure 2B ) ., The relation between the distance, traveled and the duration of the different states were however not different in, the two strains ., For both strains , the distance traveled during active or, inactive states was different , but both exhibited a linear relationship with the, duration of a given event ( μ\u200a=\u200a0 . 113, and 0 . 117 in active phase for wt and ß2−/− mice ,, and μ\u200a=\u200a0 . 02 and 0 . 023 in inactive, phase ) ., These relationships tended to break down for long events , but were not, different in WT ( Figure 1C ,, left ) and in ß2−/− mice ( Figure 1C , right ) ., The, distance traveled was then roughly reflected in the time spent in inactive or, active states ., These results suggest that higher locomotor activity in, ß2−/− mice is not due to a modification of the, velocity distribution ( either in the active or inactive phase ) , but rather to a, significant change in the organization of the behavior ., A change in the time spent in inactive states does not give any insight into the, modification of the temporal structure of behaviors ., Analysis of transition, frequencies and conditional probabilities between different states of the animal, were then carried out ( Figure, 3A1 ) ., Using only four states {PI , PA , CI , CA} did not allow to build a, first-order Markov description of the sequence of states ., Indeed , when checking, for all possible combinations of states X , Y , Z whether, P ( X|YZ ) =\u200aP ( X|Y ) was satisfied , revealed that, the probability of states X after Y\u200a=\u200aPA did, not depend only on the present state PA , but also on the previous one Z ( Figure 3A2 ) ., In order to, obtain a first order Markov dynamics , PA symbols had to be differentiated into, peripheral movement that follows central movement ( CA ) , and peripheral movement, that follows inactivity in the periphery ( PI ) ., They will be designated by the, symbols PAc and PAp , respectively ., Using the five symbols {PAc , PAp , PI , CA , CI}, allowed to describe open-field activity by a first-order process ( Figure 3A3 ) ., This implies, that , with such a state description , the animal movement depends only on the, preceding state , suggesting a very local organization of decision-making ., The, same description could be applied to ß2−/− mice ., However , in mutants , the percentage of transitions from periphery to center ( PA, → CA ) was enhanced , while the “stops in the center”, transitions ( CA → CI ) were reduced ( Figure 3B ) ., Stationarity has been tested by comparing transition probabilities obtained, during the first and the second 15 minutes of the experiment ., We observed, ( i ) a, slight modification of ( PI → PA ) probability of transition ( it, decreases from 97 . 7% to 95 . 5% , and from 98 . 0%, to 96 . 0% in WT and ß2−/− respectively ) ,, and, ( ii ) an increase of ( CA → CI ) transition with time ( from, 22 . 3% to 32 . 2 and from 13 . 9% to 25 . 3 in WT and, ß2−/− mice respectively ) ., This last modification, indicates that animals have a higher tendency to stop at the center in the, second part of the experiment ., This increase is similar in WT and in, ß2−/− mice ., Distributions of residence times were also modified in, ß2−/− mice ( Figure 3C ) ., Comparison of the mean of, residence times in individual using the Wilcoxon test indicated that PI , PAc, and PAp residence times were significantly modified in, ß2−/− mice ., PI average duration was reduced, 13% ( Δ mean\u200a=\u200a0 . 58 sec ,, p\u200a=\u200a0 . 028 ) , while PAp and PAc average duration, were increased 15 . 2 and 35 . 3% ( Δ, mean\u200a=\u200a0 . 72 ,, p\u200a=\u200a0 . 0017 and 1 . 66 sec, p\u200a=\u200a1 . 7e-6 ) , respectively ., Mean of CI or CA, states were not statistically modified , despite an apparent difference in the, distribution of CI ( not shown ) ., In the state sequence , CA is preceded either by PAp , PAc or CI ., In WT , there was, no significant difference between time distributions of CA , depending on the, preceding state ( Wilcoxon test ) ., In contrast , CA resident time was increased, after a CI when compared with PI preceding a PAp or a PAc, ( mean\u200a=\u200a3 . 09 against 2 . 7 and 2 . 8 sec , Wilcoxon, test , p<0 . 001 in both cases ) ., Similar dependencies on preceding state, were observed for PI state duration ., Mean duration varied significantly, ( mean\u200a=\u200a4 . 01 , 5 . 16 and 4 . 11 sec , Wilcoxon test ,, p<0 . 001 in pair comparison ) after CI , PAc or PAp , respectively, ( mean\u200a=\u200a4 . 01 , 5 . 16 and 4 . 11 sec , Wilcoxon test ,, p<0 . 001 in all pair comparisons ) ., Similar properties were observed in, ß2−/− mice, ( mean\u200a=\u200a3 . 08 , 4 . 32 and 4 . 08 sec , Wilcoxon test ,, p<0 . 001 in all pair comparisons ) ., Deletion of the ß2-subunit gene affected both the residence time, distribution and the transition matrix ., To identify more specifically the locus, of the behavioral sequence where the mutation effect takes place , we used a, modeling strategy ( see Methods ) ., We first checked the validity of the simulation ( see also Text S1 and, Figure, S1 and Figure S2 ) and that the numbers of occurrences of each of the five, states in 30 min experiment agreed well in both WT and, ß2−/− mice with numbers obtained with simulated, data when the respective matrix of transition and residence times were used ., Accordingly , the total traveled distance being almost linearly ( Figure 2C ) related to the, total time spent in each of the five states , it was also well-reproduced using, simulation ( Figure 4A ) ., We, also tested the impact of non-stationarity and resident time sequence dependency, ( see also Text, S1 and Figure S1 ) on the simulation ., To further dissect the respective contribution of the transition matrix and of, the residence time distributions , we modeled data based on:, ( i ) transition, matrix of WT and residence time distribution of WT ( labeled WT/WT ) ,, ( ii ), transition matrix of ß2−/− and residence time, distribution of WT ( ß2/WT ) ,, ( iii ) transition matrix of WT and, residence time distribution of ß2 ( WT/ß2 ) , and, ( iv ), transition matrix of ß2−/− and residence time of, ß2−/− ( ß2/ß2 ) , and we compared, the time spent in PI and in PAc ( Figure 4B ) for the various model configurations ., Convolving matrix, and residence time distribution demonstrated that none of them fully explained, modifications of the time spent in a given state and consequently the, “hyperactivity profile” ., Transition probabilities and, residence time distribution explained individually no more than 56%, of the total difference observed between WT and, ß2−/− , while their sum effect explained 95 and, 92% of the total mean difference observed between WT and, ß2−/− ., In terms of quantification this suggested, that both matrices and distributions of residence time should be used ., A final question was whether a single modification of a WT sequence property, could reproduce most of the ß2−/− phenotype ., The, observed behavioral changes between WT and ß2−/−, are open to a variety of interpretations ., One of them is that, ß2−/− specifically reduce some stops ., The main, advantage of such hypothesis is that modification of only one element ( decreased, number of stop ) accounts for matrix and residence time difference between WT and, ß2−/− mice ., A simple simulation ( see Methods , “stop reduction”, model ) revealed that removing 30% of stops in WT sequences reproduced, well the number of occurrences of each of the five states ( Figure 5A ) , matrices ( Figure 5B ) , and residence time distributions, ( Figure 5C ) ., More, precisely , PI was not changed , which means that the model does not explain the, decrease observed in ß2−/− mice ., However , Pap and, Pac increased to a level compatible with resident time observed in, ß2−/− mice ( Δ, mean\u200a=\u200a0 . 27 sec , Wilcoxon test ,, p\u200a=\u200a0 . 09 and Δ, mean\u200a=\u200a0 . 43 sec , Wilcoxon test ,, p\u200a=\u200a0 . 49 for Pap and Pac respectively ) ., Such, modeling identified the “stop” as an element that could, explain differences between WT and ß2−/ ., We then focused our, analysis on this particular moment ., Finite-state systems deriving from the discrete analysis of a continuous movement, necessarily coarsen the fine structure of that movement ., What has been , so far ,, identified as inactivity in this paper , is a mode of motion close to a complete, stop of the animal ., During this period of inactivity the mouse can however make, a variety of movements ., The animal can progress forward slowly ( with a small but, constant speed ) , freeze , perform a number of action patterns ( i . e . , grooming ,, rearing , scratching , etc ) , or orienting movements ( head scanning , sniffing ,, etc ) ., In order to be able to differentiate some of these patterns , we have, simultaneously recorded the position of the animal and digitized video images, ( 25 frames/second ) ., These images have been used as the input for fine off-line, movement analysis ( Figure, 6A ) ., Visual analysis of video images allowed us to distinguish periods, with rearing and head scanning movements , from periods with only reorientation, or no change in orientation ., Five classes of inactivity periods were have been, distinguished ., They corresponded to rearing , scanning , grooming , border rearing, and sniffing ( see Methods ) ., Stops at the, periphery of the open-field were differently distributed in WT, ( n\u200a=\u200a14 ) and β2−/−, ( n\u200a=\u200a11 ) mice ( Figure 6B ) ., The numbers of rearing , wall, rearing , and sniffing were not affected and were similar in both strains, ( Δ\u200a=\u200a3 . 18 , Wilcoxon test ,, p\u200a=\u200a0 . 32;, Δ\u200a=\u200a0 . 4 , Wilcoxon test ,, p\u200a=\u200a0 . 80;, Δ\u200a=\u200a6 . 18 , Wilcoxon test ,, p\u200a=\u200a0 . 12 , respectively ) ) ., Grooming patterns, were increased ( Δ\u200a=\u200a4 . 1 , Wilcoxon test ,, p\u200a=\u200a0 . 003 ) , whereas scanning was decreased, ( Δ\u200a=\u200a6 . 9 , Wilcoxon test ,, p\u200a=\u200a0 . 0008 ) in mutant mice ., Scanning behavior, being related to the “exploration” of , or the information, update about , the environment , differences observed in scanning could therefore, have a consequence on the sequence of behaviors ., New information obtained by the splitting of PI into five subtypes identified by, the dominant behavioral acts , i . e . rearing , scanning , etc . , can challenge the, description of the sequences in two ways ., First , the knowledge of the animal, acts during a PI state can modify the probabilities of consecutive states, without modifying the first-order Markov description ., Second , new information, about PI can modify not only the conditional probabilities but also the order of, the Markov description , thus requiring a more complex description of the, process ., The conditional probability of transition from PA to CA was modified by the, knowledge of the behavioral act performed during stops preceding PA ( Figure 6B , left ( top ) , ANOVA ,, F ( 6 , 91 ) =\u200a13 . 4 ,, p\u200a=\u200a8e-11 ) ., More specifically ,, P ( CA|PA ) =\u200aP ( CA|PI-PA ) , when no further, indication is given on PI , but the probability of transition was greatly, enhanced when the animal performed scanning ., That is ,, P ( CA|PA ) <P ( CA|PIsc-PA ) if PIsc was a scanning behavior, ( Δ\u200a=\u200a0 . 36 , test, p\u200a=\u200a1 . 5 e-08 ) ., These results showed that after, scanning an animal tended to engage more frequently in a transition to the, center of the arena than after a stop paired with a different activity ., Probability to stop at the center of the arena was however not modified by the, activity of mice during a PI ( Figure 6B , left ( bottom ) , ANOVA ,, F ( 7 , 104 ) =\u200a0 . 91 ,, p\u200a=\u200a0 . 49 ) ., In, ß2−/− mice , the modification of probability after, scanning disappeared , that is , the first order model was not modified by, knowledge of the behavioral act occurring during a PI ( Figure 6B , right ) ., Providing new information about the PI state modified the Markov order of the, description ., We therefore switched to Variable Length Markov Chain modeling ( see, Methods ) ., If we consider two main populations of stops , i . e . scanning and no-scanning , a, tree representation of the influence of the past behavior , i . e “the, context” , on a given decision can be built ., For this purpose , the, sequence of symbols was fitted using a Variable Length Markov Chain model ( VLMC ,, see Methods ) ., Animal trajectories were, described using six symbols CI , CA , PAp , PAc , PInsc and PIsc , the two last, states coding for stop at the periphery without or with scanning , respectively ., Sequences from different animals were concatenated for VLMC analysis ., The WT mice context tree ( Figure 7A ,, left ) showed seven contexts ., Five of them were first order ( from top, to bottom , CI , CA , Pac , PInsc and PIsc , Figure 7A ) , indicating that the next symbol, ( X ) depends uniquely on the present state ., More interestingly , two contexts with, second order also appeared ., The first corresponded to the previous demonstration, that after “scanning” an animal tended to engage more, frequently in a transition to the center of the arena ., The second indicated, that , in contrast , when mice did not perform scanning , they preferentially made, a stop in the periphery ., This is schematized ( Figure 7A , right ) by a “PI choice, point” , where the movements that follow depend on what activity the, mouse had performed during the previous PI ., The context tree of β2−/− mice was made of eight, contexts , four of them ( CI , Pap , PInSC , PIsc ) being of first order ., The, architecture of the tree was clearly modified when compared to WT ., Strikingly ,, dependence between movements during PI and “transition to, center” completely disappeared ., In contrast , the tree highlighted, different chains in the ß2−/− sequence of, behavior , with chains of second or third order that organized movements and, relations between PAc and CA ( Figure 7B ) ., In this paper we have investigated the processes underlying, ß2−/− mouse hyperactivity in an open field ., These mice, exhibit an increase in the total distance traveled in the open field by about, 40% when compared to WT ., Consistent with this hyperactive phenotype ,, ß2−/− mice spent more time in fast , and less time in, slow , movements ., To analyze mouse trajectories we developed a specific approach, based on a dissection of mouse behavior in the open field as a sequence of motor, activities organized in patterns ., We have shown evidence for two main modifications, of the behavior in ß2−/− mice:, ( i ) quantitatively ,, mutant mice show a reduced number of stops and modification of specific transition, probabilities , and, ( ii ) structurally , the organization of the sequence of behavior, was different between strains ., Streams of complex acts or movements exhibit some regularity that is the basis of the, subdivision of behaviors into units , or species-specific movements ., In rodents , a, variety of complex sequences of action have been identified 20 ., In our analysis we, focused on two classifications , active versus inactive , and central versus, peripheral movement ., Although simple , this classification captures two essential and, ethologically meaningful properties of the displacement ., The first is the, alternation between progressions and stops , observed in a number of locomotor, behaviors , and associated with prey search , vigilance or energy saving 21–23 ., The second concerns, the spatial distribution of movement ., Traveling close to the wall is an important, feature of the mice , and it has been suggested that the wall confers security while, the center is anxiogenic ., However , exploratory behaviors also drive the mouse to, explore all the open space ., A more precise definition of the different movements can, be performed 15 , 24 , but our coarse-grained decomposition allowed us, to focus on sequence properties , and to obtain sufficient stationary data in 30 min, experiments , for a robust statistical description of simple spontaneous decision, making ( engage in the center of the arena , stop… ) ., Analysis of behavior in terms of sequences and Markov processes has been already, applied to different species 25 ., Markov analysis assumes that the underlying, process that generates a sequence is homogeneous in time all along the sequence ., The, time range over which an event influences the future ones is supposed to be constant, ( i . e independent of the event and the sequence preceding it ) ., For this reason , fixed, length Markov chain analysis is a poor detector of sequence rules that operate only, after a particular portion of the sequence ., By contrast , VLMC allows identification, of particular sequences or contexts , such as those identified after scanning an, environment ., Modification of this homogeneity in sequences is often seen as an, indicator of higher organization such as “hierarchical” or, “grammatical” properties 26 , 27 or reflects specific, ‘decisions’ 26 ., The methodology applied in this paper is not, intended to be a blind modeling but rather a way of testing hypotheses , giving or, not significance to ‘a priori’ choices and categories ., It offers, the possibility of including ethological knowledge and previously established, categories ., It would then also be relevant and efficient also in more naturalistic, and complex settings ., The VLMC framework can be generalized so as to investigate, whether the grouping of categories in classes is relevant ., It thus proves to be, useful to improve the parsimony of the description 28 ., Hyperactivity in an open field can take different forms , including faster locomotion ,, longer periods of travel , fewer pauses , shorter pauses , etc ., The question is then, whether the reduction of the number of stops is sufficient to explain the, hyperactive profile ., Our experiments demonstrate that locomotion is not faster in, ß2−/− mice , and that the difference lies in the, patterns and organization of behaviors ., Furthermore , a simulation approach suggests, that hyperactivity cannot be explained only by changes in the matrix , or only by, changes in the duration of the various states , but by their joint effect ., Hyperactivity would then emerge from alterations of many different underlying, processes ., However , we here propose that in ß2−/− mice, hyperactivity is mainly due to the “lack of stops” ., Most, characteristics of the sequences of ß2−/− mice can be, explained by the fact that these mice do not observe certain, “stops” and that after a stop they organize their behavior, differently ., The significance of such a modification and the underlying changes it, reflects is , however , not trivial ., Open-field behavior , also called exploratory behavior or locomotor behavior in a, novel environment has been initially used as an indicator of anxiety/emotionality, 10 , 11 ., It is also used to study exploration and how, animal react to novelty , an approach with known limitations 10 , 13 , the most important, difficulty being that the various open-field measurements do not represent a single, dimension of behavior ( i . e , emotionality or exploration ) ., This limitation reinforces, the interest of using sequence analysis , which does not make any assumptions about, any underlying process , but focuses on the organization of behavior ( see also 29 ) ., Most, important features of an animals displacement organization can be, summarized as follows ( Figure, 8 ) : At the periphery , after a “stop” , the probability, that WT mice engage movement in the center of the arena is 36% ., This, probability is, ( i ) increased by “scanning” ( up to, 61% ) and, ( ii ) decreased by a recent excursion to the center ( down to, 24% ) ., In ß2−/− mice this probability is, different in baseline ( 48% ) , the increase caused by scanning disappears, and the decrease by recent incursion is similar ., These results point to information, gathering as a key element underlying differences between WT and ß2 in the, organization of sequence of behavior in an open field ., The ability to adapt to an unfamiliar or uncertain environment is fundamental , and an, essential point in adaptation would be that animals actively look for a modification, in the environment ., Displacement of an animal in a novel environment is, characterized by intermittent locomotion , scanning , and pauses that can be used to, gather information about environment but also to reduce unwanted detection by an, organisms predators 22 ., Organization of locomotor behavior in an open, environment is compatible with optimization theory insofar as it minimizes risk, while maximizing gain , i . e . collect information about environment 30 ., Fear, and anxiety tend to reduce center movement , while exploratory motivation tends to, increase these movements 24 ., Accordingly , increased probability of center, engagement after scanning may be viewed as caused by a reduction of anxiety ( Figure 8 ) ., Yet , WT and, ß2−/− mice have similar levels of anxiety 4 , 31 ,, furthermore the parallel evolution of CA → CI probability of transition, suggest that reduction of anxiety with time is similar in both strain ., The, observation that the structure of the displacement is modified in, ß2−/− mice and that this modification targets, “scanning” as a key feature in the organization of behavior, suggests instead a modification of information gathering and of the risk/gain, optimization ., The notion that exploratory behaviors in novel environments may serve, to optimize safety and that this behavior is modified in, ß2−/− mice also parallels previous observations, suggesting that WT mice react to novelty by increasing exploratory activity , whereas, ß2−/− mice do not adapt their behavior to a change in, the environment 4 ., It has been proposed that the alteration of behavioral adaptation in, ß2−/− mice , coupled with unimpaired memory and, anxiety , may model cognitive impairment observed in human disorders 4 such as, attention-deficit hyperactivity disorder ( ADHD ) 32 , or even in autism, 5 ., This proposition relies upon the idea that behavioral flexibility is controlled by, an adequate hierarchization of motivations , a process known to mobilize prefrontal, and cingulate cortex ., ADHD symptoms such as inattention lack of inhibitory control ,, and hyperactivity and prefrontal involvement indeed resemble, ß2−/− behavioral deficits , and fit well with nAChR, localization and function ., Yet , the possible contribution of prefrontal cortex and, higher-level top-down processes in open-field behaviors is at this stage not clear ., More complex environments and tasks , together with relevant methods of analyses , are, needed to explore this problem ., Further experiments are also needed to clearly, identify the brain loci and the nicotinic receptor subunits that are involved in the, modification of the behavioral patterns observed in, ß2−/− mice ., This fine-tuned analysis of the way, wild-type and mutant animals organize their spontaneous activity may ultimately help, to understand the contribution of nAChRs to higher brain functions in humans , and, the abnormalities that accompany many neuro-pathologies ., Exploratory activity was recorded in a 1-m diameter circular open-field ., Experiments were performed out of the sight of the experimenter and a video, camera , connected to a Videotrack system ( View-point , Lyon , France ) , recorded, the trajectory of the mouse for 30 minutes ., To characterize stopping behavior, ethologically , home-made softwares ( Labview , National instrument ) were used to, acquire film with a higher resolution ., Initially introduced in a purely mathematical context , symbolic dynamics has also, been developed as an efficient tool for data analysis 33 ., It provides a, framework to investigate generic features of a dynamical system from the, knowledge of experimental trajectories , in particular when only short series are, available , when individual variability is important , or when only a few features, within the recording are relevant ., The core idea is to encode continuous-valued, trajectories into behaviorally relevant symbol sequences associated with a, finite partition of the state space ., Velocity and position of the mice were used, to define a partition in four states ( or symbols ) , by combining two binary ones, ( see below ) : When combined , these symbols give four codewords or states {PA ,, PI , CA , CI} that, correspond to Activity or Inactivity in the Periphery or in the Center of the, arena ., Animal trajectories in the open-field are then represented by a sequence, of codewords ( Figure 1C ) ., The choice of a specific threshold value to partition symbols and the range of, validity of these values have been discussed and analyzed in a previous paper, ( see Supporting Information 6 ) ., The 2-D paths were smoothed using triangular filter ., The instantaneous velocity, can be then meaningfully computed from these smoothed data , simply implementing, its definition ( first time-derivative of the position ) ., Instantaneous velocity range was partitioned in two sub-ranges delineated by the, threshold θ1 ., A second threshold θ2 has, to be involved in order to faithfully assess activity , according to the, following rule:allowing to encode the continuous trajectory into a binary, sequence φv ( t ) ., In other words , it means, that crossing the low threshold θ1 can be considered as the, starting point of a significant active phase if and only if the velocity reaches, the high threshold θ2 ., This high threshold determines, qualitatively the active type of the period whereas the low threshold determines, quantitatively its duration ., This dual criterion avoids spurious alternation of, active and inactive phases of arbitrary small duration ., Indeed , since the, acceleration of the mouse is bounded above by some value amax , the, duration of an active phase is at least, ( θ2-θ1 ) /amax, hence the choice of the thresholds implicitly fixed a lowest bound on the time, scales ., In fact , a lowest bound on the time scale was also prescribed, explicitly: an additional temporal smoothing achieving a stronger masking of, fast velocity fluctuations is performed by fixing a minimal duration above or, below the low threshold to record

Introduction, Results, Discussion, Methods

Nicotinic acetylcholine receptors ( nAChRs ) are widely expressed throughout the, central nervous system and modulate neuronal function in most mammalian brain, structures ., The contribution of defined nAChR subunits to a specific behavior is, thus difficult to assess ., Mice deleted for ß2-containing nAChRs, ( ß2−/− ) have been shown to be hyperactive in an, open-field paradigm , without determining the origin of this hyperactivity ., We, here develop a quantitative description of mouse behavior in the open field, based upon first order Markov and variable length Markov chain analysis focusing, on the time-organized sequence that behaviors are composed of ., This description, reveals that this hyperactivity is the consequence of the absence of specific, inactive states or “stops” ., These stops are associated with, a scanning of the environment in wild-type mice ( WT ) , and they affect the way, that animals organize their sequence of behaviors when compared with stops, without scanning ., They characterize a specific “decision, moment” that is reduced in ß2−/− mutant, mice , suggesting an important role of ß2-nAChRs in the strategy used, by animals to explore an environment and collect information in order to, organize their behavior ., This integrated analysis of the displacement of an, animal in a simple environment offers new insights , specifically into the, contribution of nAChRs to higher brain functions and more generally into the, principles that organize sequences of behaviors in animals .

Understanding mechanisms underlying complex behaviors and the abnormalities that, accompany most neuropathologies is a current challenge in biomedical research ., A, number of approaches is primarily based on the identification of genes and their, associated molecular pathways implicated in complex motor or cognitive, pathologies ., However , optimal use of the large body of genetic , molecular ,, electro-physiological , and imaging data is hampered by the practical and, theoretical limitations of currently available behavioral analysis methods ., Complex behaviors consist of a finite number of actions combined in a variety of, spatial and temporal patterns ., In this paper we develop a sequential analysis of, mouse displacement in an open-field paradigm and demonstrate that a description, based on a Markov model can be used to describe quantitatively patterns of, behaviors and to detect changes in the way that animals organize their, displacement , especially in mice lacking nicotinic acetylcholine receptor, subunits ., This paper would be of broad interest not only to those concerned with, this particular mice model but also generally to those interested in modeling, complex behavior traits in mice .

neuroscience/behavioral neuroscience, neuroscience, neuroscience/animal cognition

journal.pcbi.1006341

Interactive implementations of thermodynamics-based RNA structure and RNA–RNA interaction prediction approaches for example-driven teaching

Bioinformatics analyses have become indispensable to biological research ., While platforms like Galaxy enable the setup of tool pipelines without expert knowledge 1 , 2 , one requires a general understanding of underlying concepts and algorithms to be able to successfully apply and adapt these pipelines to biological data 3 , 4 ., Thus , bioinformatics is taught n both computer science and biology studies ., It has been established that , when teaching mathematics , a combination of reflective example study and problem solving by hand fosters learning ., This learning effect is heightened when done iteratively with increasing difficulty 5 ., Thus , diverse examples covering different aspects of the topic have to be provided to guide the learning process ., This is even more important in an e-learning or self-study context , in which the study of examples that show different aspects of a problem might compensate for the missing interaction with a teacher 6 , 7 ., Here , we focus on RNA-related bioinformatics and especially on approaches for RNA structure and RNA–RNA interaction prediction ., Both are essential when investigating the vast amount of regulatory RNA that is common to all kingdoms of life 8 , 9 ., The function of many RNA species is guided by their structure that is defined by the formation of intramolecular base pairs ., For instance , prokaryotic small RNAs show evolutionary conserved unstructured regions that regulate the expression of their target mRNAs via intermolecular base pairing 10 , 11 ., Thus , the prediction of both functional intramolecular structures of RNAs as well as their intermolecular ( RNA–RNA ) interaction potentials are central bioinformatics tasks ., Most computational methods for RNA structure or RNA–RNA interaction prediction are based on thermodynamic models and provide an efficient computation , since Richard Bellmans principle of optimality 12 can be applied ., This means that optimal solutions of a problem can be composed of optimal solutions of ( independent ) subproblems ., This is used by dynamic programming approaches that decompose a problem into smaller problems and tabularize partial solutions ., Robert Giegerich and colleagues developed a rigorous framework , namely Algebraic Dynamic Programming ( ADP ) 13 , 14 , to systematically study and develop dynamic programming approaches in a computer science context ., In addition , they provided an online platform to study ADP programs for various problems also covering RNA related topics 15 ., The central idea of ADP is to separate the strategy of how a problem is decomposed into subproblems from the evaluation strategy , i . e . , the objective of the optimization ., We use the counting of structure alternatives for a given RNA to illustrate how dynamic programming can be applied to prediction problems ., In particular , we introduce the decomposition strategy for ( nested ) RNA structure models ., The teaching of dynamic programming approaches is typically split into a theoretical introduction by the lecturer showing individual examples and a subsequent manual application by students in which the methods are implemented or applied to solve small-scale problems for exercise ., This leads often to a very small set of examples discussed due to the high amount of work needed for manual application and the limited gain of knowledge by iterated usage of once-understood solution strategies ., To increase the number of examples , e . g . , to focus on different aspects of an individual method or to compare different approaches , either partial solutions have to be provided or implementations made available ., Besides single instances like the Nussinov algorithm , most state-of-the-art methods and their underlying algorithmic ideas are not covered by textbooks , e . g . , 16–18 ., Resorting to the original literature for teaching these algorithms , however , is complicated , as most approaches are introduced for very sophisticated energy models ., While these advanced energy models are required for a successful application of these tools in real-world scenarios , they often mask the basic and transferable algorithmic ideas for the nonexpert reader since they require a high level of background knowledge ., We approach the aforementioned problems in two ways ., First , we have stripped the model-specific energy details from the state-of-the-art methods for RNA structure prediction and RNA–RNA interaction prediction and present their underlying ( or basic ) algorithmic ideas ., For that purpose , we use the most simple energy model available ., State-of-the-art energy models take the structural context of base pairs into account ., To this end , RNA structures are decomposed into loops ( i . e . , a region that is enclosed by one or more base pairs ) to calculate their overall energy ., However , the algorithmic principles are essentially the same when using an energy model that considers base pairs without their structural context as basic units ., Since all methods are presented using the same mathematical nomenclature , relationships and differences are easy to understand ., Second , we provide a web interface that provides interactive implementations of all algorithms discussed with extensive visualizations ., This interface, ( i ) helps to understand and follow the algorithms ,, ( ii ) eases the generation of interesting examples for different aspects to teach , and, ( iii ) provides master solutions for comparison with your own calculations or implementations ., Each section closes with a list of advanced questions that exemplify what can be studied and answered using the provided web interfaces available at http://rna . informatik . uni-freiburg . de/Teaching/ ., RNA structure prediction topics covered within this manuscript are the formalization of RNA secondary structures and simplified energy models , computation of the number of structures with regards to the given model 19 , 20 , identification of the minimum free energy structure 21 , 22 , computation of partition functions 23 , probability calculation for single base pairs and unpaired regions 23 , 24 , and identification of the maximum expected accuracy structure 25 , 26 ., RNA–RNA interaction prediction approaches are grouped according to their algorithmic idea , as in 27 , into hybrid-only interaction prediction 28–30 , concatenation-based/cofolding interaction prediction 31 , 32 , and accessibility-based interaction prediction 24 , 33 , 34 ., Ribonucleic acid ( RNA ) is a linear molecule built from nucleotides ., The ribose sugars of the nucleotides are bound via interlinking phosphate groups ., Furthermore , each sugar is connected to a nitrogenous base , typically one of adenine ( A ) , guanine ( G ) , cytosine ( C ) , or uracil ( U ) ., The bases can form hydrogen bonds between two ( nonconsecutive ) nucleotides , which is then called a base pair ., Although other forms are possible , the typically considered base pairs are G−C , A−U , and G−U in both orientations ., Pairing between nucleotides of the same molecule ( intramolecular ) defines its three-dimensional structure ., In order to fulfill a certain regulatory function , typically a stable structure is needed ., Thermodynamic analyses have identified base ( pair ) stacking as the major stabilizing force within RNA structures 35 , and according energy estimates have been identified experimentally , e . g . , refer to 36 ., The functional structure of an RNA can regulate , e . g . , other RNA molecules by direct ( intermolecular ) base pairing , i . e . , forming base pairs between two RNAs , called RNA–RNA interactions ., While the probability of an initial contact is dependent on many factors , such as concentration or location , the subsequent formation of a stable RNA–RNA interaction is assumed to follow the same thermodynamic principles as single structure formation ., Thus , most ideas and parameters from RNA structure prediction are transfered to RNA–RNA interaction prediction approaches ., It is important to note that thermodynamics-based approaches are again models that do not consider all factors that influence structure/interaction formation , e . g . , already bound molecules , specific solution conditions , or kinetics of structure formation ., Nevertheless , they typically allow for accurate predictions for the majority of RNA molecules 37 ., In the following , we provide the mathematical framework needed to define and solve RNA-related problems ., The primary structure of an RNA molecule can be described by its sequence of bases ., That is , an RNA molecule of length n is defined by its sequence S∈{A , C , G , U}n of respective International Union of Pure and Applied Chemistry ( IUPAC ) single-letter codes 38 ., The secondary structure P of an RNA S is defined as a set of ( ordered ) base pairs , i . e . , P⊂1 , n×1 , n with ( i , j ) ∈P→i<j ., Typically , it is assumed that each nucleotide can pair with at most one other nucleotide , i . e . , ∀ ( i , j ) ≠ ( p , q ) ∈P:{i , j}∩{p , q} = ∅ , and that only the introduced Watson–Crick or G−U base pairs are allowed , i . e . , ∀ ( i , j ) ∈P:{Si , Sj}∈{{A , U} , {C , G} , {G , U}} extraneous to order ., Such base pairs are said to be complementary ., Furthermore , to restrict computational complexity of prediction algorithms , structures are constrained to be noncrossing ( nested ) , i . e . , ∄ ( i , j ) , ( p , q ) ∈P:i<p<j<q ., Using noncrossing structures generally allow a good estimate of the overall structure stability ., However , it is important to note that crossing base pairs do exist , albeit not as abundant as noncrossing base pairs , and contribute to the final stability of the three-dimensional shape ., It is typically assumed that first noncrossing structural elements are formed that subsequently are linked via few crossing base pairs 39 ., Thus , the majority of the structure can be modeled/predicted via nested base pairing , which strongly reduces the computational complexity ., Finally , it is commonly enforced that pairing bases have a minimal sequence distance of l , also called minimal loop length , to incorporate steric constraints of structure formation ., In the following , we will denote with P the set of all possible structures ( also referred to as structural ensemble or structure space ) that can be formed by a given sequence S . It has been shown that the size of the structure space P grows exponentially with sequence length n ., For a minimal loop length l of 3 , the growth is about 2 . 3n 40 ., Nested secondary structures can be visualized as outerplanar graphs in which nucleotides are represented by nodes , and edges represent base pairs or sequential backbone connections ., Furthermore , dot-bracket strings can be used that encode for each position i whether it is unpaired “ . ” , it is the smaller index ( opening ) of a base pair “ ( , ” or the larger ( closing ) index “ ) ” ., As motivated by Ruth Nussinov and coworkers 21 , we relate the stability of an RNA structure directly with its number of base pairs ., Since some algorithms require explicit energy contributions of individual base pairs ( e . g . , McCaskills algorithm to compute base pair probabilities ) , we set the energy of any base pair Ebp to −1 for simplification purposes ., Thus , the energy of a structure is given by E ( P ) = |P|∙Ebp ., Note , this is in stark contrast to state-of-the-art RNA structure prediction approaches ( e . g . , using Zukers algorithm 22 ) , which typically apply a Nearest Neighbor energy model 41 , 42 and experimentally derived energy contributions 36 ., Furthermore , all algorithms for RNA–RNA interaction prediction ignore concentration dependence and other factors influencing the duplex formation , which is typically modeled within the Nearest Neighbor model by an “initiation” energy term 24 , 33 , 34 ., Nevertheless , the use of the simplified base pair-focused model enables a much clearer presentation of the algorithms , which is better suited ( and sufficient ) to understanding their ideas and mechanisms ., The transfer from the simple base pair maximization to the advanced energy models , as done by Michael Zuker and Patrick Stiegler 22 , is generic and can be applied to all problems discussed within this manuscript ., References to extended versions and implementations are provided for each approach ., A first task that introduces the general structure of dynamic programming approaches used for RNA structure prediction is to compute the number of structures a sequence S can form , i . e . , |P| ., Since the structure space P grows exponentially , explicit enumeration is inefficient ., In order to apply dynamic programming , we first have to have a strategy of how to decompose such a problem into independent subproblems ., Let us consider the subsequence Si ., . Sj ., We can easily split the problem into two independent problems by introducing a case distinction for its last position Sj; case ( 1 ) Sj is not involved in any base pairing , and case ( 2 ) Sj is paired with some position Sk ( i≤k<j ) ., Both cases are depicted in Fig 1 . The first case can be easily reduced to a smaller problem , namely to Si ., . Sj−1 , since the unpaired position Sj does not allow any structural alternatives ., Thus , the reduced problem directly provides a count for case 1 . On the contrary , each possible base pairing of Sj in the second case decomposes the problem into two smaller independent problems ( one to the left of and one enclosed by the base pair ( k , j ) ) , since no base pair is allowed to cross ( k , j ) ( nestedness condition , see section on RNA secondary structure ) ., Since any structural alternative of the left subproblem can be combined with any of the enclosed ones , we have to multiply the numbers from these smaller subproblems to get the overall count for case 2 . Michael S . Waterman and Temple S . Smith applied this idea to solve the counting problem using a table C 19 , 20 ., An entry Ci , j provides the number of structures for a subsequence Si ., . Sj ., Thus , we initialize Ci , i = 1 for all positions i , since any subsequence of length one is confined to the unpaired structure ., The recursion for longer subsequences is given by, Ci , j=Ci , j−1+∑i≤k< ( j−l ) Sk , Sjcompl . Ci , k−1⋅Ck+1 , j−1, ( 1 ), which combines the two discussed cases to consider all possible “states” of nucleotide Sj in valid structures ., The first ( Ci , j−1 ) covers all cases where Sj is unpaired , and the second counts all cases where Sj is paired with an Sk within the subsequence ( second case ) ., Note , the base pair ( k , j ) has to respect the minimal loop length l ., The overall number of structures is accessed by |P|=C1 , n ., Given l and an RNA sequence , our user interface computes and depicts the filled matrix C . Example questions Ruth Nussinov and coworkers introduced in 1978 21 a first algorithm that efficiently predicts a nested structure with the maximal number of base pairs for a given RNA sequence S , i . e . , argmaxP∈P ( |P| ) ., The corresponding recursion, Ni , j=max{Ni , j−1Sjunpairedmaxi≤k< ( j−l ) Sk , Sjcompl ., ( Ni , k−1+Nk+1 , j−1+1 ) Sk , Sjpair, ( 2 ), is strongly related to the counting approach from Eq 1 . Here , an entry Ni , j stores the maximal number of base pairs that can be formed by the subsequence Si ., . Sj ., Thus , summation in Eq 1 is replaced by maximization and multiplication with summation , while the second case considers the formed base pair with “+1 . ”, N is initialized with 0 and can be filled in O ( n3 ) time while using O ( n2 ) memory ., A depiction of the recursion is given in Fig 2 . The maximal number of base pairs formed by any structure can be found in N1 , n , and a respective optimal structure P can be identified via traceback starting in N1 , n ., Thus , for a given cell Ni , j , the traceback discovers how the value of Ni , j was obtained ., To this end , the case distinctions of the ( filling ) forward recursion ( e . g . , from Eq 2 ) are considered ., If it holds Ni , j = Ni , j−1 ( first case ) , position j is found to be unpaired , and the traceback proceeds with cell Ni , j−1 ., Otherwise , position j has to form a base pair with some position i≤k<j , which is identified in accordance to the second case of Eq 2 . The base pair ( k , j ) is stored as part of the final structure P and the traceback proceeds for both subintervals represented by Ni , k−1 and Nk+1 , j−1 ., For the identification of functional structures or the study of structural alternatives , the enumeration of suboptimal structures is of interest ., A generic approach was introduced by Stefan Wuchty and coworkers 43 that enables the enumeration of all structures that are in a certain range of the minimal energy ., An implementation is also available in our web interface ., Our interactive user interface enables the computation of both optimal and suboptimal structures ., For a user defined sequence as well as recursion and traceback parameters , the dynamic programming table is provided along with a list of ( sub ) optimal structures ., On selection , the according traceback is highlighted within the matrix ., This is complemented with a graphical representation of the structure using Forna 44 ., Different recursions can be chosen to examine the effects of ambiguous recursions versus the original one ., In the following , such an ambiguous variant from 17 is presented ., Ni , j=max{Ni+1 , jSiunpairedNi , j−1SjunpairedNi+1 , j−1+1ifSi , Sjcompl . andi+l<jmaxi<k< ( j−1 ) Ni , k+Nk+1 , jdecomposition, ( 3 ), While this recursion also computes the same entries of N and thus maximal number of possible base pairs ( N1 , n ) , it is not using a unique decomposition of the structure , i . e . , the same structural variant is considered by different recursion cases ., This causes duplicated enumeration of ( sub ) optimal structures when using Wuchtys traceback algorithm , which can be studied in our web server for different recursions ., Furthermore , it is not possible to use variants of ambiguous recursions like Eq 3 to count structures ( consider relation of Eqs 2 and 1 ) or to compute the partition function of the structural ensemble ( as discussed next ) , since both requires a unique consideration of each structure ., In 1981 , Michael Zuker and Patrick Stiegler introduced a dynamic programming approach that efficiently computes minimum free energy structures using a Nearest Neighbor energy model 22 ., Using further restriction , the same time and space complexity compared to Nussinovs algorithm is kept ., The approach with according decomposition depictions and how it relates to Nussinovs algorithm is introduced in detail , e . g . , in 45 ., Implementations like UNAFold 46 ( formerly mfold 47 ) or RNAfold 31 , 37 are the current state-of-the-art tools for RNA secondary structure prediction ., Example questions To estimate the probability of a given structure P within the structural ensemble P , statistical mechanics typically dictate a Boltzmann distribution when using minimal assumptions 48 ., Thus , the probability of a structure P is directly related to its energy E ( P ) by, Pr ( P ) =exp ( −E ( P ) /kBT ) ∑P′∈Pexp ( −E ( P′ ) /kBT ), ( 4 ), given the Boltzmann factor kB and the systems temperature T . Note , when using an energy model with units “per mole , ” which is typically the case when using a Nearest Neighbor model with measured energy contributions , one has to replace kB with the gas constant R . Note further , the structure with minimal free energy , e . g . , predicted with algorithms discussed above , will always have maximal probability according to Eq 4 ., Thus , the most stable structure is automatically the most likely structure ., The nominator of Eq 4 is called Boltzmann weight ( of structure P ) ., The denominator is called canonical partition function Z , which is the sum of the Boltzmann weights of all structures in P . Since P grows exponentially , its exhaustive enumeration to compute Z is impracticable ., Nevertheless , it is possible to compute Z efficiently using a variant of the counting algorithm ., This approach was first introduced for the Nearest Neighbor energy model by John S . McCaskill ( 1990 ) 23 , and we rephrase a variant for the simplified base pair model ., First , we have to note that the Boltzmann weight of a structure P can be computed based on the energy of its base pairs Ebp , as follows, exp ( −E ( P ) /kBT ) =exp ( −∑ ( i , j ) ∈PEbp/kBT ) =∏ ( i , j ) ∈Pexp ( −Ebp/kBT ) ., ( 5 ), That is , the structures weight is computed by the product of individual base pair weights ., To simplify notation in the following , qbp = exp ( −Ebp/kBT ) refers to the Boltzmann weight of a single base pair ., Given this , we can alter the counting recursion from Eq 1 to, Qi , j=Qi , j−1+∑i≤k< ( j−l ) Sk , SjpairQi , k−1⋅Qk+1 , j−1⋅qbp ., ( 6 ), This directly provides the partition function Z =Q1 , n in O ( n3 ) time ., For some approaches and research questions , probabilities of individual base pairs Prbp ( i , j ) are of interest ., This is the probability that a base pair ( i , j ) is formed by some structure , which can be calculated by summing up the probabilities of all structures containing ( i , j ) , i . e . ,, Prbp ( i , j ) =∑P∈P ( i , j ) ∈Pexp ( −E ( P ) /kBT ) Z ., ( 7 ), As for counting , the base pair ( i , j ) decomposes all structures into the enclosed and outer subsequence that are independent concerning base pairing ., Thus , the partition functions of the according subsequences can be used to compute Prbp ( i , j ) efficiently ., To do so , we need an auxiliary matrix Qbp ., Each entry Qi , jbp holds the partition function for the subsequence Si ., . Sj , with the side constraint that i and j form the base pair ( i , j ) ., If this is not possible due to noncomplementarity or the minimal loop constraint , the entry is 0 . Given this , we can rewrite Eq 6 as follows, Qi , j=Qi , j−1+∑i≤k< ( j−l ) Qi , k−1⋅Qk , jbp, ( 8 ), Qi , jbp={Qi+1 , j−1⋅qbpifSi , Sjcomplementary0otherwise, ( 9 ), and compute the base pair probability using, Prbp ( i , j ) =Q1 , i−1⋅Qi , jbp⋅Qj+1 , nQ1 , n+∑p<i , j<qPrbp ( p , q ) ⋅qbp⋅Qp+1 , i−1⋅Qi , jbp⋅Qj+1 , q−1Qp , qbp ., ( 10 ), The first term in Eq 10 covers structures where ( i , j ) is an external base pair , i . e . , not enclosed by any other base pair ., The second term considers all structures in which ( i , j ) is directly enclosed by a base pair ( p , q ) and corrects the respective base pair probability Prbp ( p , q ) by the probability of the structure subensemble that contains both base pairs and no “in-between spanning” base pair ( k , l ) with p<k<i<j<l<q ., The latter probability is defined by the fraction within the second term ., Note ( again ) that by using a simple energy model , we omit all the complex case distinctions , which allows one to concentrate on the main cases of algorithmic importance ., In the full model , the first case would have been the same , whereas the second one would have been split to consider specifically each structural context a base pair can have ., In analogy to base pair probabilities , it is also possible to define and compute the unpaired probability Prss ( i , j ) of a subsequence Si ., . Sj ( Eq 11 ) , i . e . , the probability of all structures that show no base pairing in the single-stranded subsequence ., Prss ( i , j ) =∑P∈Pi ., . jssexp ( −E ( P ) /kBT ) Z, ( 11 ), withPi ., . jss={P|∄ ( k , l ) ∈P:k∈i , j∨l∈i , j}⊆P, ( 12 ), The unpaired probability is also sometimes termed “accessibility , ” as an unpaired region in an RNA is accessible for pairing to another RNA ., For the computation of Prss ( i , j ) , we only have to replace Qi , jbp with 1 in Eq 10 , since only the unpaired structure with energy zero has to be considered for Si ., . Sj , which has a Boltzmann weight of 1 . Stephan H . Bernhart and coworkers provide in 49 details for the extension of the introduced recursions to the Nearest Neighbor model , which is also nicely detailed in 45 ., Implementations are for instance available in the Vienna RNA package 37 ., The authors also show how to reduce the time complexity of the probability computation from O ( n4 ) to O ( n3 ) ., To this end , they introduce another auxiliary matrix Q^bp that provides the “outer” partition function , which reflects only base pairs not enclosed by respective subsequences ., Our web implementation enables the computation of both base pair probabilities as well as unpaired probabilities ., To provide insights into how the temperature and energy model influence structure and base pair probabilities , the user can alter the used temperature as well as Ebp ., Besides a visualization of the partition function tables Q and Qbp , the user is provided with a visualization of the base pair and unpaired probabilities using the established dot plot format ( e . g . , used also by UNAfold/mfold 46 , 47 or RNAfold 37 , 50 ) ., Within this matrix-like illustration , each base pair probability is represented by a dot of proportional size , i . e . , the higher the probability , the larger the dot and small probabilities are not visible ., With a bit of visual practice , dot plots enable an easy identification of highly probable substructures and the study of structural alternatives ., Example questions, The fastest class of RNA–RNA interaction prediction approaches focuses only on the identification of the interaction site , i . e . , only on the intermolecular base pairs , without considering the intramolecular structures of the interacting RNAs ., To this end , the prefix-based decomposition scheme of global sequence alignment 52 can be adapted ., Given two RNA sequences S1 and S2 of lengths n and m , respectively , we denote with S←j2 the reversely indexed S2 to simplify the index notation , since RNA molecules interact in antiparallel orientation ., The latter applies to both intra- and intermolecular base pairing ., When considering S1 and S←j2 , we can design a dynamic programming approach for the simplified energy model using a two-dimensional matrix H . An entry Hi , j will provide the maximal number of intermolecular base pairs for the prefixes S1 . ., i1 and S←1 . ., j2 ., The decomposition scheme for the recursion of Eq 16 to compute Hi , j is visualized in Fig 3 ., As already mentioned , Eq 16 is a variant of the global sequence alignment approach introduced by Saul B . Needleman and Christian D . Wunsch 52 using an adapted scoring scheme ( base pair instead of match/mismatch scoring for Si1 , S←j2 and no gap cost ) ., Thus , initializing all Hi , 0/H0 , j with 0 , the entry Hn , m provides the maximal number of intermolecular base pairs that can be formed , and a traceback starting at Hn , m yields the respective interaction details ., This approach enables very low runtimes ( O ( nm ) ) , as observed by Brian Tjaden and coworkers , who presented in 30 a variant of Eq 16 ., When computing hybridization-only interactions via minimizing a more sophisticated energy model , the strategy has to be altered to follow a scheme similar to local sequence alignment as defined by Temple Smith and Michael S . Waterman 53 , which is detailed in 30 ., The web interface of our implementation identifies and reports all optimal interaction sites ., For each , an American Standard Code for Information Interchange ( ASCII ) visualization of the intermolecular base pairs is provided ., Note , to reduce code redundancy , we do not use an implementation of Eq 16 but use a base pair-maximization variant of Eq 19 , which is discussed in the next section ., Adaptations of this approach to the Nearest Neighbor model have been discussed in 28 and , e . g . , implemented in the tools TargetRNA 30 , RNAhybrid 29 , or RNAplex 54 ., While such methods have been successfully applied for target site identification of very short RNAs , they often overestimate the length of target sites since intramolecular base pairing is ignored 33 , 54 ., These problems are tackled by concatenation- and accessibility-based approaches discussed next ., Example questions Among the first approaches to predict the interacting base pairs for two RNA molecules are concatenation-based or cofolding approaches 31 , 32 ., Here , two or more RNA sequences are concatenated into a single sequence with special interspacing linker sequences ., The resulting hybrid sequence is used within an adaptation of a standard structure prediction that takes special care of the linker sequences ., The linked sequences are forbidden to form base pairs , and the structural elements containing linker sequences are treated energetically as external , as discussed by Ivo L . Hofacker and colleagues 31 ., The extension of standard structure prediction approaches to RNA–RNA interaction prediction directly yields the possibility to compute according probabilities of interaction sites or intermolecular base pairs 55 ., A first implementation of concatenation-based prediction using the Nearest Neighbor energy model was reported for mfold 47 and later implemented in , e . g . , the tools MultiRNAFold 56 and RNAcofold 55 ., Our implementation extends the Nussinov recursion from Eq 2 with a special handling for linker sequence characters “X . ”, Base pairs ( case 2 ) are not allowed to involve a linker position ., No special energy treatment is necessary for the simplified energy model since we treat intra- and intermolecular base pairs equally and without considering their context ., The input is restricted to two RNA sequences that are concatenated by a linker of length l+1 ( where l is the minimal loop size ) to ensure the presence of a linker and that the concatenated sequence ends can form a base pair ., Our interactive cofolding web interface lists ( sub ) optimal hybridization structures using our generic suboptimal traceback implementation ., Within the reported dot-bracket strings , intramolecular base pairs are encoded using parentheses “ ( ) , ” intermolecular base pairs ( spanning the linker ) are represented by brackets “ , ” and the linker itself is depicted by linker characters “X . ”, For each hybridization structure , a traceback is visualized on selection along with a Forna 2D structure graph visualization ., Furthermore , an ASCII visualization of only the intermolecular base pairs is provided ., Concatentation-based approaches do incorporate the competition of intra- and intermolecular base pairing , which is a central weakness of hybridization-only prediction algorithms ., Still , not all important interaction patterns can be predicted using cofolding approaches since the hybrid structure has to be nested ., For instance , common kissing stem–loop or kissing–hairpin interactions cannot be predicted because they form a crossing structure in the concatenated model ( see Fig 4 ) ., To predict such patterns , accessibility-based approaches , discussed next , can be applied ., Example questions The previously introduced concatenation-based approaches directly reflect the competition of intra- and intermolecular base pairing by optimizing both at the same time ., Nevertheless , they are neglecting that the intramolecular structure is established before an intermolecular interaction is formed ., That is , intramolecular base pairs ( might ) have to be opened/broken such that intermolecular base pairs can form a stable interaction ., To be favorable , the interaction energy must outweigh the energy needed to make the subsequences accessible ., This two-step process is modeled by accessibility-based interaction prediction approaches ., The following formula , depicted in Fig 5 , is used to compute the final interaction energy values Ij , li , k that incorporate both the hybridization/duplex energy D as well as the penalties ΔE1 , ΔE2 for inaccessible sites of the RNAs S1 , S2 , respectively ., Note , ΔEj ., . l2 is computed for the reversely indexed sequence S←2 to ease the notation ., This revers

Background, Results and discussion, Maximum expected accuracy, Conclusion

The investigation of RNA-based regulation of cellular processes is becoming an increasingly important part of biological or medical research ., For the analysis of this type of data , RNA-related prediction tools are integrated into many pipelines and workflows ., In order to correctly apply and tune these programs , the user has to have a precise understanding of their limitations and concepts ., Within this manuscript , we provide the mathematical foundations and extract the algorithmic ideas that are core to state-of-the-art RNA structure and RNA–RNA interaction prediction algorithms ., To allow the reader to change and adapt the algorithms or to play with different inputs , we provide an open-source web interface to JavaScript implementations and visualizations of each algorithm ., The conceptual , teaching-focused presentation enables a high-level survey of the approaches , while providing sufficient details for understanding important concepts ., This is boosted by the simple generation and study of examples using the web interface available at http://rna . informatik . uni-freiburg . de/Teaching/ ., In combination , we provide a valuable resource for teaching , learning , and understanding the discussed prediction tools and thus enable a more informed analysis of RNA-related effects .

RNA molecules are central players in many cellular processes ., Thus , the analysis of RNA-based regulation has provided valuable insights and is often pivotal to biological and medical research ., In order to correctly select appropriate algorithms and apply available RNA structure and RNA–RNA interaction prediction software , it is crucial to have a good understanding of their limitations and concepts ., Such an overview is hard to achieve by end users , since most state-of-the-art tools are introduced on expert level and are not discussed in text books ., Within this manuscript , we provide the mathematical means and extract the algorithmic concepts that are core to state-of-the-art RNA structure and RNA–RNA interaction prediction algorithms ., The conceptual , teaching-focused presentation enables a detailed understanding of the approaches using a simplified model for didactic purposes ., We support this process by providing clear examples using the web interface of our algorithm implementation ., In summary , we have compiled material and web applications for teaching—and the self-study of—several state-of-the-art algorithms commonly used to investigate the role of RNA in regulatory processes .

computer applications, rna sequences, education, molecular probe techniques, applied mathematics, rna structure prediction, rna stem-loop structure, simulation and modeling, algorithms, mathematics, molecular biology techniques, thermodynamics, research and analysis methods, computer and information sciences, rna structure, probe hybridization, molecular biology, rna hybridization, free energy, physics, biochemistry, rna, web-based applications, nucleic acids, biology and life sciences, physical sciences, macromolecular structure analysis

journal.pgen.1002075

Epistasis between Beneficial Mutations and the Phenotype-to-Fitness Map for a ssDNA Virus

The nature of epistatic interactions between loci or mutations is a major component of evolutionary theories ., For example , epistasis is thought to have been important in the evolution of sexual reproduction 1 , 2 and reproductive isolation between incipient species 3–6 ., In models of adaptation and fitness landscapes , epistatic interactions are the primary determinant of the topology of landscape and thus the accessibility of high-fitness genotypes 7–11 ., Previous empirical studies have provided much evidence for a variety of forms of epistasis ., Compensatory mutations , whose beneficial effects depend on the presence of a deleterious mutation , provide direct evidence of the relevance of epistasis; numerous empirical examples have been described 12–17 ., Experiments in microbial 18 , 19 and viral systems 20–24 have provided abundant evidence for antagonistic epistasis , in which the total effect of multiple mutations is less than expected on the basis of their individual effects ., Similarly , some of these same studies have provided evidence for synergistic epistasis 18 , 22 , 23 , in which the combined effects of mutations are greater than expected ., Some evidence suggests that the predominance of antagonistic epistasis is a feature of simpler genomes , whereas synergistic epistasis is more common in more complex genomes 25 ., The majority of commonly cited effects of epistasis in evolution are the results of interactions between deleterious alleles , but interactions between beneficial alleles can significantly affect the rate of adaptation ., Epistasis has been shown to constrain pathways of molecular adaptation severely 26–28 ., One of the major advantages of sexual reproduction is the presumed benefit of recombining separate beneficial mutations or alleles into the same genome 2 ., Discussions of microbial evolution are dominated by the phenomenon of clonal interference 29–37 , in which , because of their asexual mode of reproduction , clonal organisms suffer a reduced rate of adaptation because individual beneficial mutations must compete for fixation rather than being combined into the same genome by recombination ., These results rest on the assumption that mutations that are individually beneficial remain beneficial when combined ., Furthermore , many models of adaptation rely on the assumption that the effects of beneficial mutations are additive 29 , 30 , 38 ., Though these assumptions are widely used , their validity is largely undetermined ., To explore epistatic interactions between beneficial mutations , we constructed bacteriophage mutants with pairs of previously identified beneficial mutations by site-directed mutagenesis ., We used nine beneficial mutations ( which we designate A through I , in order of their appearance in the genome; Table 1 ) identified for the ssDNA microvirid bacteriophage ID11 39 ., This phage infects Escherichia coli strain C , and the nine mutations increased growth rate of the wild type at in liquid culture with excess hosts ., We built 18 of the possible 36 pairs of these nine beneficial mutations ( designated by two-letter combinations ) and measured the fitnesses of the wild-type genotype , those of the single beneficial mutations , and the double mutants ., A similar approach was used to study epistatic interactions between deleterious mutations and between beneficial mutations for the RNA virus vesicular stomatitis virus ( VSV ) 22 , but we go beyond characterizing the patterns by constructing an explanatory model that posits that epistatic interactions arise at the level of the mapping from phenotypes to fitness and assessing the fit of our data to it ., For the 18 double mutants , the expected effect of incorporating both beneficial mutations into the genome under additivity ( i . e . , without epistasis ) was greater than the observed effect ( Figure 1 ) ., Because our fitness was measured as a growth rate ( i . e . , log fitness ) , the expectation under additivity was that the effect of the two mutations in combination would be the sum of the single-mutant effects on growth rate ., We can measure the deviation from additivity by calculating ( 1 ) where is the effect of the double mutant with mutations and relative to the wild type , and is the effect of single mutant relative to the wild-type ., An of 0 implies additivity; implies synergistic epistasis , and implies antagonistic epistasis 22 ., The average deviation from additivity over the 18 double mutants was ., We could easily reject additivity ( , ) ., All deviations were less than zero ( for all and ) , and the deviation of smallest magnitude , , was more than 5 standard errors less than zero ., We therefore found no evidence of synergistic epistasis between beneficial mutations and could strongly reject additivity ., Epistasis between beneficial mutations of ID11 was entirely antagonistic ., Previous work with the RNA virus VSV looking at the effects of pairs of beneficial mutations also found evidence for a predominance of antagonistic epistasis and no significant cases of synergistic epistasis for beneficial mutations ., This result confirmed the prediction by Martin et al . 40 based on a generalized version of Fishers geometrical model 41 that values of between pairs of beneficial mutations should be skewed toward negative values ( see below for a full treatment of this model ) ., Although , under antagonistic epistasis , the beneficial effect of a second mutation is reduced , that second mutation might still increase fitness to some lesser extent ., We are also therefore interested in decompensatory epistasis 22 , under which a beneficial mutation actually becomes deleterious in the presence of another beneficial mutation ( analogous to compensatory mutations , which are beneficial only in the context of a deleterious mutation ) ., Decompensatory epistasis is also a special case of sign epistasis 9 and would indicate that the set of beneficial mutations available for the wild-type genotype may be quite different from the set of beneficial mutations available after the first fixation event in adaptation ., This situation would be consistent with , for example , the standard implementation of the mutational landscape model 42–45 , which uses a random fitness landscape ., After a mutation becomes fixed in the population , an entirely new set of beneficial mutations ( if any ) becomes available to the evolving population ., Figure 2 illustrates the cases in which the mean fitness conferred by the double mutation is less than the mean fitness conferred by one or both beneficial mutations on their own ., To test for significance , we performed three different sets of tests of increasing stringency ., For the first , we simply asked whether the fitness conferred by the double mutation was significantly less than the higher of the two fitnesses conferred by the single mutations of which it was composed ., We called the situation in which it was conditional decompensatory epistasis , as it merely guaranteed that at least one of the two mutations was deleterious in the presence of the other and did not preclude the case where the double-mutant fitness lies between the two single-mutant fitnesses ., Using a one-sided Welch two-sample t-test and a Bonferroni correction for 18 tests , we found six double mutants that showed evidence of conditional decompensatory epistasis with : BE , BG , BI , CE , DI , and EI ., The second test was to determine whether the double mutant was less fit that the lower-fitness single mutant ., We refer to the case in which it was as unconditionally decompensatory epistasis , as regardless of the order mutations might be added to the genome , the second was always deleterious ., Using the same test as above , we found only two double mutants that were unconditionally decompensatory with : CE and EI ., Finally , our most stringent test was to ask whether the double mutant was less fit than the wild-type genotype ., This situation would imply that the two mutations together constituted a deleterious mutation , i . e . , a population in which both mutations became fixed would be worse off than one in which neither had ., Using the same test as above , we found two double mutants that were significantly less fit than the wild type with : CE and EI , the two unconditionally decompensatory doubles ., The presence of decompensatory epistasis for beneficial mutations is consistent with a random fitness landscape , but clearly not all pairs of beneficial mutations show this pattern ., In fact , at least one double mutant is significantly more fit than mutants bearing either of its constituent single mutations ( see below ) ., Nevertheless , in a number of cases , both beneficial mutations could not become fixed in the population because they could not outcompete one or both of the single mutations from which they were formed ., A similar observation about beneficial mutations was made for VSV 22 ., Under landscape models such as the block model 10 , 11 or model 7 , 8 , the ruggedness of the landscape can be adjusted if the extent of epistatic interactions is changed from a smooth , additive landscape with no epistasis to a highly rugged , highly epistatic random landscape ., We can clearly reject the nonepistatic model , but just as clearly , the random landscape is too extreme ., Under a random-landscape model , the probability that a second mutation increases fitness ( i . e . , is not decompensatory ) is the same as the probability that a random mutation is beneficial , which is generally assumed to be small ., Our observation of nondecompensatory mutations is therefore inconsistent with this model ., One of the major proposed advantages of sexual reproduction is that it facilitates recombination , which can increase the rate of adaptation by allowing beneficial mutations arising in different genomes to be combined in the same genome ., This advantage is contigent on the assumption of a fitness increase for the recombinant over its composite single mutations ., To test this assumption , we asked whether any of the 18 double mutants had significantly higher fitness than the higher of the fitnesses of mutants bearing the single mutations of which it was composed ., Using a one-sided Welch two-sample t-test and Bonferroni correction for 18 tests , we found only a single double mutation that could outcompete its constituent single mutations: AG ( with Bonferroni correction ) ., Even without the Bonferroni correction , only two doubles are significantly higher at the 5% significance level: AG ( ) and AH ( ) ., Therefore , recombination would not increase the rate of adaptation in this phage system ., This observation , together with the presence of decompensatory epistasis described above , indicates that the patterns predicted by clonal interference models 29 , 30 may actually arise even in the presence of recombination ., The assumption of the model is that , because of their asexual mode of reproduction , clonal organisms have a lower rate of adaptation because individual beneficial mutations must compete with one another for fixation rather than be combined into the same genome through recombination for simultaneous fixation ., If combinations of beneficial mutations confer less fitness or not more fitness than the single mutations , however , even with recombination , the single mutations must compete for fixation because of a kind of epistatic interference or epistatic repulsion ., Our results suggest that the types of theoretical results derived for asexuals have broader applicability even in sexual organisms , while at the same time calling into question the underlying impetus for the models , if similar results are found in other systems ., In other words , in our phage , sexual reproduction would provide little or no increase in the rate of adaptation , because ultimately one of the single mutants will outcompete the other singles and any double mutants that could be produced by recombination ., Clearly , our results and Figures 1 and 2 reveal significant epistatic interactions between the nine beneficial mutations in our data set ., Recent theoretical and empirical work has suggested that mutations produce additive biochemical effects 26 , 46 , and bacteriophage growth is merely a somewhat complex biochemical reaction ., If phenotypic ( e . g . , biochemical ) effects are completely additive , epistatic interactions might still arise through nonlinearity in the mapping from phenotype to fitness 40 ., In addition , work with the nine beneficial mutations we studied revealed a distinct upper bound on fitness effects for beneficial mutations 47 ., Such an upper bound could arise naturally with an intermediate phenotypic optimum ( i . e . , stabilizing selection ) ., To determine whether such a scenario might apply to the ID11 system , we developed a simple model of the phenotype-to-fitness mapping and fit it to our data ., Our model is analogous in structure to the model of Martin et al . 40 , who assumed a fitness landscape based on Fishers geometrical model 41 in a multidimensional phenotype space , additivity of phenotypic effects of mutations , and a Gaussian fitness function to map phenotypes to fitness ( see below for a comparison of the two models ) ., DePristo et al . 46 also assumed additivity of phenotypes in their model ., For our model , we assumed the phenotype-fitness relationship took the form of a gamma curve , with shape ( ) , scale ( ) , height ( ) , and shift ( ) parameters ., We also assumed that the mutations were all affecting a single underlying and unknown phenotype ., Under the model , we assumed that the phenotype of the double mutant with single mutations and with phenotypes and was given by ., We treated the phenotypes of the single mutations as missing data and imputed their values and estimated the values of the gamma parameters , , , and ., For our nine single mutants and the 18 constructed double mutants , we found that the model provides a good fit to our data ( Figure 3 ) , with a coefficient of determination ., We rejected a null model that assumed the fitnesses of the doubles and the singles to be independent draws from a normal probability distribution with giving ., The parameter estimates for the phenotype-to-fitness map were , , , and ., This distribution is right skewed and suggested that our wild-type ID11 is close to the phenotypic optimum ., Our gamma model and the model of Martin et al . 40 make similar assumptions but differ in the number of phenotypic dimensions and the shape of the phenotype-fitness map ., Martin et al . assume a Gaussian map ., To compare the performance of the models , we produced predicted distributions of epistatic effects ( Figure 4 ) ., The gamma model provided a 12 log-likelihood improvement over the model of Martin et al . but requires imputation of nine phenotypes and estimates of five parameters ( four for the gamma and one for the error distribution ) ., The model of Martin et al . has only two parameters , leaving a difference of 12 parameters ., Therefore , when Akaike Information Criterion ( AIC ) scores were used to penalize for over-fit , the two models explained the data equally well ( Figure 4 ) ., Both models predicted a pattern of negative epistatic effects , which was reflected in the data , but the model of Martin et al . predicted more extreme antagonistic epistasis than was observed ., The lack of fit for this model is due primarily to this prediction of extreme negative epistasis ., The pattern of epistasis predicted by the gamma model is consistent with the data , but this model is penalized for extra parameters ., The gamma model assumes that the phenotypic optimum is intermediate , and our fitted values suggested that five of the nine single mutants actually overshoot this optimum ., Therefore , adding two of these effects together had an overall tendency to reduce fitness , except for those mutations conferring the smallest phenotypic effects , A , D , and H ( Figure 3 ) ., Note that all cases in which the second mutation appeared to have increased fitness involved at least one of these three mutations ( Figure 2 ) ., In addition , the strongest epistatic interactions ( i . e . , those involving the unconditionally decompensatory mutations ) involved at least one of the mutations with the largest phenotypic effects , E and I ( Figure 3 ) ., Therefore , the model did explain the major patterns in our data , and it also made a number of testable predictions ., For example , we can predict which of the 18 unconstructed possible double mutants will have low or high fitness or predict the fitness of triple mutants and beyond ., To test the predictive power of the model , we conducted a series of analyses , each of which involved the removal of one of the 18 double mutants from the data set ., The model was fit to each reduced data set , then used to predict the removed value ., The model generated accurate predictions for 17 of the 18 double mutants ( Table 2 ) , suggesting good predictive power ., More interestingly , the model predicts that , if we can change the phenotypic optimum by , for example , changing the environment , we can entirely alter the patterns of epistasis ., Increasing the distance of the wild type from the optimum might produce additive effects or even synergistic epistasis rather than the uniform antagonistic effects we observed ., Intriguingly , recent work on the phage , a close relative of our phage ID11 , showed that epistatic interactions between different amino-acid residues at two particular sites in the phage coat protein can change from antagonistic to synergistic depending on the environment in which fitness is measured 23 ., Our simple model can evince such behavior in response to simple changes in the optimum ., The isolation and initial characterization of the nine beneficial mutations of the microvirid bacteriophage ID11 48 have been described in detail previously 39 , 48 ., These mutations confer an increased growth rate on the wild-type ID11 ., The isolates used were confirmed by full-genome sequencing to have the mutations of interest and no other mutations ., PCR-based construction of the double mutants was based on published techniques 23 , 49 ., Pairs were selected such that each mutation was found in multiple genotypes , and all combinations of large- , intermediate- , and small-effect mutations were included ., To construct the double mutants , we added the second mutation into a sequence-confirmed isolate of the first ., We PCR amplified the circular genome in two halves , in which the forward primer for one half and the reverse of the other had the mutation to be incorporated ., The other primers were selected to result in an overlap of the resulting genome halves ., These halves were cleaned with a Qiagen QIAquick PCR purification kit and combined in a PCR ( no primers ) ., This reaction was cleaned with the QIAquick kit and electroporated into E . coli ., The resulting plaques were picked and plaque purified by replating ., We then subjected the final isolate to full-genome sequencing to confirm the incorporation of the mutation and the lack of secondary mutations ., Fitness assays were performed as described previously 12 ., We measured fitness as the increase in the phage population per hour on E . coli strain C at ., Assays were performed in an orbital water bath shaking at 200 rpm ., We measured each genotype at least five times ( Table 3 ) ., Let be the fitness effect of mutation and let be the fitness effect of the double mutant with mutations and ., We assumed the phenotype-to-fitness mapping followed a gamma curve given byNote that this is not a probability density function ., We view as the shape parameter , as the scale parameter , as the height parameter , and as the shift parameter ., The phenotypic effect is denoted by ., Our model is then given by where is normally distributed with mean zero and variance ., Our data consisted of the fitness effects of single mutations , , and fitness effects of double mutants , ; average effects are given by and additivity of phenotypic effects was modeled on the assumption that ., For model fitting , the estimate of the shift parameter , denoted by , was based on the fitness of the lowest-fitness genotype ( see below ) ., We treated , , and as parameters and the phenotypes as missing data ., We first imputed the phenotypes and estimated the parameters from nonlinear least squares regression ., Let the array of phenotypes be ., We then minimize ( 2 ) We denote the estimates and imputations by , , , , and ., Then the predicted fitness are and ., To assess model fit , we used a simple null model where and are draws from some probability distribution and vary about some mean such that and , where follows a normal distribution with mean zero and variance ., Under this null model , the fitnesses of the single mutations and double mutations are completely independent of one another ., We can therefore consider to be our estimate of , where is the total number of mutants considered ( doubles and singles ) ., Then , the coefficient of determination isWhen was close to 1 , the model explained a large amount of the variation ., For a formal test , we used an approach analogous to an test ., The sum of squared error is defined byand the sum squared total isThe sum of squares model is then the difference ., The degrees of freedom for SST is , and the degrees of freedom for SSE is , where is the number of single mutants ., The degrees of freedom for SSM is then ., Therefore the statistic would beThe standard distribution may not hold because of the nonlinear nature of the model ., All statistical analyses were done in R 50 ., To analyze our data , we shifted all fitnesses , which are given in units of doublings per hour , by subtracting a fitness value of from each ., This shifting allowed our model to address only fitnesses in the observed range without making predictions about the phenotype-fitness relationship for very low fitness values ., Because of the simplicity of the model , it may not accurately describe the behavior far outside the range of our data ., We could not shift by the wild-type fitness because two double mutants had fitnesses below that of the wild-type , which would have given negative fitness values ., Therefore , we shifted by the largest integer value that was less than all observed fitnesses ., The degrees of freedom for SSE becomes , and the degrees of freedom for SSM becomes ., Note that the scale of the phenotypes is arbitrary , as a change in the phenotype scale can be absorbed by a change in the gamma scale parameter ., The minimization problem given by equation ( 2 ) is an dimensional problem , where is the number of single mutations ., We used the following algorithm to solve this problem ., To compare the gamma model to the model of Martin et al . 40 , we simulated the expected distributions of the deviations from additivity ( in our notation ) under the two models ., Parameter values were selected such that the two models yielded the same distributions for single beneficial mutations ., For both models , we assumed the distribution of fitness effects followed the generalized Pareto distribution ( GPD ) with shape parameter as estimated previously for the single mutations 47 , 51 ., The GPD with corresponds to a uniform distribution ., We used the maximum observed fitness of the single mutations as our estimate for the upper bound and used the smallest observed fitness for a beneficial mutation as the lower bound ., To simulate s under the gamma model , we chose nine fitness effects from the uniform distribution and mapped them to phenotypes using the inverse of the fitted gamma function ., Each fitness value could be mapped to either side of the optimum; we selected the side at random ., We assumed additivity of the phenotypes and generated the phenotypes of 18 double mutants ., Double mutants were selected to match the pattern in our empirical data ., Fitness was calculated for each on the basis of the gamma curve with normal error added from the estimated error ., Deviations from addivity were calculated as described above ., We generated 1 , 000 replicate data sets ., This model requires imputation of nine phenotypes and estimation of four gamma parameters and the error parameter ., To simulate s under the model of Martin et al . 40 , we noted that Fishers geometrical model predicts a GPD distribution of beneficial fitness effects with , where is the number of phenotypic dimensions 52 ., Therefore , the number of dimensions for our data is ., We used the same upper and lower bounds on fitness as for the gamma model and a two-dimensional geometrical model with a Gaussian phenotype-fitness map ., The wild type was assumed to be one phenotypic unit from the optimum ., Given phenotype values and , the fitness function iswhere is the fitness of the wild type , and is the difference between the maximum fitness and the wild-type fitness ., This form was selected to satisfy several constraints ., We wanted and , for simplicity , when ., The final constraint shifted the floor of the function to ; the location of this floor was not found to affect the results significantly ., To generate our distribution of deviations from additivity , we simulated nine phenotypes at random within the circle defined by the fitness of the smallest-effect mutation , created 18 double mutants by vector addition , and mapped the single and double mutants to fitness to calculate the deviations from additivity ., Double mutants were selected to match the pattern in our empirical data ., We simulated 1 , 000 replicate data sets ., This model requires the estimation of two parameters ., To compare the fit of the two models , we calculated AIC scores for each model , where ., The number of parameters for the gamma model is and for the model of Martin et al . We approximated likelihoods ( ) from the histogram densities .

Introduction, Results/Discussion, Materials and Methods

Epistatic interactions between genes and individual mutations are major determinants of the evolutionary properties of genetic systems and have therefore been well documented , but few quantitative data exist on epistatic interactions between beneficial mutations , presumably because such mutations are so much rarer than deleterious ones ., We explored epistasis for beneficial mutations by constructing genotypes with pairs of mutations that had been previously identified as beneficial to the ssDNA bacteriophage ID11 and by measuring the effects of these mutations alone and in combination ., We constructed 18 of the 36 possible double mutants for the nine available beneficial mutations ., We found that epistatic interactions between beneficial mutations were all antagonistic—the effects of the double mutations were less than the sums of the effects of their component single mutations ., We found a number of cases of decompensatory interactions , an extreme form of antagonistic epistasis in which the second mutation is actually deleterious in the presence of the first ., In the vast majority of cases , recombination uniting two beneficial mutations into the same genome would not be favored by selection , as the recombinant could not outcompete its constituent single mutations ., In an attempt to understand these results , we developed a simple model in which the phenotypic effects of mutations are completely additive and epistatic interactions arise as a result of the form of the phenotype-to-fitness mapping ., We found that a model with an intermediate phenotypic optimum and additive phenotypic effects provided a good explanation for our data and the observed patterns of epistatic interactions .

Epistasis , the extent to which the effects of a mutation depend on its genetic context , can have profound effects on the evolutionary process and strongly affects our understanding of the prevalence of sexual reproduction ., It has been investigated in a diverse array of organisms but almost exclusively for deleterious mutations ., Interactions between beneficial mutations can impede adaptation , and we therefore investigated epistasis between beneficial mutations by constructing 18 bacteriophage genomes , each with two mutations that had been previously identified as beneficial , and measuring their fitnesses ., We found universal evidence for epistasis—every pair of mutations conferred fitness lower than that expected from the single mutations alone ., In many cases , a beneficial mutation became deleterious when in combination with another , and in fact , only one pair out of 18 could be shown to confer significantly greater fitness than its constituent mutations alone ., To explain these results , we developed a model of the relationship between phenotype and fitness that posits an intermediate phenotypic optimum and assumes no epistasis at the phenotypic level ., This model fit our data well and showed that the patterns we observed could result because mutants have phenotypes that overshoot the optimum .

genetics, biology, microbiology, evolutionary biology, genetics and genomics

journal.pgen.0030072

Regional Variation in the Density of Essential Genes in Mice

In the era of complete genomes , the total number of genes in a sequenced organism can now be predicted , but the function and selective importance of a substantial fraction of genes remains unknown ., Some gene functions may be of central importance to the organism , whereas other gene functions may be useful , but not critical , or may have functions that are partially redundant ., Genes are classified as essential if an organism cannot develop to maturity without them ., Here , employing balancer chromosome mutagenesis studies on specific regions of the mouse genome , we evaluate the distribution of essential genes in these regions ., Our data also show that in mammals , similar to worms 1 , essential gene clusters are located in genomic regions with high linkage conservation ., Essential genes in two genomic regions were targeted using balancer chromosome screens: a 35-Mb region of mouse Chromosome 11 between the Trp53 and Wnt3 loci 2 and a 20-Mb region of mouse Chromosome 4 between markers D4Mit281 and D4Mit51 3 ., For comparison , we also analyzed results from an earlier mutagenesis study that identified nine essential loci in a 20-Mb deletion region on mouse Chromosome 7 4 ., In our study , we considered essential genes to be those that when mutated cause lethality at or before birth ., To improve the accuracy of the analysis , we performed pair-wise complementation tests of fully penetrant mutant lines from each screen to identify alleles at each locus ., From 785 pedigrees bred in the Chromosome 11 balancer screen , we isolated 45 mutant lines that die at or before birth ( Table 1 ) ., These 45 lines formed 40 complementation groups , and thus only five loci were detected more than once ( Table 1 ) ., From 551 pedigrees bred in the Chromosome 4 balancer screen , we isolated 16 mutant lines that die at or before birth ( Table 1 ) ., These mutants formed 12 complementation groups ( Table 1 ) ., In comparison , the deletion screen on Chromosome 7 bred 4 , 557 pedigrees to generate 24 fully penetrant lethal mutant lines that fell into nine complementation groups 4 ., Notably , only a third of the number of pedigree groups were screened on Chromosome 11 as compared to Chromosome 7 ., However , we obtained about two and a half times as many mouse lines carrying essential genes , and almost six times as many complementation groups ., To predict the number of essential genes in each chromosomal region , we employed a Bayesian approach that incorporates variation in the degree of mutability among loci to provide a credible range of values rather than a point estimate 5 ., This analysis requires knowledge of the number of complementation groups in each region , and cannot be applied to studies that fail to consider allelism ., Evidential support for gamma and mixture models that incorporate variation in mutability among loci was minimal based on the datasets alone , although previous analyses show that variation in mutability is the norm 5 ., When mutabilities vary , genes with low mutabilities tended to be under-counted if a model with a single mutability rate ( Poisson ) is assumed; the numbers of lethal mutations predicted from a Poisson distribution are therefore probably an underestimate 6 , 7 ., To obtain an accurate measurement , we considered gamma-distributed mutabilities with the shape parameter constrained to reasonable values ( a = 0 . 2–5 . 0 ) based on previous observations 5 ., There were 222 essential genes ( between 98 and 943 based on a very conservative 99% credible region ) predicted in the Chromosome 11 balancer region ( Figure S1A; Table S1 ) ., Similarly , 31 essential genes ( 16 to 124 ) were predicted in the Chromosome 4 balancer region ( Figure S1B ) ., The Chromosome 7 mutagenesis experiment was more highly saturated , with 12 essential genes estimated ( 10 to 25 , Figure S1C ) ., These three regions clearly vary considerably in their density as well as their number of essential genes ., The predicted mean density of essential genes per Mb in the Chromosome 11 balancer region is four times greater than the density on Chromosome 4 , and 11 times greater than the density on Chromosome 7 ., All density differences between chromosomes are significant , and the chromosome 11/4 density ratio is at least 2 . 26 ( p < 0 . 05 ) , while the 11/7 ratio is at least 7 . 0 ( p < 0 . 05 ) ., The number of essential genes predicted in each region is also significantly different ( p < 0 . 05 ) as a proportion of the total number of predicted genes ( 739 , 373 , and 237 , respectively ) ., The Chromosome 11 balancer region has unusually high synteny in addition to its high essential gene density: human Chromosome 17 is entirely conserved with this region of mouse Chromosome 11 , making it the most conserved mouse–human autosomal linkage group ( Figure S2 ) ., Chromosomes 4 and 7 have less synteny conservation with human chromosomes ( data not shown ) ., Although gene density ( as well as essential gene density ) is high on Chromosome 11 , we found that on other mouse chromosomes the relationship between gene density and synteny conservation was weak ( Figure S3 ) ., The number of essential genes appears to be predictive of microsynteny and sequence conservation as well as large-scale synteny ., We examined homologs among mouse , rat , human , dog , and cow to determine which genes had the same neighbors in all five species , and found that 26% of the genes on mouse Chromosome 11 had conserved microsynteny ., In contrast , only 22% of the genes on Chromosome 4 and 13% of the genes on Chromosome 7 had conserved microsynteny in all five species ( Table 1 ) ., These frequency differences are significant ( Table 1 ) ., At the sequence level , a previous comparison between the C57BL/6J and 129S5 mouse strains demonstrated that Chromosome 11 has much higher sequence conservation than Chromosomes 4 or 7 8 ., Overall , Chromosome 11 is the third most-conserved chromosome between these two strains 8 ., In this first comparative study of essential gene densities in a mammalian genome , we have identified surprising differences as large as an order of magnitude ., Our region-specific mutagenesis screens combined with complementation testing were laborious but necessary for these calculations ., Our statistical accommodation of variation in mutability , although more complex than most previous studies , allowed a more accurate assessment of the variability in essential gene density ., Sequence conservation of regions dense in essential genes is perhaps not surprising , but synteny conservation is more so ., A weak correlation between essential gene density estimates and synteny was previously observed in roundworms based on RNAi 1 , but our observations in mammals use a more precise assessment of essential function and a more definitive assessment of large-scale synteny among more species , as well as an assessment of microsynteny ., Thus , it is reasonable to consider a general causal relationship between essential genes and reduced rates of chromosomal translocation and rearrangement ., If adjacent essential genes generally reduce the probability of productive chromosomal translocations between them , essential gene-dense regions would be expected to expand over time as essential genes randomly join a cluster , but then have a reduced probability of departing ., Thus , it appears that the large number of densely packed essential genes on the balancer region of mouse Chromosome 11 may have forced it to remain as a unit in spite of millions of years of divergence and speciation ., This also predicts that syntenically conserved regions should be especially attractive targets for future essential gene detection ., It is traditional to use regional estimates of essential gene density to estimate the total number of essential genes in the genome ., If we extrapolate the number of essential genes as a proportion of predicted genes in each region , there would be 5 , 749 essential genes overall ( 20% of the genome ) ., If we extrapolate based on the density of essential genes per Mb , we predict about twice as many ( 10 , 849 ) ., The results of our own research , however , indicate that the variability on this extrapolation is huge ., If the variability of the regional estimates , as well as the variability among the regional estimates ( up to 11-fold ) , is taken into account , the estimate ranges from ∼1 , 100 essential genes up to more genes than the total predicted number of genes in the genome ( 28 , 594 ) ., It is a near certainty that such variability is not specific to our study , but applies to all previous estimates of essential genes that utilized one or a few genomic regions ., If the relationship between essential genes and synteny , particularly microsynteny , is consistently upheld in a variety of organisms , more accurate and believable estimates could be obtained by using microsynteny and conservation in essential gene predictions ., The fraction of lethal mutations remaining to be isolated from each screen was calculated using Saturate 5 ., We considered gamma-distributed mutabilities with the shape parameter constrained to reasonable values ( a = 0 . 2–5 . 0 ) based on previous observations ., For the gamma model , alpha was constrained to be less than 5 . 0 ., Genomic sequences of mouse , human , chimp , rat , cow and dog were downloaded from Ensembl v . 38 ( http://www . ensembl . org/info/data/download . html ) ., Each region of mouse sequences was divided into 150-kb fragments , which were then blasted using Megablast ( http://www . ncbi . nlm . nih . gov/BLAST/download . shtml ) ., The sequence comparison was carried out on a Sun cluster with SunFire 280R ( http://www . sun . com ) ., Mouse genomic annotation was downloaded from Ensembl BioMart v . 38 ( http://www . ensembl . org/Multi/martview ) ., To visualize the blast results , we developed in-house software written in Microsoft Visual Basic ( http://www . microsoft . com ) ., All blast results were uploaded in a MS SQL server database , and the results displayed on a PC ., Microsynteny comparisons were performed using gene annotation from Ensembl Biomart v . 38 ., A list of genes with conserved microsynteny will be provided upon request ., An explanation of calculations is found in Table S2 ., All predictions are based on protein-coding known genes found in Ensembl Biomart v . 39 ., The extremes of two distributions such that they were as similar as possible but the joint probability was no less than 5% was taken to obtain the minimal ratio of the two essential gene predictions ., In no case did the density distributions overlap with greater than 5% probability .

Introduction, Results, Discussion, Materials and Methods

In most species , and particularly in vertebrates , the percentage of genes absolutely required for survival , the essential genes , has not been estimated ., To obtain this estimation , we used the mouse as an experimental model to carry out high-efficiency N-ethyl-N-nitrosourea ( ENU ) mutagenesis screens in two balancer chromosome regions , and compared our results to a third previously published screen ., The number of essential genes in each region was predicted based on allele frequencies ., We determined that the density of essential genes differs by up to an order of magnitude among genomic regions ., This indicates that extrapolating from regional estimates to genome-wide estimates of essential genes has a huge variance ., A particularly high density of essential genes on mouse Chromosome 11 coincides with a high degree of regional linkage conservation , providing a possible causal explanation for the density variation ., This is the first demonstration of regional variation in essential gene density in the mouse genome .

The genome sequences of many organisms are now complete ., However , speculation remains regarding the function of many newly discovered genes ., There is also debate about the percentage of genes that are required to build an organism ., These genes , which are necessary for the development of the organism , are essential genes ., We have performed mutagenesis screens that allow the identification of mutations in essential genes from specific regions of the mouse genome ., From these data we have predicted the number of essential genes in three regions of the mouse genome ., When we compared these predictions , we found that the density of essential genes varies in different regions of the mouse genome ., We then analyzed these regions of the genome to identify similar regions in other mammals ., We found that regions of the mouse genome with a high density of essential genes are more similar to other species than those regions with fewer essential genes , suggesting that throughout evolution genomic regions with many essential genes remain intact .

mus (mouse), genetics and genomics

journal.pgen.1003028

Genomic Study of RNA Polymerase II and III SNAPc-Bound Promoters Reveals a Gene Transcribed by Both Enzymes and a Broad Use of Common Activators

The human pol II snRNA genes and type 3 pol III genes have the particularity of containing highly similar promoters , composed of a distal sequence element ( DSE ) that enhances transcription and a proximal sequence element ( PSE ) required for basal transcription ., In pol II snRNA promoters , the PSE is the sole essential core promoter element whereas in type 3 pol III promoters , there is in addition a TATA box , which determines RNA pol III specificity 1 , 2 ., The PSE recruits the five-subunit complex SNAPc , one of the few basal factors involved in both pol II and pol III transcription ., Basal transcription from pol II snRNA promoters requires , in addition , TBP , TFIIA , GTF2B ( TFIIB ) , TFIIF , and TFIIE , and from pol III type 3 promoters TBP , BDP1 , and a specialized GTF2B-related factor known as BRF2 3 , 4 , 5 ., The DSE is often composed of an octamer and a ZNF143 motif ( Z-motif ) that recruit the factors POU2F1 ( Oct-1 ) and ZNF143 ( hStaf ) , respectively 1 , 2 ., POU2F1 activates transcription in part by binding cooperatively with SNAPc and thus stabilizing the transcription initiation complex on the DNA ( see 6 , and references therein ) ., In addition to requiring some different basal transcription factors for transcription initiation , pol II and pol III transcription at SNAPc-recruiting promoters differ in the way transcription terminates ., In pol III genes , there are runs of T residues at various distances downstream of the RNA-coding sequence , which direct transcription termination ( 7 and references therein ) ., In pol II snRNA genes , a “3′ box” starting generally 5–20 base pairs downstream of the RNA coding sequence directs processing of the RNA , with transcription termination reported to occur either just downstream of the 3′ box 8 , or over a region of several hundreds of base pairs 9 ., Although model snRNA promoters have been extensively studied , it is unclear how broadly SNAPc is used , and to what extent the highly similar pol II and pol III PSE-containing promoters are selective in their recruitment of the polymerase ., It is also unclear how generally the use of the basal factor SNAPc is coupled to that of the activators POU2F1 and ZNF143 , and by which mechanisms ZNF143 activates transcription ., To address these questions , we performed genome-wide immunoprecipitations followed by deep sequencing ( ChIP-seq ) to localize four of the five SNAPc subunits , GTF2B , BRF2 , and a subunit of each pol II and pol III ., These studies define a set of SNAPc-dependent transcription units and show that although most loci are primarily bound by one or the other polymerase , the RPPH1 ( RNase P RNA ) gene is occupied by both enzymes ., Pol II is detectable up to 1 . 2 kb downstream of the end of the RNA-coding regions of pol II snRNA genes , thus defining a broad region of transcription termination ., Localization of POU2F1 and ZNF143 shows widespread usage of these activators by PSE-containing promoters , and we find that several of these promoters also bind the activator GABP 10 , which has not been implicated in snRNA gene transcription before ., Activators are recruited before the polymerase in G1 , and this process is less efficient when ZNF143 levels are decreased by RNAi ., We performed ChIP-seq with antibodies against SNAPC4 ( SNAPC190 ) , the largest SNAPc subunit , SNAPC1 ( SNAP43 ) , and SNAPC5 ( SNAP19 ) in IMR90Tert cells ., To localize SNAPC2 ( SNAP45 ) , we used an IMR90Tert cell line expressing both biotin ligase and SNAPC2 tagged with the biotin acceptor domain for chromatin affinity purification ( ChAP ) -seq ( see 11 ) ., We also used antibodies against GTF2B , which should mark pol II snRNA promoters , BRF2 , which should mark type 3 pol III promoters , and POLR2B ( RPB2 ) , the second largest subunit of pol II ., We used POLR3D ( RPC4 ) ChIP-seq data 11 to localize pol III ., Most of the human pol II snRNA and type 3 pol III genes are repeated and/or have given rise to large amounts of related sequences within the genome ., We therefore aligned tags as described before 11 , excluding tags aligning with one or more mismatches but including tags with several perfect matches in the genome ( see Methods ) ., We selected regions containing at least two SNAPc subunits and either BRF2 and pol III , or GTF2B and pol II , as described in Methods ., We obtained loci encompassing all known type 3 pol III genes as well as most annotated pol II snRNA genes ., In addition , we obtained a few novel loci occupied by SNAPc and pol II ., Table S1 shows these loci as well as the annotated snRNA genes that did not display any tags , namely four RNU1 and one RNU2 snRNA genes ( in red in the first column ) ., It also shows , in grey , RNU2 genes that are still in the “chr17_random” file of the human assembly and were thus not in the reference genome used for tag alignment ., In some cases , we noticed adjacent POLR2B peaks separated by only one or a few nucleotides , which often corresponded to annotated SNP positions ., Inclusion of tags aligned with ELAND , which allows for some mismatches , often resulted in the fusion of adjacent peaks , as for the SNORD13 gene shown in Figure S1A ( compare upper and lower panels ) ., Such loci are likely to be occupied by POLR2B –indeed their promoter regions are occupied by significant amounts of GTF2B and SNAPc subunits– and they are labeled in yellow in the first column of Table S1 ., In a few cases , however , this did not result in fusions of adjacent peaks , as shown in Figure S1B for a RNU1 gene ( U1-12 ) ., Such peaks probably result from attribution of tags with multiple genomic matches to an incorrect genomic location and are thus likely to be artifacts ., Consistent with this possibility , U1-11 , U1-12 , U1-like-8 , U3-2 , U3-2b , U3-4 , and U3-3 , all labeled in orange in Table S1 , had POLR2B , GTF2B , and SNAPc subunits scores with either 0% or , in the cases of U3-4 , less than 15% , unique tags ., We consider these loci unlikely to be occupied by pol II in vivo ., In contrast , the POLR2B peak on the RNU2 snRNA gene on chromosome ( chr ) 11 , even though interrupted about 500 base pairs downstream of the snRNA coding region , is constituted mostly of unique tags , as are the GTF2B and SNAPc subunit peaks ., This gene is likely , therefore , to be indeed occupied by pol II and other factors , and is labeled in striped yellow in the first column ( Table S1 ) ., We calculated occupancy scores for all loci by adding tags covering peak regions , as described in Methods ( see legend to Table S1 for exact regions ) ., We first examined the POLR2B , POLR3D , GTF2B , and BRF2 scores ., For most genes there was a clear dominance of either POLR2B and GTF2B or POLR3D and BRF2 ( Figure 1A ) ., Further , there was a good correlation between POLR2B and GTF2B ( 0 . 89 ) or POLR3D and BRF2 ( 0 . 80 ) scores , but not between POLR2B and BRF2 ( 0 . 075 ) , or POLR3D and GTF2B ( 0 . 22 ) ( Figure S2 ) ., This is consistent with GTF2B and BRF2 being specifically dedicated to recruitment of pol II and pol III , respectively , and indicates that most SNAPc-occupied genes are transcribed primarily by a single polymerase ., Strikingly , among SNAPc-occupied promoters , only thirteen loci were occupied primarily by BRF2 and pol III ( listed on top of Table S1 ) , corresponding to the known type 3 genes previously shown to be occupied by pol III in IMR90hTert and other cell lines 11 , 12 , 13 , 14 ., We identified a larger number of SNAPc-bound loci occupied primarily by GTF2B and pol II ., They included genes coding for the U1 , U2 , U4 and U5 snRNAs , all involved in splicing of pre-mRNAs; U11 , U12 , and U4atac snRNAs , which have similar functions as U1 , U2 , and U4 but participate in the removal of a smaller class of introns referred to as AT-AC introns; U7 snRNA , involved in the maturation of histone pre-mRNAs; U3 , U8 , and U13 small nucleolar RNAs ( snoRNAs ) , involved in the maturation of pre-ribosomal RNA , as well as snRNA-derived sequences ., The relationship of these loci with previously described snRNAs and snoRNA genes is described in the Results section of Text S1 ., We also uncovered a few non-annotated loci harboring SNAPc subunits , as well as GTF2B and POLR2B , peaks constituted by at least 20% of unique tags and , therefore , likely to correspond to new actively transcribed regions ., These are labeled Unknown-1 to 7 ( rows 76–82 in Table S1 ) ., As described below , these sequences harbor a PSE as well as some other sequence elements typical of pol II snRNA promoters , and contain similarities to the 3′ box ., Although most genes were occupied mostly by either BRF2 and POLR3D , or GTF2B , and POLR2B , there were a few exceptions ., The most notable was the RPPH1 gene , which is considered a type 3 pol III gene 15 but was in fact occupied not only by BRF2 and POLR3D but also by significant amounts of POLR2B and GTF2B , comparable to those found on the RNU4 snRNA genes ( Figure 1A and 1B ) ., This suggested that this gene could be transcribed in vivo by either of two RNA polymerases , pol II or pol III ., To explore this possibility further , we treated cells with a concentration of α-amanitin known to inhibit pol II but not pol III transcription 16 ., As expected , this treatment reduced the POLR2B signal of the pol II RNU2 gene but not the POLR3D signal on the pol III hsa-mi-886 gene ( Figure 1C , upper panels ) ., To determine the effects of α-amanitin for the RPPH1 gene and the U6-2 gene , which also displayed some POLR2B signal in addition to the expected POLR3D signal ( see Figure 1A ) , we set the POLR2B and POLR3D signals obtained in the absence of α-amanitin at 1 ., In each case , addition of α-amanitin to the medium reduced the POLR2B but not the POLR3D signal ( Figure 1C , lower panels ) ., Thus , the RPPH1 gene can be transcribed either by pol II or pol III in vivo ., One of the criteria used to select the genes in Table S1 was the presence of at least two of the four SNAPc subunits examined ., We obtained a good correlation between scores for the four SNAPc subunits tested ( Figure S3 ) , consistent with SNAPc binding as a single complex to snRNA promoters 17 ., Figure 2A shows the peaks obtained for the SNAPc subunits , BRF2 , GTF2B , POLR3D , and POLR2B on the pol III TRNAU1 gene and the pol II RNU4ATAC gene , and Figure 2B shows two non-annotated genomic loci occupied by POLR2B , GTF2B , and SNAPc subunits ., Whereas the polymerase subunits were detected over the entire RNA coding sequence of the corresponding genes ( and further downstream in the case of POLR2B ) , the other factors were located within the 5′ flanking region , with GTF2B and BRF2 close to , or overlapping , the TSS ., Although peaks were sometimes constituted of too few tags to allow an unambiguous determination of the peak summit location ( see for example the SNAPC4 peak in Figure 2A ) , we could nevertheless detect clear trends ., The GTF2B or BRF2 peaks were generally the closest to the TSS , the SNAPC4 , SNAPC1 , and SNAPC5 peaks were within the PSE sequence , and the SNAPC2 peak was upstream of the PSE ( Figure 2C ) ., Figure S4 shows an alignment of the PSEs and TATA boxes of the 14 pol III type 3 promoters ( including the RPPH1 gene ) , and Figure S5 an alignment of the PSEs of all pol II loci listed in Table S1 ., The non-annotated loci occupied by POLR2B and factors contain clear PSEs ., Moreover , as noted previously 1 , 2 , the PSE is located further upstream of the TSS in pol III than in pol II snRNA genes ., The corresponding LOGOs revealed similar but not identical consensus sequences for the PSEs of pol II and pol III genes ( Figure 2D ) ; for example , adenines were favored in positions 11 and 12 of pol III , but not pol II , PSEs ., Thus , although the TATA box is the dominant element specifying RNA polymerase specificity –indeed the U2 and U6 PSEs can be interchanged with no effect on RNA polymerase recruitment specificity 16– the exact PSE sequence may also contribute to specific recruitment , for example in the context of a weak TATA box ., The U1 and U2 snRNA genes are followed by a processing signal known as the 3′ box 18 , 19 , which is also found downstream of several other pol II snRNA genes 1 ., We could identify 3′ boxes in most of the pol II genes in Table S1 ., An alignment of these motifs allowed us to generate a matrix with GLAM2 20 , which we then used to search for 3′ boxes in all pol II with GLAM2SCAN 20 ., As shown in Figure S6 , we could identify putative 3′ boxes downstream of all annotated pol II genes in Table S1 ( except for the non-expressed RNU1 ( U1-9 ) and RNU1 ( U1-13 ) genes ) , as well as for the non-annotated genes ., For the RPPH1 gene , the best match to a 3′ box was located within the RNA coding sequence , from −73 to −61 relative to the end of the RNA coding sequence ( Figure S6 ) ., The resulting 3′ box LOGO derived from all sequences aligned in Figure S6 is shown in Figure 3A ., Pol II transcription termination has been reported to occur either shortly after , or several hundred base pairs downstream of , the 3′ box 8 , 9 ., Our POLR2B ChIP-seq data reveal the extent of pol II occupancy downstream of the RNA coding region ., Whereas on average , the POLR3D ChIP-seq signal dropped quite abruptly downstream of the RNA coding region of pol III genes ( see 7 ) , POLR2B could be detected as far as about 1200 base pairs past the RNA coding region of pol II snRNA genes ( Figure 3B ) ., Moreover , examination of the POLR2B peak downstream of individual pol II genes revealed a gradual decrease of tag counts over regions of 500 or more base pairs ( see for example Figure 2A and 2B , and Figure 4A below ) ., Thus , transcription termination occurs well downstream of the 3′ box and over a broad region ., snRNA promoters are characterized by an enhancer element ( DSE ) typically containing an octamer motif and a ZNF143 binding site ( Z-motif ) , which in some specific genes has been shown to recruit , respectively , the POU domain protein POU2F1 and the zinc finger protein ZNF143 ( see 1 , 2 and references therein ) ., To determine how general the binding of POU2F1 and ZNF143 is among SNAPc-binding promoters , we localized POU2F1 by ChIP-seq in HeLa cells and we analyzed ChIP-seq data obtained by others in HeLa cells ( JM , VP , and Winship Herr , personal communication ) for ZNF143 and , as ZNF143 was found to bind often together with GABP ( JM , VP , and Winship Herr , personal communication ) , for the α subunit of GABP ( GABPA ) ., The scores for all genes are listed in Table S1 and , in a summarized form , in Table S2 ., The pol III genes in Table S1 , which were all occupied by basal factors ( see above ) , were each occupied by at least one activator ., Among pol II genes , those not occupied by basal factors ( labeled in red in the first column of Tables S1 and S2 ) did not display peaks for any of the activators , and those with interrupted POLR2B peaks ( orange in the first column ) had peaks composed solely of tags with multiple matches in the genome , consistent with the possibility raised above that these genes are , in fact , not occupied by factors ., Of the genes clearly occupied by basal factors , all displayed peaks for at least one activator with three exceptions , U1-like-11 , unknown-2 , and unknown-3; these last three loci had basal factor peaks with relatively low scores and thus may bind some of these activators at levels too low to be detectable in our analysis ., Most genes had a POU2F1 peak ( 93% ) , a large majority had a ZNF143peak ( 81% ) , and about half had a GABPA peak ( 45% ) ., Interestingly , some genes had specific combinations of activators; for example the RNU5 and U5-like genes as well as most pol III genes had peaks for both POU2F1 and ZNF143 but not for GABPA ., In contrast RNU6ATAC , SNORD13 , and RNU3 genes had POU2F1 and GABPA peaks but no ZNF143 peak ., Only few genes had only one activator ( RMRP , RNY4 , RNU2-2 , U3b2-like , RNU7 , and Unknown-5 ) suggesting that most snRNA genes require some combination of the three activators tested for efficient transcription ., Indeed , altogether 23 genes had peaks for all three factors and 23 had peaks for both ZNF143 and POU2F1 but not GABPA ., Thus , the very large majority ( 79% ) of SNAPc-binding genes bound both POU2F1 and ZNF143 ., The scores for the various activators were surprisingly correlated ( see Figure S7 ) , perhaps indicating that these factors bind to snRNA promoters interdependently ., Figure 4A shows two examples ( RNU4ATAC and U1-like-5 ) with the three factors present , and two examples ( Unknown-6 and tRNAU1 ) with only POU2F1 and ZNF143 ., In all cases , the factors bound upstream of the PSE with GABP , when present , generally binding the furthest upstream ., We analyzed 5′ flanking sequences for motifs and identified POU2F1 ( octamer , see 21 ) , ZNF143 22 , 23 , and GABP 24 , 25 , 26 binding sites ( Figure 4B , Figure S8A and S8B ) ., This analysis revealed a high concordance between occupancy as determined by ChIP-seq and presence of the corresponding motif , with only a few cases ( GABP and ZNF143 for U1-like-10 , and GABP for U5E-like , U4-1 , and unknown-7 genes ) where no convincing motif could be identified ., We then aligned all occupied motifs ( see Figures S9 , S10 , and S11 ) to generate the LOGOs shown in Figure 4C , which thus reflect the ZNF143 , POU2F1 , and GABP binding sites in SNAPc-recruiting genes ., Transcription of RNU6 and probably RNU1 and RNU2 is known to be low during mitosis and to increase as cells cycle through the G1 phase 27 , 28 , 29 , 30 , 31 , hence we measured the levels of U1 , U2 , and U6 snRNA during mitosis and at several times after entry into G1 ., Since snRNA transcripts are very stable , making it difficult to measure transcription variability , we generated HeLa cell lines containing RNU1 or RNU6 reporter construct expressing unstable transcripts whose levels therefore better reflect ongoing transcription ., For U2 snRNA , we measured its precursor , which has a short half-life 16 ., Cells were blocked in prometaphase with Nocodazole and released with fresh medium ., RNA levels were low during mitosis and , in the case of the U1 reporter RNA and pre-U2 RNA , increased to a maximum 6–7 h after release , around the middle of the G1 phase ( as determined by FACS analysis , see Methods ) ., For the U6 reporter RNA , RNA levels reached a maximum 3 h after release , at the beginning of the G1 phase ( Figure 5A ) ., POLR2B occupancy was apparent 4 h after the mitosis release and peaked after 6 h , as measured by ChIP-qPCR analysis of both RNU1 and RNU2 loci ( Figure 5B ) ., This was specific , as no significant amounts of POL2RB were detected on the control region ., In comparison , increased POLR3D occupancy of RNU6 ( but not the control region ) was apparent 3 h after release and peaked after 6 h , consistent with the accumulation of U6 RNA earlier in G1 than U1 and U2 RNA ., We then examined promoter occupancy by transcription activators ( Figure 5B ) ., ZNF143 occupancy increased over time on both the RNU1 and RNU6 promoters , becoming clearly detectable at 3 h and reaching a maximum at 6 h for RNU1 and 4 h for RNU6 ., In contrast , ZNF143 was undetectable on the RNU2 promoters ., POU2F became detectable at 3 h on the RNU1 , RNU2 , and RNU6 promoters and then remained at a more or less constant level ., GABP was detected only on the RNU1 promoters and was recruited early , starting 2 h after the release and reaching a maximum at 5 h ., Thus , activators were recruited on the promoters expected from the ChIP-seq data above , with kinetics slightly faster than the polymerase ., Among activators , GABP was recruited the earliest , followed by concomitant recruitment of ZNF143 and POU2F1 ., Some basal transcription factors such as TBP are thought to remain bound to chromatin , and hence probably promoters , during mitosis 32 , 33 ., To explore whether this is the case for SNAPc , GTF2B , and BRF2 , we monitored occupancy by these factors at mitosis ( 1 h after release ) and in mid-G1 ( 7 h after release ) ., On the pol II RNU1 snRNA promoter , we observed enrichment of GTF2B and SNAPc subunits , as well as the pol II subunit POLR2B , the activators ZNF143 , POU2F1 , and GABP , and H3 acetylated on lysine 18 ( H3K18Ac ) at mid-G1 compared to mitosis ( Figure 5C , upper panel ) ., This was specific as the pol III subunit POLR3D was not enriched ., On the pol III RNU6 promoter , we observed enrichment of POLR3D , BRF2 , SNAPc subunits , ZNF143 , POU2F1 and H3K18Ac , but not POLR2B nor GABP , as expected ( Figure 5C , lower panel ) ., This suggests that at snRNA promoters , both basal transcription factors and activators are removed from promoter DNA during mitosis and are recruited de novo upon transcription activation in G1 ., To explore the role of ZNF143 in transcription factor recruitment , we targeted endogenous ZNF143 by siRNA and synchronized the cells as above ., Total protein levels measured both at mitosis and in mid-G1 were reduced by more than 70% ( Figure 6A ) , and in mid-G1 , ZNF143 bound to the U1 promoter was decreased by 50% ( Figure 5B ) ., Under these conditions , binding of the activators POU2F1 and GABP , the basal transcription factors GTF2B and SNAPC1 , and POL2RB were reduced by 40 to 70% ., In contrast , the H3K18Ac levels were not reduced ( Figure 6B ) ., Thus , ZNF143 contributes to efficient recruitment of other activators , basal transcription factors , and the RNA polymerase , but not to H3K18 acetylation , at the pol II U1 promoter ., Using stringent criteria of co-occupancy by two SNAPc subunits and either GTF2B and pol II , or BRF2 and pol III , we identified a surprisingly small number of SNAPc-occupied promoters comprising the 14 known type 3 pol III promoters , some 40 pol II snRNA genes , and 7 novel pol II-occupied loci ., It seems , therefore , that in cultured cells , SNAPc is a very specialized factor participating in the assembly of transcription initiation complexes at fewer than 100 promoters ., We have not explored , however , the possibility that some of the SNAPc subunits participate in transcription of other genes or in other functions as part of complexes other than SNAPc ., Indeed , in a previous localization of SNAPc subunits on genomic sites also binding TBP , a correlation analysis on non-CpG islands split the SNAPc subunits into two subgroups , one containing SNAPC1 and SNAPC5 and the other SNAPC2 , SNAPC3 , and SNAPC4 34 , consistent with the possibility that other SNAP -subunit-containing complexes exist ., A peculiarity of SNAPc is its involvement in transcription from both pol II and pol III promoters , promoters that differ from each other mainly by the presence or absence of a TATA box ., We found that most SNAPc-occupied promoters were predominantly occupied by either pol II or pol III with two exceptions , the U6-2 and most notably the RPPH1 genes , which were occupied not only by BRF2 and pol III , as expected , but also by levels of GTF2B and pol II comparable , in the second case , to those found on some pol II snRNA genes ., We showed that pol II occupancy of the RPPH1 gene was obliterated by levels of α-amanitin shown before to inhibit pol II transcription in cultured cells 16 ., Previous experiments comparing the 3′ ends of pol II and pol III transcripts derived from wild-type and mutated versions of the human RNU2 and RNU6 promoters have shown that pol II-synthesized transcripts end downstream of a signal referred to as the “3′ box” whereas pol III-synthesized transcripts are not processed at such boxes and instead end at runs of T residues 16 ., The best similarity to a 3′ box lies within the RPPH1 RNA coding region ., However , we detect only one type of transcript , terminated at the run of T residues downstream of the RPPH1 gene , in endogenous RNA from proliferating IMR90Tert cells ( data not shown ) , suggesting that the transcript synthesized by pol II is highly unstable , at least under the conditions tested ., It is conceivable that the ratio of RPPH1 genes transcribed by pol II and pol III , as well as the ratio of stable pol II and pol III RNA products , change in different cell types or under different conditions ., The observation that a gene can be transcribed by two different polymerase in vivo thus raises the possibility of an added layer of complexity in the regulation of gene expression ., It is not clear why the U6-2 and RPPH1 promoters are capable of recruiting significant levels of pol II ., The RPPH1 promoter has a short TATA box , but the U6-7 and U6-8 promoters have the same TATA box and are not promiscuous ., An intriguing possibility is that the presence of a 3′ box at a correct distance downstream of the TSS , together with a weak TATA box , allow pol II recruitment ., The locations of the occupancy peaks for the four SNAPc subunits we tested are remarkably consistent with what is known about the architecture and DNA binding of SNAPc ., SNAPC4 , the largest SNAPc subunit and the backbone of the complex , binds directly to the PSE through Myb repeats located in the N-terminal half of the protein 35 ., SNAPC1 and SNAPC5 associate directly with SNAPC4 , N-terminal of the Myb repeats ( aa 84–133 , see 36 ) ., Consistent with this architecture , we find that SNAPC4 , SNAPC1 , and SNAPC5 generally peak very close to each other within the PSE ., In contrast , SNAPC2 , which associates with the C-terminal part of SNAPC4 ( aa 1281–1393 , see 36 ) , peaks upstream of the PSE ., This suggests that the N-terminus of SNAPC4 is oriented facing the transcription start site whereas the C-terminal part is oriented towards the upstream promoter region ., This is consistent with the orientation of D . melanogaster SNAPC4 37 on the U1 and U6 D . melanogaster snRNA promoters as determined by elegant studies combining site-specific protein-DNA crosslinking with site-specific chemical protein cleavage ( 38 , see also 39 and references therein ) ., The 3′ end of pol II snRNAs is generated by processing at a sequence called the 3′ box 2 , 40 ., The 3′ box is efficiently used only by transcription complexes derived from snRNA promoters , suggesting that the polymerase II recruited on these promoters is somehow different from that recruited on mRNA promoters ., Indeed , the C-terminal domain of pol II associated with snRNA genes carries a unique serine 7 phosphorylation mark , which recruits RPAP2 , a serine 5 phosphatase , as well as the integrator complex , both of which are required for processing ( 41 and references therein; 42 , 43 ) ., Moreover , pol II transcription of snRNA genes requires a specialized elongation complex known as the Little Elongation Complex ( LEC ) 44 ., It has been unclear , however , how far downstream of the 3′ box processing signal transcription continues , with one report indicating a very sharp drop in transcription within 60 base pairs past the U1 3′ box 8 and another reporting continued transcription for several hundreds of base pairs downstream of the U2 3′ box 9 ., Our ChIP-seq data indicate that pol II can be found associated with the template more than 1 Kb downstream of the 3′ box , for both the RNU1 and RNU2 genes as well as all other pol II snRNA genes ., This suggests that transcription termination downstream of snRNA gene 3′ boxes does not occur at a precise location but rather over a broad 1 . 2 Kb region , and is triggered by passage of the polymerase through the processing signal , reminiscent of transcription termination downstream of the poly A signal , in this case in a region of several Kbs 45 ., Activation of several SNAPc-dependent promoters has been shown to depend on a DSE and on the binding of POU2F1 and ZNF143 ( see 1 , 2 and references therein , 23 ) ., Our ChIP-seq analyses show that POU2F1 and ZNF143 are associated with the large majority of SNAPc-dependent promoters and identify GABP as a new factor binding to a subset of these promoters ., During transcription activation in G1 , we observed binding of ZNF143 and POU2F1 preceding binding of RNA pol II and pol III , consistent with the possibility that binding of these activators prepares the promoters for polymerase recruitment ., Indeed , lowering the amount of ZNF143 by siRNA strongly affected recruitment of POU2F1 , GABPA , basal factors , and the polymerase itself on the U1 promoter ., Thus , ZNF143 could either recruit and stabilize POU2F1 by direct protein-protein contact , or affect chromatin structure to allow recruitment of POU2F1 , or both ., In support of the first hypothesis , ZFP143 , the mouse homolog of ZNF143 , recruits another POU-domain protein , Oct4 ( the mouse homolog of POU5F1 ) by direct association 46 ., On the other hand , ZNF143 and POU2F1 do not bind cooperatively to the human U6-1 promoter 47 , but then U6-1 is weakly POLR3D-occupied compared to other human RNU6 genes 11 ., In support of the second possibility , we have shown before that ZNF143 can bind to an snRNA promoter , in this case the pol III U6 snRNA promoter , preassembled into chromatin 48 , suggesting that it is an early player in the establishment of a transcription initiation complex ., However , promoter H3K18 acetylation , which is low just after mitosis and increases during G1 , was unaffected ., This suggests that SNAPc-dependent promoters are targeted very early in G1 by as yet unidentified factors that lead to histone modifications , in particular H3K18 acetylation ., It will be interesting to determine how this modification combines with the H3K4me3 mark observed on pol III promoters , including type 3 pol III promoters 12 , 13 , 14 , 49 ., ChIPs were performed as described 11 ., The antibodies used ( rabbit polyclonal antibodies except where indicated ) were as follows: POLR3D , CS682 , directed against the C-terminal 14 aa 50; POLR2B , H-201 from Santa Cruz Biotechnology; BRF2 , 940 . 505 #74; GTF2B , CS369 #10 , 11; SNAPC4 , CS696 #4 , 5; SNAPC5 , CS539 #7 , 8; SNAPC1 , CS47 #7 , 8; GABP , sc-22810 X from Santa Cruz Biotechnology; POU2F1 , mix of YL8 and YL15 51 , 52 or mix of two polyclonal antibodies ( A310-610A from Bethyl Laboratories ) ; ZNF143 , antibody 19164 raised against ZNF143 aa 623–638 , 48 ., The ChAPs have been described 11 ., The sequence tags obtained after ultra-high throughput sequencing were mapped onto the UCSC genome version Hg18 , corresponding to NCBI 36 . 2 , as before 11 except that we included tags mapping to up to 500 rather than 1000 different locations in the genome ., Table S3 shows the total number of tags sequenced for each ChIP and the percentages of tags mapped onto the genome ., In all cases , 75 . 5% or more of the total tags mapped onto the genome had unique genomic matches ., Peaks were detected with sissrs ( www . rajajothi . com/sissrs/ ) 53 with a false discovery rate set at 0 . 001% , as previously described 11 ., We identified 77312 POLR2B , 4838 GTF2B , 1366 POLR3D , and 2526 BRF2 peaks ., We then selected the POLR2B peaks within 100 base pairs of a GTF2B peak ( 3878 peaks ) , and the POLR3D peaks within 100 base pairs of a BRF2 peak ( 125 peaks ) ., The ChIPs with the anti-SNAPc subunit antibodies gave relatively weak signals ., We therefore divided the genome into 200 nucleotide bins , counted tags obtained for each of the four SNAPc subunits analyzed , and retained only bins displaying an enrichment for at least two of the SNAPc subunits ., Bins were considered positive only if the tag number in bin reached at least the minimum tag count determined by sissrs for enriched regions with a 0 . 001 false discovery rate as the one used in sissrs set at the default parameters ., We then considered genomic regions containing POLR2B and GTF2B , or POLR3D and BRF2 , sissrs peaks as well as a bin positive for two SNAPc subunits within 100 nucleotides of the polymerase sissrs peak ., We obtained 157 and 58 loci for the POLR2B and POLR3D lists , respectively , which were all visually inspected ., We eliminated peaks in regions of high backgr

Introduction, Results, Discussion, Methods

SNAPc is one of a few basal transcription factors used by both RNA polymerase ( pol ) II and pol III ., To define the set of active SNAPc-dependent promoters in human cells , we have localized genome-wide four SNAPc subunits , GTF2B ( TFIIB ) , BRF2 , pol II , and pol III ., Among some seventy loci occupied by SNAPc and other factors , including pol II snRNA genes , pol III genes with type 3 promoters , and a few un-annotated loci , most are primarily occupied by either pol II and GTF2B , or pol III and BRF2 ., A notable exception is the RPPH1 gene , which is occupied by significant amounts of both polymerases ., We show that the large majority of SNAPc-dependent promoters recruit POU2F1 and/or ZNF143 on their enhancer region , and a subset also recruits GABP , a factor newly implicated in SNAPc-dependent transcription ., These activators associate with pol II and III promoters in G1 slightly before the polymerase , and ZNF143 is required for efficient transcription initiation complex assembly ., The results characterize a set of genes with unique properties and establish that polymerase specificity is not absolute in vivo .

SNAPc-dependent promoters are unique among cellular promoters in being very similar to each other , even though some of them recruit RNA polymerase II and others RNA polymerase III ., We have examined all SNAPc-bound promoters present in the human genome ., We find a surprisingly small number of them , some 70 promoters ., Among these , the large majority is bound by either RNA polymerase II or RNA polymerase III , as expected , but one gene hitherto considered an RNA polymerase III gene is also occupied by significant levels of RNA polymerase II ., Both RNA polymerase II and RNA polymerase III SNAPc-dependent promoters use a largely overlapping set of a few transcription activators , including GABP , a novel factor implicated in snRNA gene transcription .

genome-wide association studies, gene regulation, dna transcription, genome analysis tools, genome databases, molecular genetics, gene expression, biology, molecular biology, rna, nucleic acids, gene identification and analysis, genomics, molecular cell biology, computational biology, genetics and genomics

journal.pbio.2003148

Tongue-driven sonar beam steering by a lingual-echolocating fruit bat

Adaptive sampling is a universal principle of animal sensing that has recently gained popularity in engineering practices 1 ., By selectively gathering information pertinent to behavioral goals , animals are able to efficiently parse and process information in complex environments 2 , 3 ., Echolocating animals , such as bats and toothed whales , emit sounds and analyze returning echoes to build representations of their surroundings for navigation and foraging 4 , 5 ., Their sound beams are highly directional and function as an “acoustic flashlight” to enhance information intake from the illuminated space while suppressing clutter in complex environments 6 , 7 ., Importantly , the narrow sampling volume of a highly directional beam makes it crucial to adaptively adjust the beam direction to build a comprehensive acoustic image of the environment 8 ., This is analogous to the process of consecutive foveal fixations and saccadic eye movements to scan a visual scene during search and reading 9 ., Just as movements of head , eyes , and ears are essential in adaptive acquisition of sensory information 9 , 10 , an important observation from previous studies is that adaptive changes in directional biosonar sampling are enabled by physical movements of sound-emitting structures ., Among the bat species that produce echolocation signals using their larynx ( laryngeal echolocators ) , beam control is achieved by either turning the head and changing the mouth gape for oral-emitting bats 11 or deforming specialized facial appendages ( noseleaves ) for nasal-emitting bats 12–14 ., However , this empirical principle of sonar beam control does not apply to lingual-echolocating species that produce echolocation signals with their tongue , such as the Egyptian fruit bat ( R . aegyptiacus ) ., The Egyptian fruit bat emits alternating left- and right-pointing broadband “clicks” in pairs , rather than singly , unlike most bat species 15 , 16 ., An early report suggested that each pair of clicks are produced by a series of complex movements , as the bat flicks the tongue away from the bottom of the mouth 17 ., The paired clicks point in directions as large as 60° apart , but are separated in time by only approximately 20 msec ., In addition , the inter-click angle within each pair of clicks is adaptively controlled for accurate target localization 15 , 18 ., Curiously , no clear head or mouth movements have been observed between the emission of these clicks ., How , then , do these bats achieve such efficient sonar beam direction steering ?, We hypothesize that these capabilities are directly linked to the bat’s tongue-driven click production mechanism ., Specifically , we hypothesize that the tongue motion during click generation results in rapid alternation of click directions within each pair , and that the sonar beam aim is adjusted by changing the relative position of the clicking tongue with respect to the emission aperture ( the narrowly-parted lips ) , which creates variation of phase differences across the aperture ., We predict that these properties will result in measurable features in the bat’s broadband sonar beam pattern , and that the observed beam pattern features can be captured by a theoretical model ., Similar mechanisms have been suggested for horizontal beam steering by nose-emitting laryngeal-echolocating bats 19 , 20 ., However , there has been no experimental evidence to validate this possibility in either laryngeal- or lingual-echolocating bats ., In this study , we test these predictions by combining broadband acoustic measurements from flying bats with a numerical model of biosonar beam formation to obtain insight into the mechanisms shaping sonar beam directionality of click pairs produced by the Egyptian fruit bat ., We reconstructed the biosonar beam pattern of free-flying R . aegyptiacus across the entire azimuth-elevation domain using a three-dimensional ultrasonic microphone array and synchronized high-speed video recordings ( Fig 1 and S1 Video ) ., The beam patterns of these broadband sonar clicks are significantly elongated in elevation ( Fig 2A , S1 Fig ) , with mean azimuthal and elevational −3 dB beam widths of 25 . 2° and 36 . 6° , respectively , and a mean aspect ratio of 1 . 49 ( characterized by the −3 dB best-fitting ellipse at 35 kHz , see Materials and methods ) ., The beam patterns also contain uneven azimuthal intensity distribution across all frequencies ( Fig 2A and S1 Fig ) , corroborating with previous results from narrowband measurements 15 ., In addition , the sonar beams exhibit an unusual multi-frequency structure ( Fig 2B ) : the main lobe is directed more laterally at lower frequencies and points more toward the medial axis ( directly in front of the animal ) at higher frequencies , with the center of the beam shifted by 17 . 8° and 14 . 0° from 25 to 55 kHz for the average left and right clicks , respectively ., This multi-frequency structure has not been observed in other echolocating animals , including bats and toothed whales ., Furthermore , the above features are conserved across all individual left- and right-pointing clicks and the average clicks ., Here , the average clicks were constructed to augment individual click measurements , which were sampled more coarsely in space ., Results from a series of Monte Carlo simulations suggest that our beam pattern measurements are not affected by potential spatial sampling bias ( S1 Text ) ., The beam pattern features measured in our study cannot be explained by the conventional “piston model , ” which models oral-emitting echolocation as a piston-shaped emitter in an infinite baffle 21 , 22 ., The piston model predicts a circularly varying beam intensity distribution with a concentric multi-frequency structure ( Fig 2C ) ., For decades , this prediction has been shown to be consistent with the sonar beam patterns of laryngeal-echolocating , oral-emitting bats as well as toothed whales 5 , 23 ., Although slight deviations from the piston model have been observed 24 , none exhibits the distinctly oval beam shape and asymmetric frequency-dependent main lobe direction variation shown in our data ., The frequency-dependent main lobe direction variation ( Fig 2B ) is reminiscent of the features of frequency-scanning phased sonar/radar arrays , whose beams point toward different directions in a frequency-dependent manner 25 ., Based on this insight and prior observations of the bat’s click production mechanism 4 , 17 , we investigated whether a beam formation mechanism similar to a phased array might reproduce the observed beam pattern features , and thus explain biosonar beam formation and steering in lingual echolocation ., The Egyptian fruit bat emits echolocation signals through clenched teeth and lips parted , or occasionally with teeth and lips parted , without rapid modulations in lip position ( Fig 3A and S2 Video ) ., Therefore , we model the narrow opening of the bat’s mouth as an array of transmitter elements , each emitting sound originating from the clicking tongue inside the mouth ., The elements can be discretely located , representing the gaps between clenched teeth ( Fig 3B ) , or be continuously distributed , representing the narrow gap between parted teeth and lips ( S3 Fig panel B ) ., In both scenarios , the array elements are naturally prescribed with a set of relative phase shifts , which vary depending on the distance between the elements and the tongue as well as the sound frequency ( Fig 3B ) ., These phase shifts can cause both constructive and destructive sound interference in different directions and produce a strong directivity pattern ( Fig 3C ) 25 ., Importantly , the frequency-dependent phase shifts give rise to the frequency-dependent beam direction changes observed in the experiment ., We implemented this model with boundary element method ( BEM ) 26 using a head shape obtained from micro computed tomography ( μCT ) scans of a bat specimen as well as a simplified artificial shape with dimensions comparable to the real head ( Fig 3D ) ., In both cases , the model successfully reproduces the beam pattern features shown in the data , including the vertically elongated beam shape , the distinct multi-frequency beam pattern structure , and the uneven azimuthal intensity distribution ( Fig 3E–3G , S2 Fig ) ., Specifically , the beam shape is similar between the data and the transmission array model ( p > 0 . 05 , Mann-Whitney U-test ) but significantly different between the data and the piston model ( S7 Fig panel B ) ., The beam centers of the data and the transmission array model exhibit similar frequency-dependent shifts in azimuth , whereas the beam center of the piston model varies only slightly about zero ( S7 Fig panel C ) ., In this model , only array elements on one side of the mouth transmit to produce clicks on the same side ( i . e . , left array elements produce left-pointing clicks ) ., This is a simplification supported by the observation that each pair of clicks are generated by successive detachment of the two sides of the tongue from bottom of the mouth 17 , during which the tongue itself could occlude the sound from one side to the other ., Importantly , the experimentally observed beam pattern features cannot be produced by sound transmission from both sides of the mouth ( S5 Fig ) ., To functionally support adaptive inter-click angle changes for accurate target localization , a simple steering mechanism is necessary to realize fine tuning of beam direction at a millisecond time scale 15 , 18 ., Our model provides such flexibility in a biologically-plausible and parsimonious manner: only a single parameter change ( tongue clicking location ) is required to change beam direction ( Fig 4 ) ., Here , a 25 . 7° beam direction shift is induced by a 6-mm shift in tongue location ., Importantly , the extent of beam direction changes due to tongue position changes can vary dramatically , depending on the array configuration ., For example , a much larger shift in beam direction ( 44 . 1° ) can be induced by the same tongue position variation simply by moving the array location more forward on the bat’s head ( S4 Fig ) ., We note that an important difference between our proposed beam steering mechanism and the so-called “phased array” in the radar and sonar literature is that the phase shifts at array elements are not individually controlled in our model ., It is the overall phase shift pattern across the array that is varied due to the tongue position change ., In addition , the proposed transmission array model is robust against natural morphological variations ., Results of sensitivity analysis suggest that the beam pattern features are generally conserved , irrespective of the head shape and the number and locations of array elements chosen in BEM calculation ( Fig 3D–3F , S3 and S4 Figs ) ., This indicates robustness against natural morphological variations , such as those that might occur across individuals or populations ., Our results motivate detailed comparisons between the biosonar employed by the lingual-echolocating Egyptian fruit bat and the more widely-studied laryngeal-echolocating bats ., By changing the relative phase shifts of sounds emitted along the parted lips through tongue movements ( Fig 3B ) , the sonar beam steering mechanism of this bat species is in sharp contrast with the all in-phase excitation over the gaping mouth described by the conventional piston model ., Recall that the piston model is widely used to model the sonar beam of mouth-emitting laryngeal-echolocating species 21 ., Instead , it is conceptually analogous to the potential beam steering mechanism postulated for nose-emitting laryngeal-echolocating bats 19 , 20 ., In these studies , simulations were conducted to investigate the extent to which the bats might steer the sonar beam in the azimuthal dimension by manipulating the path differences between the two nostrils to complement their control over vertical beam position via shape modification of the noseleaf ., However , unlike the well-documented inter-click sonar beam angle tuning of the Egyptian fruit bat 18 , it is not known if nose-emitting laryngeal-echolocating bats are indeed capable of such active manipulation , nor if such beam steering is important for their sonar sensing ., The unusual multi-frequency sonar beam structure of the Egyptian fruit bat raises the question of multi-frequency processing of echo returns ., It has been proposed that laryngeal-echolocating bats emitting frequency-modulated ( FM ) “chirps” may exploit the broad signal bandwidth to improve temporal processing resolution to sub-millisecond level , through a matched-filter-like processing mechanism 27 ., The Egyptian fruit bat , which uses broadband but extremely short clicks , may achieve comparable temporal acuity through simple energy detection for returns of highly transient clicks ( <100 μs ) , as has been suggested for dolphins 5 ., In addition , based on single-frequency measurements , it has been hypothesized that the sharper drop-off at the inner edges of the Egyptian fruit bat sonar beam enables optimal target localization by “pointing off-axis” using the alternating left- and right-pointing clicks 15 ., This hypothesis is in general supported by our observation , since the main lobe across all frequencies overlaps at the inner edges and is as sharp as in the single frequency case ( Fig 2B ) ., Furthermore , it has been suggested that the angle-dependent spectral filtering of sonar beam patterns , in which only the region directly in front of the bat’s aim is ensonified with the full signal bandwidth , may help suppress peripheral clutter in FM bats 28 ., A similar clutter rejection mechanism may operate in the Egyptian fruit bat as well , since its broadband sonar beam pattern exhibits similar spectral filtering effects ., Future investigations into the above questions and whether the paired clicks are processed together or individually will likely provide important insights to the functional link between lingual and laryngeal bat echolocation ., The Egyptian fruit bat belongs to the genus Rousettus , whose members are the only lingual echolocating species within the family Pteropodidae 6 ., Rousettus are frugivorous and rely heavily on visual and olfactory information to forage ., They use echolocation as an additional important sensory modality to acquire navigation information in low-light conditions 29 ., These factors imply a relatively relaxed constraint on echolocation efficiency compared to the challenges faced by insectivorous bats that need to track highly mobile prey in the dark ., Transmission array beam steering and left-/right-pointing target localization strategy of lingual echolocators may enable flexible biosonar behavior in the absence of evolutionary changes in skull morphology , such as those found in laryngeal-echolocating bats 30 ., Comparative studies of the hyoid and palate morphology between bats of the genus Rousettus and other genera of the same family may provide insights into the evolution of lingual echolocation ., A recent study uncovered evidence that at least three other species of Pteropodidae produce clicks with their wings , which supports navigation in the dark 31 ., However , sonar-based navigation capability using wing clicks is more limited in comparison to that of Rousettus using tongue clicks , and it is thus likely that wing-clicking bats rely largely on visual input ., Comparative studies of sonar-guided behaviors under varying light conditions may deepen our understanding of multisensory processing in these animals ., Through measurement and modeling of the sonar beam pattern of lingual-echolocating fruit bats , we present the first evidence that an echolocating animal produces and steers its sonar beam using phase differences passively induced across the side of the mouth ( the transmission aperture ) by changing location of the clicking tongue ., The resultant effects are similar to those of a phased array—an engineering solution widely employed in human-made sonar and radar systems that allows beam direction changes without physical modification to the transducer assembly ., Additional modeling investigations show that this sonar beam formation mechanism is parsimonious in operation and robust against morphological variations , both of which are essential to successful functioning of biological systems ., Our study highlights how engineering principles provide mechanistic insight into the operation of biological systems , and reveals an intriguing parallel between nature and human engineering ., Three Egyptian fruit bats ( R . aegyptiacus ) were used in the experiment ., In each experimental trial , the bat was released by hand from a podium , from which it flew freely through the experimental space ( Fig 1A ) ., The space ( approximate dimension: 2 . 3 m × 2 . 3 m × 2 . 4 m ) was constructed by partitioning a corner of a large room using multiple vertical railings ( Unistrut , Atkore International , Harvey , IL ) ., The setup was designed such that the bat’s echolocation and head movement during free flight can be measured at high resolution without complication of landing behavior in spaces of similar size but with solid walls ., All walls and railings were covered with acoustic foam and felt to minimize echoes in the acoustic recordings and spurious infrared reflections in the videos ., The bats had minimal prior experience with the experiment setup ., The bat’s echolocation clicks were recorded using a 34-element ultrasonic broadband microphone array ., The array was composed of 32 electret condenser microphones ( the microphones that are normally supplied as part of the D500X Ultrasound Detector/Recorder , Pettersson Elektroniks , Sweden ) and 2 precision microphones ( 40DP 1/8” , G . R . A . S . , Denmark and 7016 1/4 , ACO Pacific , Belmont , CA ) ., The signals were bandpass-filtered between 10 kHz and 100 kHz using 32 single channel filters ( USBPBP-S1 , Alligator Technology , CA ) and 1 dual channel filter ( VBF40 , Kemo Filters , Kent , UK ) ., The signals were sampled at 250 kHz through NI PXI-6143 ( National Instruments , TX ) and 1 MHz through NI PXI-6358 for sounds received by the electret microphones and the precision microphones , respectively ., The bat’s head movements were recorded using a high-speed motion capture system , consisting of 10 motion capture cameras ( Vicon T40 and T40S , Oxford , UK ) with a sampling rate of 200 frames per second ., The motion capture system was calibrated immediately prior to each experimental session ., The location and orientation of each microphone were also marked with reflective markers and recorded using this video system ., An average positioning error of 3 . 97 ± 1 . 93 cm was estimated by comparing the motion capture system measurements with results from time-of-arrival estimates given by speaker playback ., In addition , the experiments were monitored by four infrared imaging cameras ( Phantom Miro 310 , Vision Research , Wayne , NJ ) ., The experimental procedure was approved by the Institutional Animal Care and Use Committee at the Johns Hopkins University ( approval number BA14A111 ) ., The protocols are in compliance with the Animal Welfare Act regulations and Public Health Service Policy ., The university also maintains accreditation by the Association for the Assessment and Accreditation of Laboratory Animal Care International ., The bat’s head movements while echolocating and flying through the experimental space were reconstructed using reflective markers on the head stage ( Fig 1B; attached using Grimas Mastix water-soluble glue ) ., The three-dimensional track of each of the markers was smoothed to reduce jittering errors in the raw motion capture outputs before the head aim was estimated ., The bat’s head aim was calculated by the vector pointing from the mid-point of the rear two markers ( b and c ) toward the front marker ( a ) ., The direction pointing upward from the bat’s head ( the head normal ) was calculated by the cross product ba⃑×ca⃑ ., The head aim , head normal , and their cross product jointly form a bat-centered coordinate system , based on which the beam pattern is calculated ( see next section ) ., When clicks were emitted at locations where 1 or more markers were not captured by the motion capture video system , the head aim direction was approximated by the tangent to the trajectory projected onto the x–y plane , and the head normal direction was approximated by the normal vector of the floor plane ., Seventy-six percent of all clicks included in the analysis were projected using head directions derived from the reflective markers ., The mean angular error induced by using vectors derived by bat trajectories and room floor was estimated to be 23 . 7° in the case when only the azimuthal differences are considered , and 36 . 1° in the case when differences in all three dimensions are considered ( S8 Fig ) These errors are unlikely to significantly affect our results , because 1 ) the process of aligning individual clicks according to the center of the −3 dB best-fitting ellipse at 35 kHz ( see below ) would mitigate errors in bat head directions , and 2 ) the click directions with respect to the bat head direction is not the focus of analysis in this study , as the majority of our analyses were based on aligning and merging individual click measurements according to the center of the best-fitting ellipse to the −3-dB contour at 35 kHz ., The beam pattern of individual clicks was reconstructed by interpolating the energy spectral density ( ESD ) of the clicks across the microphones ., The ESD measurements were compensated with atmospheric absorption , spherical spreading loss , microphone sensitivity , and microphone receiving directionality prior to interpolation ., Radial basis function interpolation was used for the reconstruction , because it incorporates all microphone measurements to derive an expression involving radial basis kernels in predicting each interpolated value 32 ., At each click emission , the microphone positions in the “global” coordinate system of the experimental space were transformed into a bat-centered “local” spherical ( azimuth-elevation ) coordinate system for interpolation ., Reporting beam pattern in an azimuth-elevation domain around the bat 23 allows unambiguous comparison with conventional sonar/radar beam pattern measurements ., This is different from the projected area approach used in some previous two-dimensional biosonar beam pattern measurements ( e . g . , 33 ) ., In addition , five criteria were used to determine the spatial sampling quality of each individual click ( S1 Table ) ., Only clicks that satisfy all five criteria were used in the analysis ., The beam pattern was reconstructed using a custom-written open-source MATLAB package archived at https://doi . org/10 . 5281/zenodo . 56167 ., Microphone recordings from all left- and right-pointing clicks were merged to construct average beam patterns ( Fig 2A and 2B ) to augment observations from individual clicks ., Since the bats emit clicks in different azimuthal directions regularly 15 , 18 , an iterative procedure was employed to align the beam axes of all individual clicks before averaging ., The alignment was conducted by fitting an ellipse 34 to the −3 dB contour of each individual normalized beam pattern at 35 kHz under the Eckert IV map projection ., The azimuth and elevation locations of each microphone were iteratively shifted and rotated until the center of the best-fitting ellipse falls on the map origin ( S6 Fig ) ., The average click beam patterns were then constructed by interpolating over the averaged normalized beam intensity values in 10-degree bins across azimuth and elevation ., Using Monte Carlo simulations ( S1 Text ) , we showed that the average beam pattern approach is effective in extracting important features under the potential influence of spatial under-sampling ( through individual microphones ) and measurement noise ., In addition , the azimuth and elevation coordinates of the final best-fitting ellipse provide a simple yet quantitative measure of the vertically elongated beam shape ., This measure facilitates direct comparison of the beam shape between the experimental data and the models ( S7 Fig panel A and B ) ., In this study , the center of the sonar beam ( or the “beam center” ) at any given frequency is defined as the center of the best-fitting ellipse to the −3 dB normalized beam energy contour at that frequency in the azimuth-elevation domain ., This was favored as the definition of the beam center over the location of the microphone that received the maximum sound intensity , because the microphones may under-sample the spatial beam energy distribution , and the maximum intensity microphone location may not reflect the true center of the sonar beam ., We support this rationale by verifying that the averaged azimuth-elevation location of all interpolated points with normalized beam energy >− 1 dB is in general very close to the center of the best-fitting ellipse but further away from the maximum intensity microphone location ., The transmission array model was implemented using BEM , which allows the radiation sound field to be computed given a three-dimensional mesh representation of the object boundary and source locations 26 ., In this study , the source locations ( i . e . , elements of the array ) are nodes of the bat head mesh located along the mouth opening ., To calculate the model beam pattern , the complex sound fields generated by individual array elements are summed coherently 35 , with phase shifts determined by the distances between the tongue clicking location within the mouth and the elements rn , i . e . ,, Ptotal=∑n=1NPneikrn ,, ( 1 ), where Ptotal is the total pressure field , Pn is the pressure field predicted using the nth elements , and k is the acoustic wavenumber ( k = ω/c , where ω is the angular frequency and c is the sound speed in air ) ., The computation was implemented using openBEM 26 , an open-source MATLAB package available at http://www . openbem . dk ., Two mesh representations were used to implement the transmission array model: the head shape of a R . aegyptiacus specimen reconstructed using μCT scan and a highly simplified artificial shape scaled to match the dimensions of the scanned bat head ., The μCT scans were obtained from a thawed frozen specimen ., The head was fixed in 10% formalin and scanned in a Bruker Skyscan 1172 μCT scanner at a resolution of 26 μm ( 40 kV and 250 amp ) with a rotation step of 0 . 4° ., The image stack was reconstructed using NRecon ( Bruker , Massachusetts ) , segmented using Mimics ( v . 18 , Materialise , Belgium ) , and imported into Geomagic Studio 2014 ( 3D Systems , South Carolina ) for manual cleaning and smoothing ., The ears , which were folded midway along their length to accommodate the size of the scanner , were spliced along the fold points , rotated into a natural position based on photographs of live bats , and merged with mesh of the head ., The shape was subsequently re-meshed using triangles with an edge length of 0 . 5 mm ., The simplified artificial head shape was constructed using GMSH ( http://gmsh . info/ ) with manually prescribed nodes ( S2 Table ) connected by straight lines and spherical arcs and scaled to match the length , width , and height dimensions of the scanned bat head .

Introduction, Results and discussion, Materials and methods

Animals enhance sensory acquisition from a specific direction by movements of head , ears , or eyes ., As active sensing animals , echolocating bats also aim their directional sonar beam to selectively “illuminate” a confined volume of space , facilitating efficient information processing by reducing echo interference and clutter ., Such sonar beam control is generally achieved by head movements or shape changes of the sound-emitting mouth or nose ., However , lingual-echolocating Egyptian fruit bats , Rousettus aegyptiacus , which produce sound by clicking their tongue , can dramatically change beam direction at very short temporal intervals without visible morphological changes ., The mechanism supporting this capability has remained a mystery ., Here , we measured signals from free-flying Egyptian fruit bats and discovered a systematic angular sweep of beam focus across increasing frequency ., This unusual signal structure has not been observed in other animals and cannot be explained by the conventional and widely-used “piston model” that describes the emission pattern of other bat species ., Through modeling , we show that the observed beam features can be captured by an array of tongue-driven sound sources located along the side of the mouth , and that the sonar beam direction can be steered parsimoniously by inducing changes to the pattern of phase differences through moving tongue location ., The effects are broadly similar to those found in a phased array—an engineering design widely found in human-made sonar systems that enables beam direction changes without changes in the physical transducer assembly ., Our study reveals an intriguing parallel between biology and human engineering in solving problems in fundamentally similar ways .

It is well known that animals move their eyes , ears , and heads towards stimuli of interest to selectively gather information in complex environments ., Interestingly , lingual-echolocating fruit bats , which generate sonar signals for object localization by clicking their tongues , can rapidly switch the direction of the sonar beam without changing head aim or mouth shape ., The mechanism underlying this capability has intrigued scientists and engineers alike ., In this study , we used a combination of experimental measurements and theoretical modeling to solve this mystery ., We discovered that the focus of this bat’s sound beam shifts systematically across a range of angles as the sonar frequency increases ., This unusual multi-frequency structure can be captured by modeling the sound emission as an array of sound sources located along the side of the mouth and driven by the clicking tongue ., Changing only the position of the tongue in this model can steer the sonar beam in different directions , showing an effect broadly similar to that found in a human-made sonar phased array—a design that enables changing beam direction without changing the physical transducer assembly ., Our study thus reveals an intriguing parallel between biology and human engineering , which arrived at fundamentally similar solutions to the same problem .

medicine and health sciences, diagnostic radiology, engineering and technology, vertebrates, neuroscience, animals, mammals, animal signaling and communication, animal behavior, tongue, remote sensing, audio equipment, zoology, neuroimaging, fruit bats, digestive system, research and analysis methods, echolocation, sensory physiology, imaging techniques, behavior, tomography, computed axial tomography, auditory system, animal migration, microphones, eukaryota, mouth, diagnostic medicine, anatomy, radiology and imaging, equipment, animal navigation, bats, physiology, biology and life sciences, sensory systems, amniotes, organisms, sonar

journal.pcbi.0030118

Not All Scale-Free Networks Are Born Equal: The Role of the Seed Graph in PPI Network Evolution

In the past few years , protein–protein interaction ( PPI ) networks of several organisms have been derived and made publicly available ., Some of these networks have interesting topological properties ( e . g . , the degree distribution of the yeast PPI network is heavy-tailed; that is , there are a few nodes with many connections ) ., It has been argued that the degree distribution of these networks are in the form of a power law 1 , 2 ( some recent works challenge this by attributing the power law–like behavior to sampling issues , experimental errors , or statistical mistakes 3–7 ) ., Since well-known random graph models also have power-law degree distributions 8–10 , it has been tempting to investigate whether these models agree with other topological features of the PPI networks ., There are two well-known models that provide power-law degree distributions 11–13 ., The preferential attachment model 9 , 14 was introduced to emulate the growth of naturally occurring networks such as the web graph; unfortunately , it is not biologically well-motivated for modeling PPI networks ., The duplication model , on the other hand 15–17 , is inspired by Ohnos hypothesis on genome growth 18 by duplication ., Both models are iterative in the sense that they start with a seed graph and grow the network in a sequence of steps ., The degree distribution is commonly used to test whether two given networks are similar or not ., However , networks with identical degree distributions can have very different topologies ( e . g . , consider an infinite 2-D grid versus a collection of cliques of five nodes; in both cases , all nodes have a degree of four ) ., Furthermore , it was observed in 3 that given two networks with substantially different initial degree distributions , a partial ( random ) sample from those networks might give subnetworks with very similar degree distributions ., Thus , the degree distribution cannot be used as a sole measure of topological similarity ., In the recent literature , two additional measures have been used to compare PPI networks with random network models ., The first such measure is based on the k-hop reachability ., The 1-hop reachability of a node is simply its degree ( i . e . , the number of its neighbors ) ., The k-hop reachability of a node is the number of distinct nodes it can reach via a path of ≤k edges ., The k-hop reachability of all nodes whose degree is λ is the average k-hop reachability of these nodes ., Thus , the k-hop reachability ( for k = 2 , 3 , . . . ) of nodes as a function of their degree can be used to compare network topologies ., An earlier comparison of the k-hop reachability of the yeast network with networks generated by certain duplication models concluded that the two network topologies are quite different 19 ., The second similarity measure is based on the graphlet distribution ., Graphlets are small subgraphs such as triangles , stars , or cliques ., In 4 it was noted that certain “scale-free” networks are quite different from the yeast PPI network with respect to the graphlet distribution ., This observation , in combination with that on the k-hop degree distribution , seems to suggest that the known PPI networks may not be scale-free , and that existing scale-free network models may not capture the topological properties of the PPI networks ., There are other topological measures that have been commonly used in comparing social networks , etc . , but not PPI networks ., Two well-known examples are the betweenness distribution and the closeness distribution 20 ., Betweenness of a node v is the number of shortest paths between any pair of nodes u and w that pass through v , normalized by the total number of such paths ., Closeness of v is the inverse of the total distance of v to all other nodes u ., Thus , one can use betweenness and the closeness distributions , which respectively depict the number of nodes within a certain range of betweenness and closeness values that can be used to compare network topologies ., As mentioned above , scale-free network generation models such as the preferential attachment model and the duplication model can have very similar degree distributions under appropriate choice of parameters ., ( See Materials and Methods for exact definitions for the two network generation models . ), Moreover , the degree distribution of these models converge to a power-law degree distribution whose shape is determined solely by the edge deletion and edge insertion probabilities , and not by the initial “seed” graph 11 ., Hence , it has been tempting to assume that networks generated by these models are similar in general; moreover , the effect of the seed graph in shaping the topologies of these networks has largely been ignored in recent literature ., We start with the observation that two networks with very similar degree distributions may have very different topologies ., For example , a network generated by the preferential attachment and another generated by the duplication model may have very different k-hop reachability , graphlet , betweenness , and closeness distributions while having almost identical degree distributions ., Figure 1 depicts the degree distribution , k-hop reachability , and graphlet frequency of the duplication model and the preferential attachment model with 4 , 902 nodes ( as per the yeast PPI network 21 ) ., Both models start with identical seed graphs; we set r = 0 . 12 , p = 0 . 365 ( the two key parameters of the duplication model ) , and c = 7 ( the single key parameter of the preferential attachment model ) so that the average degree of nodes in both models is seven ( again as per the yeast PPI network 21 ) ., Figure 1 compares the k-hop reachability achieved by the two models for k > 1 ., As can be seen , the k-hop reachability is quite different , especially for k = 3 , 4 ., Figure 1 also shows how the graphlet distributions differ , especially for dense graphlets ( e . g . , graphlets 17–29 and 85–145 ) ., In terms of betweenness and closeness , there are some differences as well ., We now show that the seed graph has a role in characterizing the topology of the duplication model ., Figure 2 depicts how various topological features of the duplication model with fixed parameters ( p = 0 . 365 and r = 0 . 12 ) vary as the seed graph changes ., The first seed graph ( red ) is obtained by highly connecting two cliques of ten and seven nodes , respectively , by several random edges ., To reduce the average degree , some additional nodes were generated and randomly connected to one of the cliques ., The second seed graph ( blue ) is obtained by enriching a ring of 17 nodes by random connections so as to make the average degree match that of the first seed graph ., The third seed graph ( green ) is formed by sparsely connecting two cliques of ten and seven nodes , respectively , with some added nodes randomly connected to one of the cliques ., All three networks were grown until all had 4 , 902 nodes as per the yeast PPI network 21 ., ( We depict the “average behavior” of five independent runs of each of the models . ), It can be observed that although all of them have very similar degree distributions , their graphlet distributions may be quite different , especially for dense graphlets ., Figure 2 also compares the k-hop reachability , closeness , and betweenness distributions ., As can be seen , the k-hop reachability and the closeness distribution can vary considerably ., Note that both the graphlet and the closeness distributions are in logarithmic scale , and seemingly small variations in the figure may imply several factors of magnitude of a difference between the two distributions ., The key question we aim to address in this paper is the following ., If the seed selection has such an impact in shaping the topology of the network generated by the duplication model , is it possible to select the “right” seed graph so that all interesting topological features of the PPI networks in question can be captured ?, Also , is there a systematic way to determine a subgraph of a PPI network that can provide a good seed graph ?, We answer the above questions positively by demonstrating that the duplication model applied on the right seed graph can result in a network that accurately captures all key features of the PPI networks we considered ., The PPI networks we consider in this study include ( the largest connected component of ) the complete Database of Interacting Proteins ( DIP ) yeast PPI network 21 with 4 , 902 proteins and 17 , 200 edges ( as of July 2006 ) as well as the smaller but more accurate core yeast network from the DIP 22 ., We also tested the lesser-developed DIP worm network 21 ., ( See Materials and Methods for a detailed description of these networks . ), As will be demonstrated , we were able to closely approximate all the interesting topological features of these networks via the duplication model using specific seed graphs that largely exist as a subgraph in the corresponding PPI network ., A crucial observation toward obtaining the right seed graph is that the duplication model is unlikely to generate “large” cliques ( a set of nodes which are fully connected ) ., Notice that the only way to produce a clique of size h through the duplication model is starting with a clique of size h − 1 , duplicating one of its nodes , and making sure that none of the new nodes edges that are connected to the clique are deleted ., The probability of this happening is negligible for large values of h ., The size of the maximum clique in the yeast PPI network is ten nodes ., In our experiments with the duplication model , even if we started with a seed graph that included a clique of nine nodes ( but not ten ) , the chances that we ended up with a clique of ten nodes ( in <5 , 000 steps ) turned out to be negligible ., Thus , the seed graph has to include a clique of ten nodes ., We enriched the seed graph by adding to the clique of ten nodes another ( independent ) clique of seven nodes that is present in the yeast PPI network ., We also included the edges between the two cliques and some additional nodes so that the normalized degree distribution of the yeast PPI network would be similar to that of the seed graph ., The total number of nodes in the resulting seed graph was 50 ., As mentioned before , there are two key parameters associated with the duplication model: p , the edge maintenance probability; and r , the edge insertion probability ., These two parameters alone determine the ( asymptotic ) degree distribution and the average degree of the generated network ., We chose p = 0 . 365 and r = 0 . 12 so that the degree distribution of the duplication model matches that of the yeast PPI network ( see Methods and Materials for the exact mathematical expressions for p and r ) ., Also , for the preferential attachment model , we choose the value c = 7 so that the average degree of the graph created using preferential attachment would be equal to that of the yeast PPI network ., We used the duplication model and preferential attachment model described above to generate a network with 4 , 902 nodes ., The resulting networks are compared with the yeast PPI network in terms of the k-hop reachability , the graphlet , betweenness , and closeness distributions in Figure, 3 . Under all these measures , the yeast PPI network is very similar to the network produced by the duplication model ( and not similar to the network produced by the preferential attachment model ) ., In fact , the duplication model approximates both the k-hop degree distribution and the graphlet distribution of the yeast network much better than the random graph models described earlier in the literature ( 4 and 19 ) —which were specifically devised to capture the respective features of the yeast PPI network ., Another evidence of the power of the duplication model in capturing the topological features of available PPI networks is through comparing the duplication model with the main component of the core subset of the yeast network ., The core subset contains the pairs of interacting proteins identified in the yeast that were validated according to the criteria described in 22 ., It involves 2 , 345 nodes and 5 , 609 edges ., The values of r and p were set to r = 0 . 12 , p = 0 . 322 as prescribed by the average degree formula a = 2r / ( 1− PS − 2p ) and the fact that PS is a function of r and p ( see the next section for explanation ) ., The seed graph we used was very similar to that used for the complete yeast network ., Also , for the preferential attachment model , we set a value c = 4 . 8 so that the network generated using the model has the same average degree as the CORE yeast PPI network ., The results are shown in Figure, 4 . Although the yeast PPI network is the most reliable PPI network available , it is still far from completion ., Following up on 3 , we also considered the effect of sampling errors on the duplication model with respect to all the topological features used ., In order to emulate the effect of sampling and thus the ( potential ) presence of false negatives in the yeast PPI network , we used the duplication model to generate larger networks than the available ones and applied the sampling strategy proposed in 3 to “shrink” them to the size of the available networks ., The sampling strategy of 3 involves two parameters: the bait sampling probability ( the probability that a node is kept in the network during sampling ) and the edge sampling probability ( the probability that an edge of a bait is kept in the network ) ., We demonstrate the effect of sampling as per 3 on the emulation of both the full yeast and the CORE yeast PPI networks below ., We used a bait sampling probability and an edge sampling probability of 0 . 7 each ( resulting in 70% “bait coverage” and again 70% “edge coverage” ) for our emulation of the full yeast PPI network ., A comparison of the features of the resulting network with that of the full yeast PPI network is given in Figure, 5 . We then used a bait sampling probability and edge sampling probability of 0 . 5 each for emulating the core yeast PPI network ( resulting in 50% “bait coverage” and 50% “edge coverage” ) ., A comparison of the core yeast PPI network against the resulting network is given in Figure, 6 . As can be seen , the topological features of both the full yeast PPI network and the core yeast PPI can still be closely captured by the networks obtained via the duplication model , which have been subject to sampling errors ., The seed graphs used in both tests involving sampling are identical to those used in the tests that do not involve sampling ., Uniform sampling reduces the size of the maximum clique in the resulting networks significantly , as can be seen at the tail end of the graphlet distributions ., In reality , the sampling errors are not uniform ., Very dense subnetworks such as cliques are better covered by both the full yeast network and the core yeast network of the DIP ., It is interesting to note that although the core yeast network has only 5 , 609 edges in comparison to the full yeast networks 17 , 200 edges , the maximum clique size in the former is nine nodes , whereas it is ten nodes in the latter ., Perhaps the best-known PPI network database is DIP 21 , which includes the Saccharomyces cerevisiae ( yeast ) PPI network ( the best-developed PPI network available , with 4 , 902 proteins and 17 , 200 interactions ) ., DIP also includes a more accurate but much smaller core yeast network ( 2 , 345 proteins and 5 , 609 interactions ) 22 ., Our results are mainly on these two networks ., Although there are other PPI networks available through DIP 23 ( e . g . , those of the fruit fly , human , and mouse ) as well as through BIND 24 , IntAct 25 , and MINT 26 databases , they are not sufficiently well-developed to perform a conclusive analysis ., For comparison purposes , we also provide results on the DIP Caenorhabditis elegans ( worm ) PPI network ( which includes 2 , 387 proteins and 3 , 825 interactions ) as Text S1 ., The two network models we study here , namely the preferential attachment model and the duplication model , both start with a small seed graph and create an additional node in each iteration as described below ., For notational convenience , let G ( t ) = ( V ( t ) , E ( t ) ) be the network at the end of time step t , where V ( t ) is the set of nodes and E ( t ) is the set of edges/connections ., Let vt be the node generated in time step t ., Given a node vt , we denote its degree at the end of time step t by dt ( vt ) ., The preferential attachment model ( as analyzed in 9 , 11 , 14 , 27 ) generates a network as follows ., In iteration t a new node vt is generated and is connected to every other node vt in the network independently with probability, ., Here , c is the average degree of a node in G ., The duplication model ( as analyzed in 15–17 ) , in contrast , generates a network as follows ., In iteration t , an existing node vt of G ( t − 1 ) is picked uniformly at random and “duplicated” ( i . e . , an exact copy of vt as vt is generated ) ., The edge set of vt is then updated: first , each existing edge of vt is deleted independently with probability ( 1 − p ) ; then each node vt not connected to vt is connected to vt independently with probability r / |V ( t ) | ., Here , p and r are user-defined parameters ., Note that it is possible to maintain a constant average degree ( a ) throughout the generation of the network by setting r = ( 1/2 − p ) · a ., As mentioned earlier , the degree distributions of both the preferential attachment model and the duplication model asymptotically approach a power law 9 , 11 , 12 , 14 ., More specifically , the frequency of nodes with degree d is proportional to d−b , where b is a constant typically between 2 and 3 ., The value of b is solely determined ( asymptotically ) by the values of p and r in the duplication model or the value of c in the preferential attachment model ., Both the preferential attachment and the duplication model produce many singletons 13 ( i . e . , nodes that are not connected to any other node ) ., ( For example , in the duplication model where r = 0 , p = 1/2 , the proportion of singletons asymptotically approaches 1 . ), In contrast , the number of singletons in known PPI networks is very small ( this is not surprising , as genes with “no functionality” are not maintained by evolution ) ., To avoid the generation of singletons , it is possible to use a slightly modified duplication model that deletes each singleton node as soon as it is generated ., This modified duplication model has also been shown to achieve a power-law degree distribution 13 ., However , it is not known which values of p and r ensure that the expected average degree can be set to a desired value and is kept fixed through all iterations ., In this paper , we derive conditions on p and r that are necessary for having a constant expected degree ., We later use these conditions so that the modified duplication model can approximate the degree distribution of the yeast PPI network as tightly as possible ., Perhaps the ultimate way to test whether two networks are topologically similar or not is through the use of graph isomorphism as described below ., Unfortunately , graph isomorphism and approximate graph isomorphism are computationally hard problems ., Thus , it is very common to use some of the topological features of networks as a basis of checking their similarity ., In this paper , we focus on five such features: the degree distribution , the k-hop reachability , the graphlet frequency , the betweenness distribution , and the closeness distribution ., Graph isomorphism ., Two networks , G and G′ , are called isomorphic if there exists a bijective mapping F from each node of G to a distinct node in G′ , such that two nodes v and w are connected in G if and only if F, ( v ) and F, ( w ) are connected ., G and G′ are called approximately isomorphic if by removing a “small” number of nodes and edges from G and G′ , they could be made isomorphic ., Ideally , a random graph model that aims to emulate the growth of a PPI network should produce a network that is approximately isomorphic to the PPI network under investigation ., Unfortunately , the problem of approximate isomorphism is NP-complete ( through a trivial reduction from subgraph isomorphism—a known NP-complete problem ) ; thus , this measure cannot be used to practically test similarity of two networks ., k-hop reachability ., Let V ( i ) denote the set of nodes in G whose degree is i ., Given a node v , denote by d ( v , k ) its k-hop degree ( i . e . , the number of distinct nodes it can reach in ≤k-hops ) ., Now we define f ( i , k ) , the k-hop reachability of V ( i ) , as, Note that f ( i , k ) is the “average” number of distinct nodes a node with degree i can reach in k-hops ( e . g . , f ( i , 1 ) = i by definition ) ., Graphlet frequency ., The graphlet frequency was introduced in 4 to compare the topological structure of networks ., A graphlet is a small connected induced subgraph of a large graph ( e . g . , a triangle or a clique ) ., The graphlet count of a given graphlet g with r nodes in a given graph G = ( V , E ) is defined as the number of distinct subsets of V ( with r nodes ) whose induced subgraphs in G are isomorphic to g ., In this paper , we consider all 141 possible graphlets/subgraph topologies with three , four , five , and six nodes ., In addition , we consider cliques of sizes seven , eight , nine , and ten ., We enumerate these graphlets as shown in the final figure in Text S2 ., Betweenness distribution ., The betweenness of a fixed node of a network measures the extent to which a particular point lies “between” point pairs in the network G = ( V , E ) ., The formal definition of betweenness is as follows ., Let sx , y be the number of the shortest path from x ∈ V to y ∈ V for all pairs x , y ∈ V . ( Note that sx , y = sy , x in undirected graphs . ) Let sx , y, ( v ) be the number of shortest paths from x ∈ V to y ∈ V which go through node v . The betweenness Bet, ( v ) of node v is now defined as, Closeness ., For all x , y ∈ V , we define dx , y as the length of the shortest path between x and y ., The closeness of a node v ∈ V is defined as, Thus , closeness of a node v is simply the inverse of the average distance of v to all other nodes in G ., The network comparison methods in use: The yeast PPI network versus the Erdos–Renyi random graph model ., The network features described above can be used to test whether a given random graph model can emulate an available PPI network ., Here , we consider the standard Erdos–Renyi random graph model 28 in comparison to the yeast PPI network ., As shown in Figure 7 , each of the features we consider point to significant differences between yeast PPI ( red ) and ( five independent runs of ) the Erdos–Renyi ( green ) model ., In this section , we show how to determine the deletion probability 1 − p with respect to the insertion probability r so that the expected average degree of the network can be set to any given value ., For this , we make the assumption that the degree frequency distribution and the average degree of nodes are fixed asymptotically once the values of p and r are determined ., Let G ( t ) = ( V ( t ) , E ( t ) ) be the network generated by the modified duplication model and let n ( t ) = |V ( t ) | and e ( t ) = |E ( t ) | ., Also , let nk ( t ) be the number of nodes in time step t with degree k and a ( t ) be the average degree of nodes in G ( t ) ., Finally , let Pk ( t ) = nk ( t ) / n ( t ) , the frequency of nodes with degree k at time step t ., We assume that Pk ( t ) is asymptotically stable ( i . e . , Pk ( t ) = Pk ( t + 1 ) for all 1 ≤ k ≤ t for sufficiently large values of t ., In other words , we assume that Pk ( t ) = dk for some fixed dk ., By definition:, Now we can calculate the average degree a ( t + 1 ) under the condition that degree frequency distribution is stable and a ( t ) = a , a constant ., Let Prs ( t ) be the probability that vt+1 ends up as a singleton ., Since this probability does not depend on t asymptotically , we can set Prs ( t ) = Prs ., Now we can calculate the expected number of nodes and the expected number of edges in step t + 1 ., Comparing the above equation with the first equation for Expe ( t + 1 ) , we get, Solving Equation 10 results in a = 2r / ( 1 − Prs − 2p ) , where Prs is a function of p , r , and dk only ., The discussion above demonstrates that the two key parameters p and r of the ( modified ) duplication model are determined by the degree distribution and the average degree of the PPI network we would like to emulate .

Introduction, Results/Discussion, Materials and Methods

The ( asymptotic ) degree distributions of the best-known “scale-free” network models are all similar and are independent of the seed graph used; hence , it has been tempting to assume that networks generated by these models are generally similar ., In this paper , we observe that several key topological features of such networks depend heavily on the specific model and the seed graph used ., Furthermore , we show that starting with the “right” seed graph ( typically a dense subgraph of the protein–protein interaction network analyzed ) , the duplication model captures many topological features of publicly available protein–protein interaction networks very well .

The interactions among proteins in an organism can be represented as a protein–protein interaction ( PPI ) network , where each protein is represented with a node , and each interaction is represented with an edge between two nodes ., As PPI networks of several model organisms become available , their topological features attract considerable attention ., It is believed that the available PPI networks are ( 1 ) “small-world” networks , and ( 2 ) their degree distribution is in the form of a “power law . ”, In other words , ( 1 ) it is possible to reach from a protein to any other protein in only a small ( approximately six ) number of hops , and ( 2 ) although most proteins have only a few interactions ( one or two ) , there are a few proteins with many more interactions ( 200 or more ) and that act as “hubs . ”, It has thus been tempting to develop simple mathematical network generators with topological features similar to those of the available PPI networks ., One such model , the “duplication model , ” is based on Ohnos model of genome growth ., It starts with a small “seed network” and grows by “duplicating” one of the existing nodes at a time , with an identical set of interactions; a randomly selected subset of these interactions is then deleted , and a few new interactions are added at random ., It has been mathematically proven that the duplication model provides a small-world network and also has a power-law degree distribution ., What we show in this paper is that by choosing the “right” seed network , many other topological features of the available PPI networks can be captured by the duplication model ., The right seed network in this case turns out to include two sizable “cliques” ( subnetworks where all node pairs are connected ) with many interactions in between ., In this paper , we also consider the preferential attachment model , which again grows by adding to a seed network one node at a time and connecting the new node to every other node with probability proportional to the existing degree of the second node ., Because the preferential attachment model also provides a small-world network and has a power-law degree distribution , it has been considered equivalent to the duplication model ., We show that the two models are vastly different in terms of other topological features we consider , and the preferential attachment model cannot capture some key features of the available PPI networks .

saccharomyces, computational biology

journal.pcbi.1006562

Ten simple rules for writing statistical book reviews

Extensive resources now support the statistical programmer and analyst ., The learner , reader , and general problem solver is thus faced with a choice of how to learn what is needed 1 , 2 ., This brief synthesis is not intended to be a comment or criticism on the pedagogy associated with successfully acquiring statistical and coding expertise , but there is evidence suggesting that up to 80% of coders do not read books to learn how to code 6 ., This seems like an unfortunate statistic , but the philosophy of “learning statistics by doing statistics” is not without merit and can be a viable approach to both introductory and expert learners alike 4 ., Nonetheless , R , Python , SAS , and MATLAB/C++ are quite literally deep languages that need to be mastered ., Fluency in a written or spoken language conveys reason and semantics 5; statistical reasoning 4 with a corresponding representation of the associated mathematics 3 can likely be secured by both doing and reading 7 ., Different problems and topics can also require the statistical programmer to embrace a diversity of resources to illuminate a solution , and the depth required must be defined by the prior knowledge of an individual and nature of the challenge ., Many statistical texts can be a significant time commitment , and open electronic resources are abundant ., The decision to read a statistical programming book is not necessarily trivial ., Short syntheses , i . e . , a review , of the relative merits of a specific resource can provide a critical decision tool to the potential reader ., The “ten simple simple rules” paper format was pioneered by Philip Bourne in PLOS Computational Biology 14 , and it has proliferated to nearly 100 papers , all functioning as a succinct , unique form of synthesis in itself 8 ., Sometimes extensive resources are summarized that support how to describe a focused process or get a task done in many domains of the scientific endeavor 11 ., Of these “ten simple rules” papers , there have been three that address the review process , including how to be an effective referee 9 , how to write a literature review 12 , and how to write a reply paper 10 ., Many of these rules certainly support improvements in how to write a review of statistical books and should be consulted ., Yet , book reviews in the Journal of Statistical Software , e . g . , strongly suggest that the importance of this topic warrants specific treatment because these reviews can serve many functions from descriptive summary to critical analysis to a launchpad for the importance of a statistical test , approach , program , language , and/or package ., All are important functions that advance statistics , but at least some of the rules here can enhance their capacity to assess merit and need for the end practitioner ., ( Appropriately ) defend books ., Write reviews ., Use reviews ., Book reviews that effectively support the decision process for better statistical reasoning are needed ., These rules promote this paradigm shift ., The book title is an excellent starting point for the reader to assess whether this is the resource for her but not the only mechanism ., The book cover or sleeve synopsis and publisher description can also fail to capture the whole story , and some statistical treatises , both introductory and advanced , necessarily invoke related principles and topics 13 ., As the objective expert of that specific text , an introduction to the necessity , scope , depth , and breadth of the topic in general can inform the reader on the challenges and solutions , including types of data or domains of inquiry that this field examines ., Place the work within the span of the literature with a brief explanation of the area in which it is embedded ., The goal of the first rule is therefore to ensure that the reader is in right place—conceptually , at least ., Most technical book reviews state the level of expertise required by the reader ., This is a critical form of synthesis that should be mentioned , even in brief , in a book review for statistics ., The most typical categories range from introductory to advanced , with relatively higher-level offering described by “graduate student” and beyond as the reader ., If the text is a blend of theory and practice with significant programming , the review should further explain the relative expertise needed for each and whether both dimensions are aligned in the assumed relative audience ., Book reviews can also take the opportunity here to frame this assessment by the expertise of the referee ( i . e . , it is sometimes useful to know if the book reviewer is a statistician , a programmer , or a domain-specific end-user ) or by the intended use of the text , such as primer , guide , in-depth treatise , or textbook appropriate for instruction at a given level ., If more than one edition exists , it is useful to describe the revisions to the more recent version of a book ., Professional and teaching textbooks can be relatively expensive , and a critical assessment of value can provide instructors with the merits associated with potentially higher costs to students of purchasing a newer text ., At minimum , a list of additions will facilitate a more informed choice for the reader and instructor , and mention of case studies , updates to code and data sets , and addition of supplements are all important criteria for the choice to learn or seek solutions from a specific edition ., Organization of the content matters to all learning 15 , and content provides context 16 ., The structure of statistical and programming texts can vary significantly ., The length and complexity of chapters , use of headings and subsections within chapters , and end-of-chapter summaries are not always needed but often do no harm ., Case studies , appendices , data sets , and location of supplements and supporting materials should be described ., Contemporary texts in statistics are often a hybrid of print and electronic resource materials , and description of the extent that a text functions in this capacity can influence the decision by the reader based on her preferred modality of learning and the rationale for exploring this topic ., This is also a good place to mention the different formats of the book ( if available in print and online ) ., As the reviewer , use parsimony and caution in deciding what level of detail to describe for the structural elements of a book—focus only on those elements that promoted the most effective learning ., These are the results , so to speak , similar to a conventional scientific publication or study report ., The description should be brief , topological , and highlight the most substantive elements of the book ., This component of the book review need not be unduly critical but should provide an overview of the what the text offers ., Some reviews take this description of what is done to also highlight what is done best and list the most valuable chapters to the reader ., This can be a useful guide to the potential reader and a means to assess expectations from the book as a whole ., If there are data sets or case studies that are revisited throughout the book or across multiple chapters , the extent that the chapters connect to one another can also be summarized ., Mention whether the content of the book is serialized or if chapters can be read piecemeal ., Readability is an intuitive concept ., It is the ease that one can comprehend writing 17 , 18 ., Complexity in syntax , vocabulary , and sentence structure should be described in a review of a statistical book ., A technical book need not be a technical challenge to read ., More broadly , appeal , style , and interest are important to all but the most committed readers , and it is reasonable to assume that a book on statistics provide some sense of enthusiasm for the topic , compel the reader to think deeply , and engage one with the challenges explored ., Composition is critical , particularly in long-form writing endeavors ., Within the R statistics community , there are now over 11 , 000 packages that extend the base language archived on https://cran . r-project . org ., SAS Procs and libraries in Python and MATLAB are also extensive ., Some statistical texts are associated with not only a single statistical program or language but with a single package or library ., A review of a statistical book should thus describe the specificity of the book , explain the extent that the book relies on single solution sets , or conversely contrasts alternatives in different languages , applications , packages , and/or libraries , and frame the programming ( if provided ) to general statistical theory and reasoning ., At times , this can be self-evident if the title of the book includes mention of the programming language or software , but the breadth of the statistics and case studies illustrated is typically not evident without review of the book ., If the book is not tied to a specific computation tool in any form , then the reviewer should mention that this is the case and state that the concepts described can be applied and transferred broadly from the book ., Compare and contrast ., There is a wealth of both short- and long-form documentation available for many open coding languages used in statistics and data wrangling ., There is also an extensive opportunity to seek specific solutions through numerous forums such as Stack Overflow ( https://stackoverflow . com/questions/tagged/statistics ) , Cross Validated ( https://stats . stackexchange . com ) , and Stack Exchange Mathematics ( https://math . stackexchange . com ) ., Online tutorials , blogs , carpentries , massive online open courses ( MOOCs ) , and webinars often provide useful , and at times , deep-learning opportunities ., A book review will certainly not comprehensively list all these options and compare and contrast to the principal subject text discussed , but if there is a significant alternative to consider , it should be mentioned ., Finally , there are also other books ., The reviewer should explicitly state the extent that she is contrasting to other resources , and due diligence by the reviewer suggests at the minimum a mention of the relative novelty and niche of the text in question ., Reading a book is a relationship ., The content , style , and perspective of the author ( s ) becomes a shared , internalized form of knowledge in a good book ., As the reviewer , it is legitimate and useful to others to mention the extent that one enjoyed the text , connected with the writing and concepts , or struggled with certain elements ( i . e . , comment on the quality of the relationship with the book ) ., A review should also mention the time that the reader should set aside to read and/or fully digest the content ., If the “summarize content” rule proposed above did not mention the standout , best chapters , this is an excellent spot to describe the chapters that provided the most for your buck and should not be skipped ., This is also an ideal opportunity to consider describing whether this is was a read-the-book-straight-through or piecemeal technical read for critical needs ., In general , it is best to be decisive in writing reviews 9 ., Evaluate the capacity that the book delivers on its stated goal ., Accept that you are part of the review process and likely have your own , specific purpose in reading this text ., Admit this in the review by articulating the need , success of text , and decision ( or not ) to use the described tools , framework , or theory ., Being specific and listing criteria point-by-point is useful to editors , authors , and readers 9 ., Similar to the peer review process for papers , be balanced , fair , and professionally critical by mentioning both strengths and weaknesses from your perspective ., Do your best to reveal implicit biases in your review ., Reading , writing , and statistics ., By putting oneself on the hook for a book to take notes and annotate or further synthesize these efforts and provide a review profoundly changes how one approaches a statistical and programming text 19 , 20 ., Higher education in the sciences and statistics has largely done away with book reviews and/or reports , but application and dissemination of critical thinking in statistics in the form of reviews is a learning opportunity ., Capitalize on this process , particularly when using a text to solve a problem and write a review ., Reviewing is a both a collegial and educational service that includes oneself as the beneficiary ., The rules proposed herein for writing a book review for statistics and increasingly for the associated coding or implementation of statistics and data do not mean to imply that reading texts in this domain is a burden ., On the contrary , the gratification of immersion in the structured reasoning inherent in these fields is a powerful form of literacy that merits discussion by people , for people ., Recommendation algorithms certainly influence many aspects of human behavior , and a book review is a reminder to take a moment and savor the story .

Introduction, Rules, Summary

Statistical books can provide deep insights into statistics and software ., There are , however , many resources available to the practitioner ., Book reviews have the capacity to function as a critical mechanism for the learner to assess the merits of engaging in part , in full , or at all with a book ., The “ten simple rules” format , pioneered in computational biology , was applied here to writing effective book reviews for statistics because of the wide breadth of offerings in this domain , including topical introductions , computational solutions , and theory ., Learning by doing is a popular paradigm in statistics and computation , but there is still a niche for books in the pedagogy of self-taught and instruction-based learning ., Primarily , these rules ensure that book reviews function as a form of short syntheses to inform and guide readers in deciding to use a specific book relative to other options for resolving statistical challenges .

Book reviews are a useful tool to inform learners in statistics and computational biology ., As an ecologist , I teach biostatistics and use many resources in the analysis and coding of research data ., In-depth texts can provide a critical resource , but well-written reviews can faciliate the decision to use a specific book .

learning, education, engineering and technology, sociology, pedagogy, social sciences, neuroscience, learning and memory, cognitive psychology, editorial, computer and information sciences, human learning, software tools, psychology, programming languages, reasoning, computer software, software engineering, biology and life sciences, computational biology, cognitive science

journal.pgen.1000527

A Network of Conserved Damage Survival Pathways Revealed by a Genomic RNAi Screen

Cellular damage is a normal component of life , with constant damage exposure from both endogenous and exogenous sources ., Damage to DNA is considered to be the most biologically relevant lesion with the potential of mutagenic results , though most exogenous agents have the potential to damage many components of the cell ., Responding appropriately to such insults , either mitigating cellular toxicity or initiating an appropriate cell death response , is critical , particularly in multi-cellular organisms ., Inappropriate responses may facilitate deleterious effects , such as a destabilized genome and diseases such as cancer 1 ., As such , DNA damage response ( DDR ) components are critical suppressors of deleterious effects of genotoxic agents by controlling cell cycle progression , DNA repair , and apoptosis 2 ., For this reason , there have been many investigations using a variety of model organisms to identify components of DDR and subsequently to determine how they function and the consequences of their dysfunction ., Recent reports suggest that DDR may involve pleiotropic cellular processes other than the central DDR components 3 , 4 , yet an intuitive systems level view of the cellular components required for damage survival , their interrelationship , and their contextual importance has been lacking ., The most comprehensive attempt at understanding the interrelationship of damage response and survival components in yeast at a systems level has been mapping identified genes to a network by integrating the general biological processes as distinct modules 5 ., It has been suggested that the inclusion of well-defined biological pathways in protein networks might provide a better understanding of biological interactions therein 6 , 7 ., In order to identify genes involved in damage response in an unbiased manner and to put them in a functional context , we used an RNAi library 8 to knock-down every predicted protein in the Drosophila melanogaster genome and assessed whether knock-down of individual proteins significantly altered cell viability following methyl methanesulfonate ( MMS ) exposure ., MMS is a prototypical alkylating agent that attacks nucleophilic groups , such as those found in nucleic acid 9 ., The resultant base methylation destabilizes the glycosidic bond , facilitating the production of the most biologically relevant cellular lesion of MMS , an abasic site in DNA ., Other common sources of alkylation damage include endogenous S-adenoysyl methionine 10 , the tobacco carcinogen N-nitorsoamine 11 and chemotherapeutics such as temozolomide 12 , carmustine ( BCNU ) 9 , and cyclophosphamide 13 ., Considering the physiological relevance of alkylation damage , several recent studies using yeast as model organism investigated the global effects of alkylation damage induced by MMS ., Viability 14 , gene transcription 15 and protein expression 16 were measured in these studies , all of which provided insights into the diverse nature of biological responses to alkylation damage ., However when transcriptional responses to various environmental stresses were compared between mammalian and yeast cells 17 , Murray et al . reported clear distinctions and suggested that this might be the result of different selective pressures between multi-cellular organisms and single-cell organisms ., In the attempt to understand diverse responses required for damage survival in a system that is evolutionarily closer to mammals than yeast , we performed a genome-wide , RNAi-based screen using cells derived from Drosophila melanogaster ., Our experiments were based on “loss of function” and an assessment for cellular viability following exposure to MMS ., Here we present results from this MMS survival screen , the pathways that were identified , and a comparative analysis of pathways to understand conservation of these pathways in yeast and mouse cells ., Additionally , we present a novel approach of assembling a protein interaction network based on defined biological pathways , which facilitates network consolidation ., By including biological pathways in our genomic data and protein network analysis , we were able to determine the commonality across species , which was obscured by direct orthologue comparison , and provide a simplified representation of the global survival responses that we identified ., An RNAi screen was designed to identify those genes that modulate cellular survival following exposure to a level of MMS-induced damage that resulted in only a limited amount of cell death ., A linear decrease in cell viability was observed from 0 . 002% ( w/v ) of MMS to 0 . 008% ( w/v ) of MMS ( Figure S1 ) ., A dose of 0 . 004% ( w/v; 40 µg/mL ) MMS was chosen , which resulted in a statistically significant decrease to approximately 65% viability , while allowing an additional , measurable decrease in cell survival ., RNAi screens were performed using the Drosophila RNAi Screening Center ( DRSC ) version 1 library ( about 23 , 000 D . melanogaster open reading frames ) 8 ., Kc167 Drosophila cells were exposed to three days of dsRNA to allow protein knock-down , followed by three days of growth in either the absence or presence of MMS exposure ( Figure 1A ) ., To identify those genes required for cell viability following MMS treatment , viability results from the MMS treated RNAi screen were compared with that of the untreated ( control ) screen as previously described 18 ., 1 , 398 different open reading frames were identified that affected MMS survival in a continuous distribution of cell survival values ( see 18 and http://gccri . uthscsa . edu/ABPublished_Data . asp for original data ) , of which 996 had a unique assigned FlyBase gene number ( FBgn; denotes known genes ) ( see Figure 1B and Table 1 ) ., Of these 996 genes , the top 537 were selected for validation analysis by a previously described , stringent validation method 18 ., Whereas in 18 we validated normalization methods , here we independently validated individual genes using dsRNA targeting a different region of each gene ., 202 protein knock-downs validated with a significant MMS viability effect , while 55 more had a notable trend effect without meeting our stringent statistical criteria ( Table S1 and Table S2 ) ., An examination of gene ontology ( GO ) on the 202 validated MMS survival genes revealed a significant enrichment for genes involved in DNA metabolism , gene transcription , and cellular metabolism ( Table 2 ) ., The overall variety of GO categories observed was quite broad , consistent with the findings reported for an analogous yeast screen 3 ( data not shown ) ., Surprisingly though , no significant enrichment between the gene orthologues for the two organisms was observed ( G-test p-values of 0 . 29 and 0 . 057 or Fisher Exact test p-values of 0 . 37 and 0 . 08 , yeast and fly , respectively ) ., To further test this , we examined the MMS sensitivity following knock-down of 183 fly genes that were orthologues of to 118 yeast MMS survival genes 3 , but only found conservation of MMS survival phenotype with 20 ( Table S3 ) ., This apparent lack of gene enrichment between species has also been reported when comparing transcriptional profiles between mammalian and yeast cells in response to a variety of stresses 17 ., Overall , these results suggest that there may be conservation of the biological processes that respond to MMS rather than the individual genes ., Assuming that biological function is a more informative measure of damage response than a requirement of individual genes , we therefore endeavoured to identify those MMS survival proteins within known signaling , metabolic , and enzymatic pathways ., Using both a priori knowledge and KegArray , we identified 13 pathways that together included 41 MMS survival proteins ( Figures 1 , 2 , 3 and Figures S2 , S3 , S4 , S5 ) , among them , five DNA repair pathways ( Base Excision Repair; Nucleotide Excision Repair; Mismatch Repair; Homologous Recombination Repair; and RECQ ) ., Many of these pathways have a statistically significant number of MMS survival proteins ( Table S5 ) , and this is without accounting for pathway “branching” or “subdivision . ”, Considering the likely role of these pathways in MMS survival , we used an RNAi assay method , which is both more sensitive and stringent than the original screen 18 , to determine whether other members of the 13 identified pathway also affected MMS survival ( Table S1 ) ., By individually interrogating the 346 pathway members of the 13 pathways that were not identified in the original screen , an additional 105 MMS survival genes were discovered ., This observation significantly enriched the number of MMS survival genes in each of the 13 pathways ( Figures 1 , 2 , 3 and Figures S2 , S3 , S4 , S5 ) and provides additional support to the hypothesis that each of the identified pathways are involved in MMS survival ., Though this result highlights the possibility of false negative screen data , compared to our previous false negative rate of 23 . 6% from randomly selected data , we observe a significant enrichment for false negatives within pertinent pathways ( ( χ2 p\u200a=\u200a3 . 9E-5 ) ., In total we identified 146 MMS survival genes in 13 “MMS survival pathways , ” encompassing 25% to 86% of all non-essential genes within each pathway ( Table S4 ) ., Using the genes we mapped to the MMS survival pathways , we then compared pathways identified from our Drosophila MMS screen with the analogous screen performed in yeast 14 using genes they found to be responsive to MMS ., As previously noted , we did not observe a significant overlap between the two screens when comparing MMS survival gene orthologues , however , with this pathway comparison , we mapped orthologues of yeast MMS survival genes to 10 of the 13 Drosophila MMS survival pathways ( Figures 2 and 3; Figures S2 , S3 , S4 , S5; Table S6 ) ., The three Drosophila MMS survival pathways without yeast MMS survival genes either are not conserved in yeast or have almost all of the pathway components are essential for viability in yeast , thus making them refractory to MMS viability analysis ( Table S5 ) ., The gene enrichment with each pathway between species is highly supportive of a conservation of processes involved in MMS survival ., A notable absentee in the Drosophila screen , which was observed in yeast , is 3-methyladenine-DNA-glycosylase ( MAG1/AlkA ) , the principal protection against MMS-induced DNA damage in yeast , but no direct orthologues exist in animals ., One of the two glycosylase found in Drosophila that act at the same step in BER , namely Thd1 , was found to be required for survival after MMS treatment ., Furthermore , several of the pathways observed in both species , such as proteasome 19 , the TOR pathway 20 , and DNA repair pathways 21 , have been shown to be functionally responsive to MMS in yeast ., Altogether , these results suggest a conservation of pathway function , if not individual genes , in response to MMS between species ., To demonstrate that the identified Drosophila pathways are functioning in the expected manner in response to MMS , five were selected for further examination – base excision repair ( BER ) , DDR , glutathione metabolism , proteolysis by proteasomal degradation ( proteasome ) , and the TOR pathway ., Two of these pathways , BER and DDR , are expected to play a role in MMS survival 9 , while the others were selected for their apparent breadth of function and their non-canonical role in damage survival ., For each pathway , an appropriate functional assay was chosen , and when possible , an appropriate upstream MMS survival protein within that pathway was knocked-down ( Figure S6A ) to demonstrate modulation of the MMS induced response ., First , we tested the two expected MMS survival pathways BER and DDR ., MMS-induced DNA damage results in the production of apurinic/apyrimidic ( AP ) sites , a DNA damage typically repaired by BER 9 ., We therefore quantified the amount of AP sites per microgram of genomic DNA following MMS treatment and observed a statistically significant increase compared to control ( Figure 4A; p≤7 . 6E-8 ) ., Knock-down of the BER component XRCC1 resulted in an increased amount of AP sites in and of itself ( p≤3E-3 ) , but following MMS treatment , the amount of AP sites in the absence of XRCC1 was increased further ( Figure 4A; compared with untreated XRCC1 knock-down , p≤3 . 3E-8; compared to MMS treated luciferase ( Luc ) control , p≤3 . 7E-3 ) ; MMS therefore produces the expected form of DNA damage to which BER appropriately responds ., p53 , a central component of the DDR pathway , is regulated at expression , protein stabilization , and posttranslational modification levels ., As expected , p53 gene expression increased in response to MMS exposure ( Figure 4B; p≤1 . 6E-3 ) ., We then examined three MMS survival pathways that are not part of the canonical DDR and DNA repair process: glutathione metabolism , proteasome , and the TOR pathway ., An increased activity of the glutathione synthesis pathway following MMS exposure was demonstrated by measuring total glutathione content per milligram of protein lysate ( Figure 4C; p≤3 . 4E-3 ) , as well as by examining the expression of the rate-limiting enzyme for glutathione synthesis , GCLc 22 ( p≤8 . 8E-4 ) ., Therefore , as expected , knock-down of this same protein , GCLc , significantly reduced the total amount of glutathione present compared with control ( p≤3 . 4E-3 ) , but its knock-down also prevented the cells from increasing the level of glutathione in response to MMS exposure ( Figure 4C ) ., Since glutathione synthesis is considered to be an oxidative stress response , we demonstrated that MMS resulted in a dose-dependent increase in the level of 8-oxoguanine , a DNA lesion normally associated with oxidative damage ( Figure S7 and Text S1 ) ., Thus it appears that MMS results in not only alkylation damage but also damage from oxidative stress ., For the analysis of the proteasome degradation pathway , we measured proteasome activity and found that it increased following MMS treatment ( Figure 4D; p≤4 . 2E-2 ) ., Protein knock-down of the proteasome components Rpn2 or Pros26 . 4 significantly reduced proteasome activity compared to cells without knock-down that were either unexposed ( Rpn2: p≤1 . 7E-2; Pros26 . 4: p≤3 . 7E-3 ) or exposed to MMS ( Rpn2: p≤1 . 4E-4; Pros26 . 4: p≤7 . 0E-5 ) ., Following Rpn2 knock-down , cells were unable to mount a detectable increase in proteasome activity following MMS exposure , although we were unable to demonstrate this for Pros26 . 4 knock-down ( p≤1 . 7E-3 ) ., Overall , it appears that MMS exposure results in increased proteasome activity ., Finally , to investigate TOR pathway activity , S6K phosphorylation status was monitored ., TOR is a kinase that phosphorylates S6K to promote growth through ribosome biogenesis and is negatively regulated by the tumor suppressor protein Tsc1 23 ., MMS exposure resulted in a dose-dependent decrease in S6K phosphorylation , suggesting an inhibition of TOR activity ( Figure 4E ) ., This MMS-induced decrease in S6K phosphorylation was dependent on Tsc1 ( Figure 4E ) ., MMS exposure therefore elicited a down-regulation of the growth promoting TOR pathway , suggesting that this pathway is also a coordinated component of DDR similar to observations in yeast 20 ., This supports a previously published observation by Matsuoka et al . 4 , who showed that some components of the mammalian TOR pathway are phosphorylated following ionizing radiation exposure ., In conclusion , the results of these functional assays validate the identification of the “MMS survival pathways” by RNAi screening and that the functionality of these pathways , or lack thereof , affects MMS survival ., Having thus identified 13 Drosophila MMS survival pathways , we went on to investigate their functional conservation in mammals ., Using primary mouse embryonic fibroblasts ( MEFs ) , we examined the same five pathways that were modulated following MMS exposure in Drosophila ., In general we observed comparable results in MEFs to Drosophila ( compare Figures 4 and 5 ) ., MMS increased the amount of AP sites in MEFs ( Figure 5A ) ; in the absence of XRCC1 , however , we observed a significant increase in AP sites following MMS exposure ( Figure 5A; p≤3 . 7E-4 ) ., For DDR in MEFs , we assessed the phosphorylation status of Chk1 ( a kinase that , once phosphorylated , phosphorylates p53 ) 24 , total p53 protein , and the phosphorylation status of p53 ( Figure 5B ) , not to examine p53 expression levels , but to obtain a result equivalent to that of Drosophila result: that DDR is activated by MMS exposure ., Also , similar to the observation in Drosophila cells , MEFs had increased proteasome activity following MMS treatment ( Figure 5D; p≤2 . 0E-4 ) ; knock-down of either proteasome component , PSMC1 or PSMD1 ( orthologues to Rpn2 and Pros26 . 4 , respectively ) , resulted in decreased proteasome activity following MMS exposure ( PSMC1: p≤8 . 1E-5; PSMD1: p≤4 . 1E-4 ) ., These results are comparable with Drosophila: it is clear that proteasome activity is responsive to MMS in MEFs ., It should also be noted that , though we were able to demonstrate the same dose-dependent decrease in S6K phosphorylation following MMS exposure in MEFs as seen in Drosophila cells ( Figure 5E ) and that knock-down of TSC1 disrupted this MMS-induced effect , the disruption was not as evident as observed with Drosophila , probably due to inefficient knock-down of the TSC1 protein in MEFs ( data not shown ) ., Having observed the functionality of these pathways in response to MMS and the ability of the five siRNA to modulate the MMS response of their respective pathway , we also examined the effect of each knock-down on MMS survival in MEFs ., Three of the protein knock-downs , GCLC , PSMC1 and PSMD1 , affected MMS survival in MEFs ( data not shown ) ., Overall , it would appear that Drosophila could be used to accurately predict MMS survival pathways in mouse ., Given that MMS causes alkylation damage , it would be reasonable to suppose the identified biological pathways are generally involved in the response to other alkylating agents ., Temozolomide is an alkylating agent used clinically in the treatment of cancer 12 , 25 ., It has already been demonstrated that BER and TOR are necessary for Temozolomide survival 12 , 25 ., Temozolomide causes DNA damage by increasing the number of AP sites , and similar to our MMS results , inhibition of BER increases sensitivity to temozolomide 12 ., It has also been reported that rapamycin inhibition of the TOR pathway increased cellular sensitivity to temozolomide 25 ., To further demonstrate the utilization of MMS survival pathways in response to temozolomide , we examined the DDR pathway , glutathione levels , and proteasome activity in MEFs and observed the same responses as observed following MMS exposure ( Figure S8 ) ., Further , using HEK293 cells , we observed that both MMS and temozolomide exposure repressed Notch reporter activity ( Figure S9 and Text S1 ) ., Absence of functional Notch protein has been shown to repress the activity of a downstream transcriptional activator ( RBP-Jκ ) 26 , therefore this result demonstrates a functional Notch pathway response in alkylation damage exposure ., Taken together , these data suggest a functional conservation of the MMS survival pathways responses with other similarly acting agents ., Of the 307 identified MMS survival genes ( 202 validated screen hits plus 105 from pathway analyses ) , we were able to assign 146 to 13 pathways ( Figures 1 , 2 , 3 and Figures S2 , S3 , S4 , S5 ) ., Because 161 MMS survival protein remain unassigned , we re-examined the inter-relationship between the identified MMS survival proteins using the currently available Drosophila protein:protein interactome ( PPI ) map 27–29 ., To measure the connectivity among the MMS survival genes , we took the largest connected components of the PPI network and discarded the other , smaller components ., The largest connected component contained 7364 of the 7504 proteins in the original PPI network ., Taking the 202 original validated MMS screen hits as an unbiased starting point , we determined the number of proteins that were directly connected to one another , compared to a random set of the same number of proteins from the PPI , and observed a significant enrichment ( Figure 6A and Table S7; p≤2 . 1E-10 ) ., Similar results were obtained for other assessments of network connectivity ( Table S7 ) ., Because the MMS screen hits had more connections on average than a set of random genes , which might have biased the above analysis , we also compared the connectivity of the MMS screen hits in the real PPI network with the connectivity of the same set of proteins in a randomly rewired PPI network ., Degrees were preserved in the random rewiring process ., As shown in Table S7 , the MMS screen hits have statistically significantly more direct interactions than would be expected in a randomly rewired network ( p\u200a=\u200a0 . 01 ) ., On the other hand , the randomized network had smaller average distance and higher global efficiency than the real network , which could be attributed to the well-known property of real-world networks: they usually have slightly longer average distances ( and correspondingly , lower global coefficients ) than their degree-preserving randomly rewired counterparts 30 ., Because we were unable to assess the effect of essential proteins ( proteins whose knock-down resulted in cell death regardless of treatment ) on MMS survival , we repeated the connectivity analysis while including those essential proteins that were connected to two or more MMS survival proteins , since these are the ones most likely to have a functional role in MMS survival ., This increased the size and significantly improved the connectivity within the resultant network ( Figure 6B and Table S7; p≤1 . 3E-26 ) ., After pathway analysis , we identified an additional 105 MMS pathway hits , which were validated ., In order to expand the network to include the proteins in these relevant pathways , these hits were then incorporated and a larger , and equally well-connected network was observed ( Figure 6C and Table S7; p≤4 . 6E-26 ) ., Considering the apparent importance of pathways over the individual genes , we then included all components of the 13 identified MMS survival pathways and observed that the connectivity of the resultant enlarged network was significantly improved yet again ( Figure 6D and Table S7; p≤4 . 6E-52 ) ., Similar to the analysis of the 202 validated MMS screen hits , we also compared the connectivity of the subnetworks containing the essential genes or pathway hits with that of a randomly rewired network consisting of the same nodes ., Although at a lower statistical significance , the same general trend was observed ( Table S7 ) ., This more inclusive network allows a view of all interactions both within and connecting to a pathway , even if the pathway components themselves are not critical to survival after MMS ., To qualitatively visualize this result , each protein known to be in a pathway was grouped together and assigned to a “pathway node” ( a single node within the interactome that retains the interactions of its constitutive proteins to proteins that are external to that pathway ) ., This resulted in a highly connected interactome , or “MMS survival network” ( Figure 6E ) , that now encompassed 179 of the 233 MMS survival proteins that are present in the PPI network ( 77% ) ., Of the remaining 54 orphan proteins , 47 are only one protein removed from the MMS survival network , six are two proteins away , and only one is not connected at all ., At each step in this analysis , we observed an improved network , either enlarged or better connected , not only when comparing the entire set of proteins in that network to a random set of proteins in the PPI , but also when randomizing the additional proteins added at each step ( data not shown ) ., Overall , our observations with the MMS survival network suggest that despite the general interconnectivity within protein interactomes , a pathway analysis is highly relevant because it may improve interactome connectivity and it simplifies a systems biology overview ., Studies performed using yeast as a model organism to predict network response to MMS reported an astounding involvement of diverse biological pathways 31 ., Considering that the genes that respond to environmental stress differ between mammalian and yeast cells 17 , we presumed that damage response might be different or more complex in higher eukaryotes , especially considering the presence of paralogues and thus increased genetic redundancy ., We therefore performed a genome-wide , RNAi based screen with Drosophila cells to investigate which genes are essential for survival following damage exposure with MMS ., We were able to identify and validate 307 MMS survival genes , the majority of which had not previously been associated with alkylation damage survival ., Of these genes , 146 were components of 13 different MMS survival pathways ., With the five pathways examined in detail , we observed that four were functionally conserved in yeast and all five conserved in mouse with regard to their utilization following MMS treatment ( Figure 5 ) ., In yeast , experimental validation of response to MMS by proteasome 19 , the TOR pathway 20 , and DNA repair pathways 21 was previously reported ., Similarly , glutathione response to MMS was observed in mammalian cells 32 , and our observation of an increase in GCLc expression provides an underlying mechanism for this phenomenon ., Our demonstration of a dose-dependent increase in 8-oxoguanine after MMS exposure indicates that MMS also results in oxidative stress damage , as previous studies suggested 33 ., Additionally , several recent studies have demonstrated a role for the proteasome in regulating several DNA repair pathways ( reviewed in 34 ) , supporting our observation of increased proteasome activity in response to MMS ., Thus , our screen and pathway identification have revealed a conserved set of MMS survival pathways ., Our Drosophila based study has provided novel insights to the global cellular response to alkylation damage by identifying biological pathways whose functions are required for survival after this damage ., The only other analogous genome-wide , loss-of-function screen for MMS survival genes was performed in yeast 14 ., That study highlighted the general biological processes required for MMS survival based on gene ontology and integrated the identified proteins into a disorganized network 14 ., Considering that pathways , whether signaling , metabolic , or enzymatic , have long been identifiable entities , it is logical to consider them as units within a network ., Thus our experiments focused on identifying pathways required upon exposure to damage and validating the biological responsiveness of pathways following this damage exposure ., Our results confirmed that these biological pathways are indeed functional in yeast , Drosophila , and mouse cells and therefore functional contribution of these biological pathways are pertinent to damage response in a network representation ., In addition to the functional conservation of the survival pathway in response to alkylation damage , these same biological pathways appear to have roles in response to other types of damages ., It is interesting to compare our results with an elegant study by Matsuoka et al . 4 , which identified proteins that are phosphorylated following ionizing radiation in human cells ., Their study identified proteins that are components of nine of our 13 MMS survival pathways , including four of the DNA repair pathways , DDR , mTOR , proteasome , basal transcription , and ribosome ., These results suggest that different types of damage , not just alkylation damage , may utilize different components of a DNA damage survival network in a functionally conserved manner and reemphasize the functional conservation of pathways , if not the individual genes , between species ., Our emphasis to reorganize the MMS survival network based on pathways is to facilitate the observation of biologically relevant interactions ., Often protein:protein interaction networks may appear chaotic , but may be interrogated for simple sub-networks associated with protein ( s ) and pathways of interest 21 ., However , when working with pleiotropic responses that encompass so many different biological processes , such as DDR , a chaotic network representation appears non-intuitive ( Figure 6D ) ., Thus , the integration of pathways within the conceptual framework of systems biology networking is logical ., An additional advantage of including pathways is highlighted by our demonstration of MMS survival protein enrichment by detailed examination of pathways ., Even with this detailed analysis and convincing evidence that the pathways were indeed functioning as expected , we were not able to assign every protein within each pathway as an “MMS survival protein . ”, There are many possible explanations for this , but nonetheless , considering the pathways as a whole provide a framework within the network that highlights novel interactions , cross-talk , and identified proteins not mapped to a canonical pathway , but present within the network , would unlikely be observed in a “chaotic network view” and encourages their investigation ., This approach is similar to computational clustering of networks based on signaling pathways using interactome datasets 35 , but our approach includes both identified proteins and pathway components ., Our representation is simplified , using a single node to represent the entire pathway rather than a complex display of interactions for every component in a pathway 36 ., This simplified pathway inclusive representation reveals a highly organized network , consistent with the requirement of each pathway for cellular survival ( Figure 6E ) , and provides an effective strategy to integrate modular components into the network 37 and thereby inferring biological properties 38 ., The interconnectivity between the survival pathways ( Figure 5E ) suggests potential pathway cross-talk ., If such cross-talk exists , it would be highly pertinent to cancer therapy ., Recent studies have demonstrated the utility of a global level analysis , allowing identification of altered pathway function in complex diseases such as the Notch pathway in pancreatic cancer 39 and similarly the DDR pathway in breast and colorectal cancers 40 ., Considering our identification of Notch , TOR , DDR , and the proteasome as “survival pathways , ” all of which are currently being explored as targets for cancer therapy 41 , 42 , our identified survival network would suggest the possibility of combining pathway-specific pharmacological agents in cancer therapy ., Some of the pathway connections and potential cross-talk represented in our survival network ( Figure 6E ) have already been observed ., For example , protein phosphatase , PP2A , a downstream component of the TOR pathway 23 , interacts with the DDR component to regulate phosphorylation of ATM and ATR 43 and vice versa 4; DDR interacts with BER via CHK2 and XRCC1 44; Notch interacts with DDR via Mastermind and p53 45; the glutathione pathway interacts with the nucleotide excision repair pathway ( NER ) 46; and the proteasome interacts with various DNA repair components 34 , 47 ., Together , it would appear that our model of an integrated network of conserved damage survival pathways is both valid and biologically relevant ., In conclusion , we have identified a network of pathways that have a functional role in damage response by affecting viability; we also demonstrated the functional conservation between species of the MMS survival pathways ., By considering the protein interactions between the MMS survival proteins and by incorporating the MMS

Introduction, Results, Discussion, Methods

Damage initiates a pleiotropic cellular response aimed at cellular survival when appropriate ., To identify genes required for damage survival , we used a cell-based RNAi screen against the Drosophila genome and the alkylating agent methyl methanesulphonate ( MMS ) ., Similar studies performed in other model organisms report that damage response may involve pleiotropic cellular processes other than the central DNA repair components , yet an intuitive systems level view of the cellular components required for damage survival , their interrelationship , and contextual importance has been lacking ., Further , by comparing data from different model organisms , identification of conserved and presumably core survival components should be forthcoming ., We identified 307 genes , representing 13 signaling , metabolic , or enzymatic pathways , affecting cellular survival of MMS–induced damage ., As expected , the majority of these pathways are involved in DNA repair; however , several pathways with more diverse biological functions were also identified , including the TOR pathway , transcription , translation , proteasome , glutathione synthesis , ATP synthesis , and Notch signaling , and these were equally important in damage survival ., Comparison with genomic screen data from Saccharomyces cerevisiae revealed no overlap enrichment of individual genes between the species , but a conservation of the pathways ., To demonstrate the functional conservation of pathways , five were tested in Drosophila and mouse cells , with each pathway responding to alkylation damage in both species ., Using the protein interactome , a significant level of connectivity was observed between Drosophila MMS survival proteins , suggesting a higher order relationship ., This connectivity was dramatically improved by incorporating the components of the 13 identified pathways within the network ., Grouping proteins into “pathway nodes” qualitatively improved the interactome organization , revealing a highly organized “MMS survival network . ”, We conclude that identification of pathways can facilitate comparative biology analysis when direct gene/orthologue comparisons fail ., A biologically intuitive , highly interconnected MMS survival network was revealed after we incorporated pathway data in our interactome analysis .

Cellular damage is known to elicit a pleiotropic response , but the relative importance of the constituent components in cell survival is poorly understood ., To provide an unbiased identification of the proteins utilized in damage survival , we performed an RNAi survival screen in fly cells with methyl methanesulfonate ( MMS ) ., The genes identified are involved in 13 biologically diverse pathways ., Comparison with analogous yeast data demonstrated a lack of conservation of the individual MMS survival genes but a conservation of pathways ., We went on to demonstrate the MMS responsiveness for five representative pathways in both fly and mouse cells ., We conclude that identification of pathways can facilitate comparative biology analysis when direct gene/orthologue comparisons fail ., Incorporation of pathway data in interactome analysis also improved connectivity and , more importantly , revealed a biologically intuitive , highly inter-connected “MMS survival network . ”, This pathway conservation and inter-connectivity implies extensive interaction between pathways; for diseases such as cancer , such crosstalk may dictate disparate cellular responses not necessarily expected and confound treatments that are not tailored to the individual molecular context .

genetics and genomics/genomics, genetics and genomics/comparative genomics, genetics and genomics/functional genomics, genetics and genomics, genetics and genomics/bioinformatics

journal.pntd.0005002

Spatial Heterogeneity of Habitat Suitability for Rift Valley Fever Occurrence in Tanzania: An Ecological Niche Modelling Approach

Rift Valley fever ( RVF ) is a mosquito-borne zoonotic disease of major public health and economic concern occurring mainly in Africa 1–6 and the Arabian Peninsula 7 , 8 ., The potential for further geographical spread of RVF to other areas of the world has been suggested 9–11 ., The disease is caused by the RVF virus ( RVFV ) of the genus Phlebovirus and family Bunyaviridae 12 , 13 and affects both humans and livestock ., In this study , RVF outbreak was defined as occurrence in a specific location of laboratory-confirmed RVF cases affecting domestic ruminants ., A RVF outbreak wave ( epidemic ) referred to sequential reports of the outbreaks at various locations within Tanzania from date of onset of the first outbreak during a particular time period of the year until outbreaks were no longer reported in the country ., Tanzania has a long history of RVF outbreaks , and it is not known how RVFV was introduced to the country ., Between 1930 and 2007 , a total of 10 RVF outbreak waves have been reported in Tanzania with average inter-epidemic period ( IEP ) of 8 years 14–17 ., There also appears to be spatial heterogeneity in the distribution of RVF ., A total of 31/90 ( 34 . 4% ) districts from 10/14 ( 71 . 4% ) regions in the eastern Rift Valley ecosystem have reported RVF outbreaks in the past compared with 12/69 ( 17 . 4% ) districts from 5/11 ( 45 . 5% ) regions in the western ecosystem 18–20 ., The past RVF outbreaks in Tanzania resulted in devastating socio-economic losses including food insecurity and threatened livelihoods ., Notably , the last RVF outbreak in Tanzania in 2006/2007 caused high mortality rates in laboratory confirmed cases amongst domestic ruminants ( 37% , n = 136 , 570 ) and humans ( 46% , n = 309 ) 14 ., Animals lost monetary value by 34% ( e . g . price of a bull dropped from US$ 238 to 158 ) , monthly internal market flow dropped by 37% ( e . g . 4 , 251 to 2 , 679 cattle ) and annual external market flow dropped by 54% ( e . g . 2 , 594 to 1 , 183 cattle ) 14 ., Additionally , the loss due to death of domestic ruminants was > US$ 6million and the government spent about US$ 4 million in the control of the disease 14 ., It is not known why RVF outbreaks have been reported mainly in the eastern Rift Valley ecosystem ., However , it should be realized that active and well-structured RVF surveillance has never been conducted throughout the country partly because of financial resources and challenging logistics ., The northern Tanzania , particularly Ngorongoro district , in the eastern Rift Valley ecosystem has remained the epicentre of all past RVF outbreaks in the country 16 ., As a result , past RVF surveillance and awareness campaign efforts have been concentrated much more in the northern than other areas of the country ., We cannot therefore , discount the possibility that sampling or reporting bias may have contributed to over-reporting of RVF outbreaks in the eastern rather than the western Rift Valley ecosystem over time ., It is probable that some of the un-sampled locations and locations without reports of RVF outbreaks in the country are also suitable for disease occurrence ., Because the disease control resources are generally limited , it is interesting to understand if heterogeneity exists in the habitat suitability for RVF occurrence in the country , as this will inform allocation of disease prevention and control resources proportional to the risk ., A number of scientific methods are available that can be used to generate information on the potential suitable habitat for species and disease occurrence ., These include general-purpose statistical methods of temporal and spatial prediction such as generalized linear models ( GLM ) 18 , 19 , generalized additive models ( GAM ) 20 , 21 and Bayesian estimation methods 22 , 23 ., However , such models require both disease presence and absence data and inferences drawn from their outputs are therefore limited to the area covered by the data ., Furthermore , these methods frequently fit linear functions between predictor variables and disease data although ecological associations are frequently highly complex and non-linear 24 , 25 ., Ecological niche models ( ENMs ) that were originally developed for ecological and conservation purposes are being used increasingly to model the spatial distribution and potential risk of occurrence of a range of diseases and vector species ., For example , they have been applied to characterize the habitat suitability for leishmaniasis 26 , malaria 27–30 , RVF 31 , 32 , bluetongue 33 , anthrax 34 , dengue 30 , Chagas disease 35 , filovirus disease 36 , Marburg hemorrhagic fever 37 , avian influenza 38 , plague 39 , 40 and lymphatic filariasis 41 ., The main advantage of ENMs , over that of the more traditional regression modelling approaches , such as generalized linear mixed models , is that they require only presence data 39 ., These data are used , together with a randomly-generated sample of background data points from the study area ( representing the available environment ) and a suite of predictor variables , to define the fundamental niche of the species or disease 42 , 43 ., In addition , as the results of such models can be extrapolated beyond the geographical areas defined by the data points used to calibrate the model , these predictive risk mapping approaches are useful for identifying other areas suitable for occurrence of the disease 42 ., These presence-only methods illustrate the likelihood of an organism’s presence or the relative ecological suitability of a spatial unit within the study area 43 ., Maximum Entropy ( MaxEnt ) is one of the presence-only general-purpose niche modelling algorithms , which has been described as efficient to estimate the probability distribution of species and diseases 42–48 and is reported to perform well , even with very small sample sizes 49–50 ., In this study , we investigated the potential effect of bioclimatic variables related to temperature and precipitation , elevation , soil type , livestock density , rainfall pattern , proximity to wild animal protected areas and proximity to forest on the spatial habitat suitability for RVF occurrence in Tanzania ., We anticipate that generation of evidence-based information on the spatial dimensions of the potential suitable habitat of RVF occurrence and understanding how much the potential predictor variables contribute in delineating these suitable habitats , will inform targeted risk assessment , surveillance and cost-effective-usage of disease control and prevention resources ., The domestic ruminants ( cattle , sheep and goats ) RVF disease outbreak data used in this study were extracted from reports of the ministry responsible for livestock development in Tanzania ., These data were anonymous , and it was therefore not possible to associate disease data with specific animal or its owner ., Serological data from domestic ruminants ( cattle , sheep and goats ) used for ground-truthing of the ecological niche modelling outputs were from the study that received ethical approval from the Medical Research Coordinating Committee of the National Institute for Medical Research in Tanzania ( ethics certificate number NIMR/HQ/R . 8a/Vol . IX/1296 ) ., This study was conducted in Tanzania Mainland , located between longitudes 29 and 41° east and latitudes 1 and 12° south ., Tanzania Mainland borders Kenya , Uganda and Lake Victoria in the north , Rwanda , Burundi and the Democratic Republic of the Congo ( DRC ) in the west ., On the south it borders with Zambia , Malawi , Mozambique and Lake Nyasa , and to the east it borders the Indian Ocean ( Fig 1 ) ., Administratively , Tanzania Mainland has 25 regions with total land areas of 883 , 343 square kilometres ., The ecological characteristics of the country vary widely ., The north-eastern regions experience a bimodal rainfall pattern ( October—December and March—May ) whereas the central , western and southern regions of the country experience a unimodal rainfall pattern ( November—May ) ., Pastoralism is mainly concentrated in Arusha and Manyara regions and agro-pastoralism in Tabora , Geita , Shinyanga , Mwanza , Dodoma and Singida regions 51 ., The plateau of the northern Tanzania is comprised of relatively higher livestock densities ( cattle ≥ 50 , goats ≥ 45 and sheep ≥ 14 head per square kilometre ) than the rest of the country 51 ., Ecological modelling of habitat suitability for RVF occurrence was implemented using the MaxEnt software version 3 . 3 . 3k 42 ., There has been no systematic surveillance of RVF in Tanzania and therefore , the spatial range of its occurrence was not explicitly known ., Prior to conducting our study , we could not differentiate whether more RVF cases were confirmed in the northern Tanzania because those locations were suitable for disease occurrence or rather because they received the largest surveillance efforts ., Our presence dataset was therefore considered small and biased because of the fact that most of past surveillance efforts have been conducted in the northern Tanzania ., We assumed that the un-sampled locations of the country could be suitable for RVF occurrence ., For this reason , the MaxEnt default setting seemed more appropriate because it assumes that the species/disease being modelled is equally likely to be anywhere in the geographical space of the study area 70 ., In addition , the regularization multiplier was set to 1 to limit over-fitting of the model and prevent prediction from being inadequately large 48 ., Regularization multiplier is a parameter that leads to smoothening of the regression line to minimizing the error function and thus prevents over-fitting of the model ., It does so by penalizing the values of the features that tries to closely match the noisy data points resulting to balanced optimal solution to avoid making the model complex ., The model containing the optimal combination of predictor variables was run with ten replicates and 500 iterations at a convergence threshold of 0 . 00001 , with cross validation replicate type ., The output was set to logistic format , so that the predictions of habitat suitability would assume probability scores between 0 and 1 42 ., To determine which set of predictor variables best fit the data , performance and selection criteria were implemented using the MaxEnt software 42 and MaxEnt extension , ENMTools 53 ., A backward stepwise approach was implemented in MaxEnt using the jackknife test of relative contribution of the predictor variables in the model as follows ., Eight models were run in MaxEnt , starting with one that included all eight predictor variables ., In the process of building the model , the variable which contributed the least was removed from subsequent models until only one variable remained ., AUC values were recorded for each model ., The raw outputs from MaxEnt were further evaluated using ENMTools 53 ., The optimal combination of predictor variables included in the final model was the one that generated the largest AUC and at least one of the smallest of Akaike`s information criterion ( AIC ) , sample-size corrected Akaike`s information criterion ( AICc ) or Bayesian information criterion ( BIC ) 71 , 72 ., The percentage contribution and permutation importance were computed for each predictor variable ., The magnitude of change in training AUC represented by the average over the 10 replicate runs was normalized to percentages ., The higher the percentage contribution , the more impact that particular variable had on predicting the most suitable habitat for RVF occurrence 53 ., In order to assess the training gain of each predictor variable , the jackknife of regularized training gain was produced by running the model in isolation and comparing it to the training gain of the model including all variables ., This was used to identify the predictor variable that contributed the most individually to the habitat suitability for RVF occurrence ., The response curves describing the probability of RVF occurrence in relation to the different values of each predictor variable were generated using only the variable in question and disregarding all other variables ., The contribution of each predictor variable to the final model was assessed using the jackknife procedure based on the AUC , which provides a single measure of model performance 42 ., The probability scores ( numeric values between 0 and 1 ) were displayed in ArcGIS 10 . 2 ( ESRI East Africa ) to show the locations in Tanzania where RVF is predicted to be more or less likely to occur ., Ground-truthing of the ecological niche modelling outputs was conducted by comparing the levels of antibodies specific to RVFV in domestic ruminants ( sheep , goats and cattle ) sampled from locations in Tanzania that presented different predicted habitat suitability values ., We assumed that locations with higher proportions of RVFV-seropositive animals represented higher levels of habitat suitability for RVFV activity than locations with low proportions of seropositive animals ., The details of sampling process and laboratory analysis of serum samples have been described by Sindato and others 73 ., In brief , MaxEnt predictive map of habitat suitability for RVF occurrence ( Fig 1 ) was used as guidance to purposively identify six villages from six districts in the eastern and western Rift Valley ecosystems of Tanzania as described elsewhere ( 73 ) ., The district veterinary officers were consulted in order to identify one district within the region perceived to be at highest risk of RVF occurrence ., Criteria used included presence of shallow depressions/locations that are subject to regular flooding , ecological features suitable for mosquito breeding and survival/experience of mosquito swarms during the rainy season , relatively high concentration of domestic ruminants , proximity to forest , rivers , lakes , wildlife and presence of areas with history of RVF occurrence ., The district within the region that was identified to have most of these epidemiological characteristics was selected for the study , even if they had never reported RVF outbreaks ., Utilizing local veterinary records , only the villages with livestock that have never been vaccinated against RVF were targeted ., Based on the above criteria for identifying the six study districts , additional discussions were then held with local veterinary/agricultural staff , community leaders and livestock keepers to identify one village within each district that was perceived to be at highest risk for RVFV activity ., The number of villages surveyed was not based on statistical considerations , but rather logistical and financial factors ., The selected villages from the eastern Rift Valley ecosystem were Chamae , Malambo and Ninchoka , and all had reported RVF outbreaks in the past ., Selected villages from the western Rift Valley ecosystem were Bukirilo , Nyakasimbi and Kajunjumele , and all had never reported RVF outbreaks ., Nyakasimbi village is located in Karagwe district in the western Tanzania bordering with Rwanda , and Kajunjumele village is located in Kyela district in the southern highland bordering with Lake Nyasa ., Ninchoka and Malambo villages are located in Serengeti and Ngorongoro districts , respectively , in the northern Tanzania bordering with Kenya ., Bukirilo and Chamae villages are in Kibondo and Kongwa districts in the western and central Tanzania , respectively ., Within each selected village a two stage random sampling process was used to select the herds and domestic ruminants ., In each of the selected villages , 20 herds keeping at least one of the three domestic ruminant species ( cattle , sheep and/or goats ) were randomly selected from the list of livestock keepers ., Within each herd , a maximum of 20 ruminant animals ( not more than 20 animals were selected from a herd ) born after the last RVF outbreak in 2006/2007 in Tanzania were bled ( i . e . 10 cattle , 5 goats and 5 sheep ) depending on the herd size and species composition within the herd at the time of sampling ., Collected serum samples were tested for the presence of anti-RVFV antibodies using IgM-capture ELISA 74 and inhibition ELISA 75 ., The results were interpreted using the cut-off threshold specified by the manufacturer of the test kit ., For IgM capture ELISA method: Sheep , goat and bovine sera producing PP values ≥7 . 9 , 9 . 5 and ≥ 14 . 3 , respectively , were considered to be positive and less than these values as negative 74 ., For RVF inhibition ELISA method: Serum samples with PI equal to or greater than 41 . 9 , 41 . 4 and 38 . 4 were considered seropositive for RVF inhibition in cattle , goats and sheep , respectively 75 ., The data were analysed using logistic regression modelling to investigate the association between various suitability habitat values ( potential predictors ) and RVFV seropositivity outcomes in domestic ruminants ., Based on the limited resources available and logistic factors , the study sites for model ground-truthing were not selected using simple random sampling approach but rather using a purposive sampling approach ., When any sampling method other than simple random sampling is used , the survey data analysis method is used to take into account the differences between the design that was used and simple random sampling ., This is because the sampling design affects both the calculation of the point estimates and the standard errors of the estimates ( e . g . regression coefficients ) ., When non-independent sampling process is not accounted for in the analysis the standard errors will likely be underestimated , possibly leading to results that seem to be statistically significant , when in fact , they are not ., The svy command was therefore used in the modelling process using Stata version 12 ( Statacorp , College Station , TX , USA ) to account for sample survey design effect ., Eight predictor variables , namely mean diurnal temperature range , precipitation of wettest quarter , elevation , soil type , livestock density , rainfall pattern , proximity to wild animal protected areas and proximity to forest were initially evaluated in the model ., The pair-wise correlation matrix for these predictor variables suggested that there was moderate correlation between mean diurnal temperature range and precipitation of wettest quarter ( r = -0 . 56 ) , precipitation of wettest quarter and livestock density ( r =—0 . 58 ) and rainfall pattern and livestock density ( r = 0 . 62 ) ( Table 1 ) ., Of the eight predictor variables evaluated in the initial model , four—proximity to forest , proximity to wild animal protected areas , elevation and mean diurnal temperature range—were dropped from the model leaving four predictor variables in the final model ( Model_4; Table 2 ) ., This model , which contained the predictor variables soil type , precipitation of wettest quarter , livestock density and rainfall pattern was selected as the model of best fit based on the highest mean AUC and lowest BIC , as well as one of the lowest values of AIC and AICc ., All subsequent results refer to this model ., Soil type and precipitation of wettest quarter together accounted for almost two-third ( 64 . 8% ) , while livestock density and rainfall pattern together accounted for just over one-third ( 35 . 2% ) of the variation in habitat suitability for RVF occurrence ( Table 2 ) ., The habitat suitability of RVF occurrence in domestic ruminants in Tanzania was displayed on continuous probability scores of least to most suitable represented by a brown-green–colour scale ( Fig 1 ) ., Probability scores were mapped at district level and the grid size was 1km2 ., It is clear from our results that the habitat suitability of RVF occurrence was heterogeneously distributed throughout the country ., About one-third ( 29% ) of Tanzania Mainland area ( n = 883 , 343 Km2 ) comprising 10 ( 40% , n = 25 ) regions ( dark-green shades in the northern and central-eastern areas of the country ) represented highest probability scores and were considered most suitable for RVF occurrence ., Almost one-fifth ( 18% ) of the land area comprising three ( 12% , n = 25 ) regions represented by light-green in the central-southern areas of the country were considered moderately suitable for RVF occurrence ., Over half ( 53% ) of the land area comprising 12 regions ( 48% , n = 25 ) represented by light- and dark-brown in the western and south-eastern areas of the country were considered least suitable for RVF occurrence ., Predictive performance of the model was considered good with mean test AUC of 0 . 812 and standard deviation of the mean probability of 0 . 014 for the 10 replicate runs ., The results of the jackknife regularized training gain indicated that the predictor variable with the highest gain when used in isolation was livestock density ., The predictor variable that decreased the gain the most when it was omitted was soil type ., Values shown are averages over 10 replicate runs ( Fig 2 ) ., Jackknife test of variable importance utilizing the AUC showed that livestock density contributed the most to the AUC ( longest dark-blue bar ) , followed by precipitation of the wettest quarter , soil type and rainfall pattern ( Fig 3 ) ., The response graphs for the final model showed that probability scores were highest in areas with impermeable soils ( planosols followed by chernozems , andosols , luvisols and acrisols ) , while the lowest probability scores were observed in locations with permeable soils ( ferralsols , cambisols and lixisols ) ( Fig 4 ) ., The areas that experienced a bimodal pattern of rainfall had much higher probability of RVF occurrence than those that experienced a unimodal rainfall pattern ., Probability of RVF occurrence was very low ( around 0 . 26 ) at minimum values livestock density of < 8 heads/km2 ., It then followed a sigmoidal pattern with an initial increase in probability occurring between 8 and 46 heads/km2 , a rapid increase between 46 and 48 heads/km2and after 150 heads/km2 the probability of RVF occurrence remained constant ( Fig 5 ) ., Probability of RVF occurrence was around 0 . 62 at the precipitation of the wettest quarter of < 275mm ., A sharp increase in the probability of RVF occurrence occurred with the precipitation of the wettest quarter between 275 and 290 mm ( Fig 6 ) ., The highest probability of RVF occurrence in relation to precipitation of the wettest quarter was 0 . 76 that occurred between 375 and 425 mm ., Then there was a sharp rate of decline in the probability between 425 and 430 mm , slower rate of decline between 430 and 590 mm and a further sharp decline to a probability of 0 . 60 between 590 and 595 mm ., Thereafter there was a further slower rate of decline in the probability to < 0 . 55 at around 1 , 075mm ( Fig 6 ) ., According to our ecological niche modelling algorithm; Ninchoka , Malambo and Chamae villages are located in the northern and central areas of the country considered most suitable for RVF occurrence while Kajunjumele , Nyakasimbi and Bukirilo villages are in the western and southern areas of the country considered least suitable areas ., A total of 1 , 435 domestic ruminants from 121 herds ( 61 herds from western and 60 herds from eastern Rift Valley ecosystem ) in these six villages were tested for antibodies against RVFV ., About an equal proportion of tested serum samples were collected in livestock from the villages in the districts within the eastern ( 51 . 9% ) and western ( 48 . 2% ) ecosystems of the Rift Valley ., The number of serum samples from each study village was: Malambo; 243 ( 16 . 9% ) , Ninchoka; 257 ( 17 . 9% ) , Chamae; 244 ( 17 . 0% ) , Nyakasimbi; 233 ( 16 . 3% ) , Bukirilo; 233 ( 16 . 3% ) and Kajunjumele; 225 ( 15 . 7% ) ., Ground-truthing of model outputs revealed a significant variation in the odds of RVFV seropositivity in livestock sampled from locations with different suitability habitat values for RVF occurrence ., The odds of an animal sampled from the most suitable location being seropositive for RVFV were two times higher than the odds of an animal sampled from least suitable areas ( OR = 2 . 0 , 95% CI: 1 . 43 , 2 . 76 , p < 0 . 001 ) ., The ecological niche modelling implemented in this study illustrates the extent of suitable habitat for RVF occurrence in Tanzania ., The results suggest that the northern and central-eastern Tanzania have a higher probability of RVF occurrence than the rest of the country ., Our model predicted areas of suitable habitat , the western and south-eastern areas of the country , beyond the known localities of RVFV activity ., The modelled most suitable habitat for RVF occurrence in this study is characterized by high livestock density , moderate precipitation in the wettest quarter , predominantly impermeable soils and bimodal rainfall pattern ., The findings of this study provide scientific evidence that can inform the design of cost-effective RVF prevention and control programmes targeting the identified high risk locations .

Introduction, Methods, Results, Discussion

Despite the long history of Rift Valley fever ( RVF ) in Tanzania , extent of its suitable habitat in the country remains unclear ., In this study we investigated potential effects of temperature , precipitation , elevation , soil type , livestock density , rainfall pattern , proximity to wild animals , protected areas and forest on the habitat suitability for RVF occurrence in Tanzania ., Presence-only records of 193 RVF outbreak locations from 1930 to 2007 together with potential predictor variables were used to model and map the suitable habitats for RVF occurrence using ecological niche modelling ., Ground-truthing of the model outputs was conducted by comparing the levels of RVF virus specific antibodies in cattle , sheep and goats sampled from locations in Tanzania that presented different predicted habitat suitability values ., Habitat suitability values for RVF occurrence were higher in the northern and central-eastern regions of Tanzania than the rest of the regions in the country ., Soil type and precipitation of the wettest quarter contributed equally to habitat suitability ( 32 . 4% each ) , followed by livestock density ( 25 . 9% ) and rainfall pattern ( 9 . 3% ) ., Ground-truthing of model outputs revealed that the odds of an animal being seropositive for RVFV when sampled from areas predicted to be most suitable for RVF occurrence were twice the odds of an animal sampled from areas least suitable for RVF occurrence ( 95% CI: 1 . 43 , 2 . 76 , p < 0 . 001 ) ., The regions in the northern and central-eastern Tanzania were more suitable for RVF occurrence than the rest of the regions in the country ., The modelled suitable habitat is characterised by impermeable soils , moderate precipitation in the wettest quarter , high livestock density and a bimodal rainfall pattern ., The findings of this study should provide guidance for the design of appropriate RVF surveillance , prevention and control strategies which target areas with these characteristics .

Rift Valley fever is a viral disease that is transmitted to livestock and humans by mosquitoes ., Humans get infected mainly through direct contact with blood or aborted materials from infected animals ., In Tanzania , a total of 10 RVF epidemics have been reported from 1930 to 2007 ., Despite the long history of RVF in Tanzania , the extent of its suitable habitat remains unclear ., As a result , disease prevention measures such as vaccination of livestock are implemented without informed risk-based resource-allocation decisions ., This study was therefore carried out to identify the locations in Tanzania where RVF is more or less likely to occur using an ecological niche modelling ( ENM ) method ., Data from 193RVF outbreak locations were used together with precipitation , elevation , soil type , livestock density , rainfall pattern , proximity to wild animal protected areas as well as to forest as inputs for ENM ., Our results show that locations at most risk for RVF occurrence were the northern and central-eastern Tanzania ., Areas at highest risk for RVF occurrence are characterised by soils with low water permeability , high amounts of rainfall , high livestock density and two rainy seasons in a year ., The findings of this study provide guidance in the design of appropriate RVF surveillance , prevention and control strategies that can be implemented cost-effectively by targeting the areas at most risk .

livestock, medicine and health sciences, ecology and environmental sciences, rift valley fever virus, ecological niches, pathology and laboratory medicine, atmospheric science, pathogens, tropical diseases, geographical locations, microbiology, viruses, rift valley fever, tanzania, rna viruses, habitats, neglected tropical diseases, africa, veterinary science, bunyaviruses, infectious diseases, veterinary diseases, zoonoses, medical microbiology, microbial pathogens, agriculture, people and places, rain, meteorology, ecology, earth sciences, viral pathogens, biology and life sciences, viral diseases, organisms

journal.pcbi.1004149

Within-Host Stochastic Emergence Dynamics of Immune-Escape Mutants

Parasites and pathogens pose a continuous threat to human , livestock , and plant health since new strains can readily emerge , via mutation or recombination , from pre-existing strains ., Generally , the focus has been on detection of emerging diseases at the population level , in order to track and control their spread 1 , 2 ., Modelling approaches to predicting emergence have therefore primarily concentrated on detecting infections arising between individual hosts 3 , 4 , and the contribution of within-host processes to pathogen emergence has often been overlooked ., It is now well known that within-host evolution has strong effects on the epidemiology of many pathogens ( reviewed in 5 ) , and can substantially affect the course of an infection , as illustrated by the cases of HIV 6 and hepatits C virus ( HCV ) 7 ., Each step of the within-host evolutionary dynamics consists of the emergence of a rare mutant that takes over the pathogen population ., Since mutated infections always initially appear as a few copies , they are prone to extinction so their emergence is best captured using stochastic dynamics ( as opposed to deterministic approaches ) ., Only a handful of previous models have investigated this stochastic within-host process ., A widespread use of within-host models in static populations is to calculate the probability that infections will evolve drug resistance , and what regime is needed to avoid treatment failure 8–11 ., One exception 12 studied viral emergence within a host , if most mutations were deleterious , in order to determine how different viral replication mechanisms affected the establishment of beneficial alleles ., Parasite and pathogen evolution can radically affect the course of infections in hosts able to mount immune responses ., For instance , HIV is known to successfully fix mutations within individuals which enable it to evade immune pressures 6 ., This is in line with evidence that target cell limitation cannot account for HIV dynamics , and that immune limitation also needs to be present 13 ., Arguably , this continual evasion of immunity prompts the chronic nature of HIV infections ( see 14 , 15 for an illustration of these ‘Red Queen’ dynamics ) ., In the case of HCV , it has also been found that chronic infections were associated with higher rates of within-host evolution 16 ., Concerning a different chronic disease , it is now well-known that the ability of malignant cells to rapidly expand in size , ignoring biological signals to arrest growth ( e . g . through mutating the p53 tumour suppression gene ) , and escaping the patient’s immune system are key steps in cancer development 17 ., All these scenarios can be analysed in the larger framework of evolutionary rescue 18 , 19 , where a change in the environment ( in this case , the activation of an immune response ) will cause the population to go extinct unless it evolves ( develops an increased replicative ability , then subsequently rise to a large enough size to avoid stochastic loss ) ., Although chronic infections are those most likely to evolve strains that evade immune pressures , evidence exists that such immune escape also plays a role in acute infections , like influenza 20 ., Note that the latter evidence arises from serial passage experiments; although these tend to maximise selection at the within-host level , they still include a slight selective pressure for increased transmission ., Modelling mutant emergence in an immune evasion context raises a technical challenge , as it is a non-equilibrium process ., In most of the within-host models presented above 8–10 , 12 , it is sufficient to know the initial state of the system to calculate emergence probability of immune or treatment escape ., A non-equilibrium model was studied by Alexander and Bonhoeffer 11 , who accounted for the reduction in available target cells when determining the emergence of drug resistance using an evolutionary rescue framework ., With immune-escape however , the problem is that the immune state when the mutant infection appears is only temporary ., In this case , there will be an initial strain present within the host , which triggers immunity ., This strain can then mutate into a faster-replicating form , but if the mutated strain arises at a low frequency , immunity can destroy it before it has a chance to spread ., This effect can be exacerbated by the fact that the mutated strain also prompts an increased immune response , so the emerging infection has a stronger defence to initially compete with ( assuming immune growth is proportional to the total size of the pathogen population ) ., This feedback , where increased immune growth prevents emergence of mutated strains with higher replication rates , can strongly affect the appearance of mutated strains within-host , and needs to be accounted for ., Existing emergence models have not yet accounted for such within-host population feedbacks , especially those arising from the immune system ., Recently , Hartfield and Alizon 21 tackled a related problem , regarding how epidemiological feedbacks affect disease emergence at the host population level ., In their model , a faster-replicating strain emerged via mutation from a pre-existing infection; however , the continuing outbreak caused by the initial strain removed susceptible individuals from the population , which limited the initial spread of the mutated strain ., It was shown that the ongoing depletion of susceptible individuals due to the initial strain spreading has a stronger effect on reducing pathogen emergence , than assumed by just scaling down the reproductive ratio by the frequency of susceptible individuals present when the mutated infection appears ., That is , the feedback produced by the first strain in removing susceptible individuals caused a drastic decrease in the emergence probability of the mutated strain ., Building on this previous study , we derive here an analytical approximation for the probability that an immune-escape mutant will emerge and maintain itself within a host ., We use ‘immune-escape’ in the sense that while the mutated strain can be killed by immunity , it can ultimately outgrow immune growth and chronically persist ., Furthermore , we use an acute infection setting , which are commonly used to study the within-host dynamics of ‘flu-like’ diseases 22–28 ., Analysis of the model demonstrates how the ongoing proliferation of immune cells acts to decrease the emergence probability of mutated strains ., Acute infection models are also useful in studying the first stages of chronic infections , where one observes exponential growth of the virus population , followed by a decline in the first weeks of infection ., In addition , there are two questions that are worth investigating with this model from a biological standpoint ., First , what is the fittest evolutionary strategy for an escape mutant: is it to overgrow the immune response ( that is , increase its inherent replication rate to enable its persistence , even when immune cells are at capacity ) , or to tolerate it ( prevent immunity from killing as many pathogens per immune cell ) ?, Both these actions will lead to an increase in the pathogen’s net reproduction rate and can thus be described as immune evasion , but it is unclear if one process is favoured over the other ., In addition , note that disabling the immune response is not possible , since the wild-type infection activates it ., Second , to what extent do we need to account for the ongoing proliferation of the immune response ?, In other words , if we calculate the emergence probability based on the system state when the mutation occurs , how inaccurate would this estimate be ?, We end by discussing our results in the light of what is known about within-host evolution for several human infections ., In order to find an analytical solution for the within-host emergence of a mutated strain , we follow the approach of Hartfield and Alizon 21 , which investigated the appearance of a new infectious pathogen from a pre-existing strain at the population level ., To consider the within-host case , we construct and subsequently analyse a specific scenario of acute , immunising infections ., Here , the first strain will go extinct because it does not replicate at a high enough rate ., However , before vanishing , it can mutate into a form that grows unboundedly; we are interested in calculating the probability of this event occurring ., We focus on this case to cover a general range of within-host evolution scenarios , which is important since there is yet no consensus on how to best model within-host infections ( reviewed in 29 ) ., Although it is feasible that the mutated strain has a maximum population size , mathematical analysis of pathogen emergence only needs to consider the dynamics of mutants when they are present in a few copies , rather than the long-term behaviour once they have already established ., Therefore , the general results outlined in this paper are also broadly applicable to cases where the infection population sizes are bounded ., Our analytical approach involves using a set of deterministic differential equations to ascertain pathogen spread in a stochastic birth-death process , where an infection ( or immune cell ) can only either die or produce 1 offspring ., This process is one of the most common ways of investigating stochastic disease spread 30 ., A list of nomenclature used in the model is outlined in Table 1 . Assume there exists an initial pathogen , or infected cell-line , the size of which at time t is denoted x1 ( t ) ., This line grows in size over time according to the following equation , which is well-used for within-host infection models 29:, d x 1 d t = x 1 ( φ 1 - σ 1 y ) ( 1 ), Here , φ1 is the growth rate of the infection , σ1 is the rate of destruction of the pathogen per immune cell , and y ( t ) is the number of immune cells ( i . e . lymphocytes ) ., For simplicity , we assume that there is complete cross-immunity between the various pathogen strains , so it is not necessary to model immune cell diversity ., The growth of the immune cell population is modelled using a logistic-growth curve:, d y d t = r x 1 y 1 - y K ( 2 ), Here , r is the proliferation rate of immune cells , and K is the maximum population size they can achieve ., This formulation is an extension of the model developed by Gilchrist and Sasaki 24 to study acute infections ., The main difference in that in their model , the density of immune cells is allowed to reach any value in order to clear the infection ., Here , we impose that immune density does not go above a maximum threshold K , which correspond to an intrinsic limitation in the host resources allocated to immunity ., Profile plots of typical infection responses are shown in Section 1 of S1 Text ., If φ1/σ1 < K , then the first strain will increase in size until the immune-cells reach a maximum ., After this point , the infection will decrease towards eventual extinction , while the immune response will be maintained at a non-zero size ., However , if φ1/σ1 ≥ K then the first strain will continue to expand ., A formal rational for this behaviour will be shown below ., To proceed with finding an analytical solution for the emergence probability , we proceed as in previous analyses 21 , 24 , 31 , and note that since y is monotonically increasing , we can use the immune cell population size as a surrogate measure of time ., By dividing Equation 1 by Equation 2 , we obtain a differential equation for x1 as a function of y:, d x 1 d y = ( φ 1 - σ 1 y ) x 1 r x 1 y ( 1 - y K ) ( 3 ) To simplify subsequent analyses , we make the following substitutions ., We define the reproductive rate of the infection , where there is a single immune cell ( y = 1 ) equals R1 = φ1/σ1 ., We also set ρ = r/σ1 ( this can be formally shown by rescaling time by τ = σ1t ) ., We use the notation R1 to draw parallels between the scaled pathogenic replication rate , and the reproductive ratio R0 in population-level , epidemiological models 32 ., After making the required substitutions , Equation 3 can be rewritten as:, d x 1 d y = K ( R 1 - y ) ρ y ( K - y ) ( 4 ) Equation 4 formally shows that the infected cell line increases in size ( dx1/dy > 0 ) if y < R1 = φ1/σ1 , and decreases if y > R1 ., A corollary of this result is that if R1 of a infection exceeds K , then it cannot go extinct in the long term ., This differential equation is straightforward to solve ( Section 1 of S1 Text ) , and yields the following function for x1 ( y ) :, x 1 ( y ) = x i n i t + 1 ρ log y y i n i t R 1 K - y K - y i n i t ( K - R 1 ) ( 5 ), where xinit and yinit are , respectively , the number of pathogen and immune cells at the start of the process ( time t = 0 ) , and log is the natural logarithm ., It is clear from Equation 4 that the maximum value of x1 occurs for y = R1 ., By substituting this value into Equation 5 , we obtain the maximum value of x1 ( denoted xM ) as:, x M = x i n i t + 1 ρ log R 1 y i n i t R 1 K - R 1 K - y i n i t ( K - R 1 ) ( 6 ) Note that ρ has no effect on the position of the peak ( that is , the value of y leading to the maximal infection level ) ., Since it affects the growth rate of the immune response ( and therefore also of the pathogen ) , it does determine the maximum value itself ., As this maximum is inversely proportional to ρ , smaller immune growth rates lead to larger peaks ., Finally , the maximum infection time ( as a function of y ) needed for the first infection to go extinct can be determined by solving x1 ( y ) = 0 numerically ( Section 1 of S1 Text ) ., Our goal is to calculate the probability of ‘evolutionary rescue’ ., That is , the initial strain has R1 < K so is guaranteed to go extinct in the long-term ., However , a mutated form ( with reproductive ratio R2 ) could arise with R2 > K , and if it does not go extinct when rare , it can outgrow immune proliferation ., Previous theory on the emergence of novel pathogenic strains 21 showed that if mutated strains arise at rate μ per time step , the overall emergence probability P is given by:, P = 1 - exp - μ ∫ y i n i t y M x 1 ( y ) Π ( y ) d y ( 7 ), for yM the maximum immune size for which it is possible for immune escape to arise , and Π ( y ) the emergence probability of an escape mutation were it to appear ., The formulation of each of these will be discussed in turn ., We verified our analytical solution by comparing it to simulation data ., Simulations were written in C and based on the Gillespie Algorithm with tau-leaping 35 , 36; source code has been deposited online in the Dryad data depository ( doi:10 . 5061/dryad . df1vk ) ., The time step was set to be very low: Δτ = 0 . 00005 ., This is because the tau-leaping algorithm is accurate only if the expected number of events per time step is small 37; since the growth rates of the pathogen strains and the lymphocytes are both large , a small time step is needed to make the simulation valid ., The growth of both the original and mutated strains are simulated using scaled parameter rates ., That is , the birth rate per time step for each pathogen is Poisson-distributed with mean ( Ri ⋅ xi ⋅ Δτ ) , and death rate with mean ( xi ⋅ y ⋅ Δτ ) for i = 1 , 2 . This is done to reduce the number of parameters in the model , and also enables accurate comparison with the scaled results ., The change in size of the immune response is determined by standard logistic-growth dynamics for the Gillespie algorithm ., That is , the Poisson mean number of births per time step equaled ρb ( x1 + x2 ) y , for ρb is the birth rate parameter , and the mean number of deaths equalled y ( x1 + x2 ) ( ρd + ρ ( y/K ) ) , where ρd is the death rate and ρ = ρb − ρd ., Since there are no distinct ρb and ρd terms in the model ( just ρ ) , then we set ρd = 1 and varied ρb , so ρ = ρb − 1 . The net growth rate ρ was varied , generally between 0 . 5 and 19 depending on the size of the other parameters ., Note that in the analytical solution , it is assumed that the immune response does not die off ., In order to maintain this assumption , we set yinit = 20 in the simulations ., This also makes intuitive sense , because it is unlikely that the initial immune response is limited to just one cell ., If the immune response goes extinct before both infected strains go extinct , or the mutated strain emerges , then that run is discarded , the simulation is reset and restarted ., In simulations , R1 is either set to 60 if K = 100 , R1 = 100 if K = 250 or 1000 , or R1 = 2 , 000 for K = 10 , 000 ., R2 was varied , ensuring that R2 > K so the infected strain can outgrow the immune response ( see Equation 4 ) ., The mutation rate was also varied over several orders of magnitude ., The first strain is reintroduced ( from an initial frequency of 1 cell ) 10 , 000 , 000 times as separate replication runs ., Since the emergence probabilities were predicted to be low , a large number of runs were needed in order to produced a meaningful estimate of emergence probability ., The second strain is said to have emerged once it exceeded 20 cells; since meaningful values of R2 were large , only a handful of cells would have been needed to guarantee emergence , hence a low outbreak threshold could be used 38 ., The first results we checked were individual profiles of the stochastic simulations , since these can be used to demonstrate the behaviour of within-host emergence ., In the majority of cases , the first strain increase in size , but once immune proliferation also reaches its maximum then the first strain goes extinct soon after ( Fig ., 2, ( a ) ) ., However , emergence of the mutated strain can occur even if the immune cells are at their maximum size ( Fig ., 2, ( b ) ) ., This is to be expected , given the form of Π ( Equation 10 ) , which demonstrates that emergence is likeliest once the immune cells reach a steady-state , so ongoing proliferation does not restrict their establishment ., Conversely , emergence probability is reduced if the immune response is spreading; this is because when the mutated infection is rare , it is less likely to reproduce as immune cells increase in number ., Since the mutated infection population is only present in a few copies , this negative effect this immune growth has on reproductive ability will be drastic ( see also equivalent population genetics results by Otto and Whitlock 39 ) ., Fig . 3 compares the full analytical solution ( Equation 7 , with Π given by Equation 10 ) against simulation data ., If K = 100 , R1 = 60 , we see that there is an accurate overlap between the two for a variety of mutation rates that span several orders of magnitude ( Fig ., 3, ( a ) and, ( b ) ) ., This match demonstrates how our analytical solutions can be used to accurately predict emergence probabilities in the face of different scenarios , including antibiotic resistance and tumour formation , as both these processes are characterised by high mutation rates ., We also tested a parameter set where the carrying capacity and initial growth rate was much higher ( K = 1 , 000 and R1 = 100 , or K = 10 , 000 and R1 = 2 , 000 ) ., Fig . 3, ( c ) - ( f ) demonstrates that the analytical results slightly underestimate the simulation results to a small degree , especially if the mutated strain’s growth rate is high and R2 is close to K , but becomes more accurate as R2 increases and generally provides a good approximation ., These inaccuracies probably arise due to our analytical solution not fully accounting for the increased variance in both infection and immune growth rates that can arise if parameters are large , as in this model 30 ., However , the error does not appear to be great , so the model can still be used to provide accurate estimates of emergence probability for this parameter set ., Finally , we also tested how well analytical solutions work for cases where R1 < R2 < K . Although such a mutated strain replicates more quickly , Equation 4 shows that it will die out in the long-term ., Hence our analytical solutions might not correctly reflect the emergence probability of these infections ., Nevertheless , even in this case , Equation 7 accurately matches up with simulation results in this parameter range , although some inaccuracies arise for K = 1 , 000 ( S1 Fig ) ., To exemplify why it is important to account for ongoing immune growth , we compared our full solution for the emergence probability ( Equation 7 , with Π equal to Equation 10 ) , to a ‘naive’ estimate that does not assume ongoing immune proliferation ( Π = 1 − 1/R2 ) ., We see that our result leads to a greatly reduced emergence probability , with values from our model being ∼1 , 000 times lower than the naive estimate ( Fig . 4 , and Section 3 of S1 Text ) ., Similar results apply if feedbacks only affect the mutant after it has arose ( that is , Π = ( 1 − P e x t * ) as given by Equation 9 ) ., Hence , as in previous non-equilibrium models 21 , one needs to account for dynamical feedbacks , both before and after the mutated strain appears , in order to account for the reduced emergence probability in these scenarios ., We next studied what process has a larger effect on pathogen emergence ., Immune escape can either be achieved by overgrowing the immune response ( increasing the intrinsic pathogen growth rate φ ) , or by tolerating it ( reducing the immune-mediated death rate σ ) ., One might expect that the two processes would lead to similar increase in escape probability , since both affect the effective reproductive rate R* ., However , this intuition need not hold in the face of immune-mediated feedbacks ., We commence with a heuristic analysis based on the deterministic model to predict general behaviour for a newly-emerging strain , then use numerical analyses to check this reasoning ., If there are two strains spreading concurrently , the deterministic rate of change of immunity , y , and the second strain x2 , is given by the following set of differential equations:, d x 2 d t = x 2 ( φ 2 - σ 2 y ) ( 11a ), d y d t = r ( x 1 + x 2 ) y 1 - y K ( 11b ) Further recall that that the basic pathogen growth rate with only one immune cell is R2 = φ2/σ2 ., In order to augment its spread , the infection can either increase its intrinsic growth rate by a certain factor ( i . e . change φ2 → φ2 c , where c > 1 is a numerical constant ) , or instead tolerate the immune response ( mathematically equivalent to the transformation σ2 → σ2/c ) ., We can investigate the effect of both these rescaled variables on the rate of change of pathogen spread by substituting them into Equation 4 ., After making the substitution φ2 → φ2 c , the pathogen rate of change becomes:, d x 2 d y = K ( c R 2 - y ) ρ y ( K - y ) x 2 x 1 + x 2 σ 2 σ 1 ( 12 ), The x2/ ( x1 + x2 ) term arises due to the presence of the pre-existing initial strain x1 , and the death-rate ratio σ2/σ1 appears since ρ was initially scaled by σ1; this term disappears if we assume equal death rates ., Apart from these terms , Equation 12 is conceptually the same as Equation 4 , but with R scaled by a factor c , as expected ., However , if we instead scale σ2 → σ2 / c , a different result emerges:, d x 2 d y = K ( c R 2 - y ) c ρ y ( K - y ) x 2 x 1 + x 2 σ 2 σ 1 ( 13 ) Here , not only is the reproductive ratio R2 scaled , but the immune growth term ρ is also altered ., Mathematically , this is a simple consequence of the fact that the growth rate is scaled by 1/σ1 , so any rescaling of σ2 also affects ρ through its effect on the σ2/σ1 ratio ., Biologically , this outcome reflects the fact that a rescaling of immune-mediated death would not have the same effect on the growth rate than changing φ2 by the same ratio , since pathogen death is also a function of the immune population , y ( see Equation 1 ) ., So although tolerating immunity might increase the infection growth rate , it comes at a cost of increasing the effective growth rate of immunity , as each immune cell would have a larger average impact on pathogen death ., Therefore , while a rise in growth rate ( φ2 ) would only affect R2 , reducing the death rate ( σ2 ) comes at a cost of increasing the effective proliferation rate of immune cells ., Hence , one expects that it is more advantageous for an infection to increase φ2 and outgrow the immune response , instead of reducing σ2 and tolerating it ., Furthermore , note that the emergence probability Π in the presence of epidemiological feedbacks contains a term of order 1/ρ ( as with dx2/dy ) , so the effective rise in immunity will further lead to a negative overall effect on emergence probability ., We verified this intuition by comparing the analytical results for cases where φ2 is increased by a set factor ( so that only R2 is changed by this rescaling ) , to outcomes where σ2 is scaled ( which will not just affect R2 but will also increase ρ by the same factor , as outlined above ) ., Fig . 5, ( a ) demonstrates how , for large parameter values , increasing φ2 only produces a higher overall emergence probability ., Qualitatively similar results arise if using different parameter values ., Our model can also be used to shed light on different processes of within-host emergence that have been observed in clinical studies ., Different diseases show varying outcomes with regards to the production of immune-escape mutations ., Two extensively-studied human diseases , HIV and HCV , are both characterised by a successive emergence of new strains over time ., In particular , HIV is well-known to produce ‘escape mutations’ that evade T cell-mediated immunity 43 , 44 ., On a within-host phylogeny , this behaviour is characterised by creation of new subclades 5–7 ., This behaviour can be intuitively explained by noting that HIV has extremely high mutation rates , estimated at ∼0 . 2 errors made per replication cycle and the ability to produce 1010 – 1012 virions per day 6 ( although mutation creation might be limited by a lower Ne , which has been estimated to lie between 1 , 000 50 and 100 , 000 51 ) ., Furthermore , a sizeable proportion of mutations are beneficial , with estimates of adaptive substitutions in the env gene placed at ∼55% 52 ( although the actual proportion of spontaneous beneficial mutations will be less than this , as observed from mutagenesis studies with viruses 53 , 54 ) ., What may seem surprising is that given this evolutionary potential , we do not see a more pronounced increase in HIV replication rate throughout the course of an infection ., Yet , conversely , immune escape is very strongly selected for 14 ., One possibility could be that given the constrains on RNA virus genomes 55 , evading the immune response might be easier to achieve than increasing the replication rate ., Furthermore , if the maximum immunity population size is high , it could be too complicated mechanistically to outgrow it , rather than than simply evading it ., HCV also shows a similar propensity to produce immune-escape mutations , with an estimated substitution rate of 1 . 2×10−4 per replication cycle and 1012 virions generated each day 7 ( although , as with HIV , the effective population size is greatly lower than this value 7 ) ., Acute hepatitis C infections are characterised by little diversity accumulating over time , except in one specific genetic region ( NS5B; 56 ) ., Chronic infections accumulate much more diversity , in line with the hypothesis that they continuously evolve to evade the immune system 16 ., These infections are also characterised by different immune profiles: acute infections lead to a high , sustained immune response , which tends to be greatly lowered in chronic infections 57 ., Our model suggests that if the immune response naturally increased rapidly ( i . e . a higher intrinsic ρ in the model ) , it can greatly limit the emergence of new pathogens as it rises , preventing immune escape and a chronic illness ., There exists evidence that mutations in hosts correlate with infection outcome , mainly due to SNPs at the IL28B loci , which could cause this effect 58–60 ., However , it remains controversial as to what determines virus clearance , especially since there exist evidence for virus control over infection outcome 61 ., Therefore , the effect on immune response on the creation of escape mutations requires further study ., Other diseases are characterised by low probability of emergence despite frequent mutation , for which immune feedbacks could be the cause ., Cancer growth is characterised by evading pre-programmed cell death and extremely rapid cell replication 17 ., Furthermore , genomes present in cancer cells usually carry ‘mutator’ alleles , causing additional cell instability 62 ., Therefore , cell mutation can be extremely common , but can generally be stopped by the immune response ., Yet it is known that tumours can mutate immune-checkpoint networks to generate protection from the immune system ., One prominent example is the up-regulation of ligands for the programmed cell death protein 1 ( PD1 ) pathway , which can block antitumour immune responses 63 ., Our model suggests that due to the rapid replication of tumour cells , they could also strongly trigger a heightened immune rate , the increased spread of which will greatly prevent cancer emergence ., This mechanism could explain why tumour emergence is rare relative to the potential mutation rate ., By accounting for the ongoing spread of immunity , we have quantified how this particular effect can feedback onto emergence of mutated pathogens intra-host , and inhibit the appearance of mutated strains ., Further analysis of the model demonstrates how it is beneficial for a pathogen to increase its replication rate and attempt to outgrow immunity , as opposed to tolerate it ., This model can therefore shed light on expected within-host evolutionary dynamics of infections , as well as determine why extremely rapidly replicating pathogens do not emerge as often as expected .

Introduction, Materials and Methods, Results, Discussion

Predicting the emergence of new pathogenic strains is a key goal of evolutionary epidemiology ., However , the majority of existing studies have focussed on emergence at the population level , and not within a host ., In particular , the coexistence of pre-existing and mutated strains triggers a heightened immune response due to the larger total pathogen population; this feedback can smother mutated strains before they reach an ample size and establish ., Here , we extend previous work for measuring emergence probabilities in non-equilibrium populations , to within-host models of acute infections ., We create a mathematical model to investigate the emergence probability of a fitter strain if it mutates from a self-limiting strain that is guaranteed to go extinct in the long-term ., We show that ongoing immune cell proliferation during the initial stages of infection causes a drastic reduction in the probability of emergence of mutated strains; we further outline how this effect can be accurately measured ., Further analysis of the model shows that , in the short-term , mutant strains that enlarge their replication rate due to evolving an increased growth rate are more favoured than strains that suffer a lower immune-mediated death rate ( ‘immune tolerance’ ) , as the latter does not completely evade ongoing immune proliferation due to inter-parasitic competition ., We end by discussing the model in relation to within-host evolution of human pathogens ( including HIV , hepatitis C virus , and cancer ) , and how ongoing immune growth can affect their evolutionary dynamics .

The ongoing evolution of infectious diseases provides a constant health threat ., This evolution can either result in the production of new pathogens , or new strains of existing pathogens that escape prevailing drug treatments or immune responses ., The latter process , also known as immune escape , is a predominant reason for the persistence of several viruses , including HIV and hepatitis C virus ( HCV ) , in their human host ., As a consequence , the within-host emergence of new strains has been the intense focus of modelling studies ., However , existing models have neglected important feedbacks that affects this emergence probability ., Specifically , once a mutated pathogen arises that spreads more quickly than the initial ( resident ) strain , it potentially triggers a heightened immune response that can eliminate the mutated strain before it spreads ., Our study outlines novel mathematical modelling techniques that accurately quantify how ongoing immune growth reduces the emergence probability of mutated pathogenic strains over the course of an infection ., Analysis of this model suggests that , in order to enlarge its emergence probability , it is evolutionary beneficial for a mutated strain to increase its growth rate rather than tolerate immunity by having a lower immune-mediated death-rate ., Our model can be readily applied to existing within-host data , as demonstrated with application to HIV , HCV , and cancer dynamics .

journal.pgen.1000187

The Genomic Distribution and Function of Histone Variant HTZ-1 during C. elegans Embryogenesis

In genomes ranging from protozoa to humans , specialized regions of chromatin are created by the local incorporation of variant histones into nucleosomes ., The histone H2A variant H2A . Z is one such highly conserved variant , though the biophysical and biological function of H2A . Z incorporation into chromatin remains unresolved ., Evidence from Tetrahymena suggests a function for H2A . Z in transcriptional activation due to its localization to the transcriptionally active macronucleus 1–3 ., This function is consistent with genome-wide studies of Htz1 occupancy in S . cerevisiae ( hereafter “yeast” ) , which revealed Htz1 incorporation flanking a nucleosome-free region upstream of most genes ., It has been hypothesized that H2A . Z-containing nucleosomes may contribute to transcriptional activation by being less stable than H2A-containing nucleosomes 4–6 ., However , others have reported that H2A . Z-containing nucleosomes are in fact slightly more stable than canonical nucleosomes 7–9 ., This seeming contradiction may have been resolved by studies examining H2A . Z in combination with the histone H3 variant H3 . 3 ., In combination with histone H3 , H2A . Z nucleosomes were at least as stable as H2A nucleosomes , but the combination of H2A . Z and H3 . 3 results in highly unstable nucleosomes 4 ., This instability in conjunction with H3 . 3 could facilitate timely and efficient gene activation ., Indeed , in yeast cells lacking H2A . Z , the activation of genes in response to heat shock or galactose is delayed , and recruitment of RNA polymerase II and TATA-binding protein to responsive promoters is diminished 10 , 11 ., H2A . Z is also required for a form of “transcriptional memory” in yeast , in which recently transcribed chromatin is retained at the nuclear membrane to allow rapid re-activation of the gene 12 ., Recent high-resolution mapping of H2A . Z in human cells also revealed a positive correlation between H2A . Z occupancy and transcription , providing additional support for an H2A . Z function in transcriptional activation 13 ., Despite the wealth of evidence for a positive association between H2A . Z and transcription , other genetic and cytological evidence suggests that H2A . Z also functions in gene silencing ., The functional homolog of H2A . Z in Drosophila , H2Avd , is localized to both euchromatin and heterochromatin on polytene chromosomes , including the heterochromatic chromocenter 14 , 15 ., By genetic criteria , H2Avd is considered to have a repressive function ., H2Avd mutations are enhancers of Polycomb mutant phenotypes , suppressors of Trithorax group mutant phenotypes , and suppressors of position-effect variegation 16 ., Further evidence for a repressive function is found in mice , where H2A . Z promotes heterochromatin protein HP1α binding and co-localizes with HP1 at pericentric heterochromatin 17 , 18 ., In mammalian cells , mono-ubiquitylation of the H2A . Z C-terminus may distinguish “repressive H2A . Z” from “activating H2A . Z” , particularly on the silent X chromosome 19 ., Even within the yeast literature , there are conflicting conclusions regarding correlation with transcriptional activity and RNA Polymerase II ., One study found no correlation between Htz1 occupancy and transcription rate of the downstream gene 11 , while others reported an inverse correlation with transcription rate 20–22 ., The resolution of these apparently contradictory activating and silencing functions could be explained by a requirement for H2A . Z in regulating the precise timing and kinetics transcription , rather than simply promoting an “on” or “off” transcriptional state ., This potentiation of transcription would be especially critical during periods of dynamic transcriptional regulation , such as occurs in development and environmental responses ., There is growing evidence for this hypothesis ., In C . elegans , knockdown of htz-1 by RNAi caused expression of genes dependent on the FoxA transcription factor PHA-4 to be delayed 23 ., Furthermore , HTZ-1 and components of the C . elegans Swr1 complex ( SSL-1 ) required for HTZ-1 deposition have been identified in genetic screens for suppressors of vulval induction , a process highly dependent on precise timing of transcriptional cascades and tightly coordinated with cell divisions 24 , 25 ., Another clue to the function of H2A . Z may lie in the fact that it is required for viability in all metazoans tested 26–29 , but is not required for viability in single-celled yeast ., A lack of H2A . Z during metazoan development typically causes defects that lead to late embryonic lethality 23 , 26 , 27 , 29 , 30 ., This is consistent with expression of H2A . Z in mice , where the undifferentiated cells of the inner cell mass have low H2A . Z protein levels , with H2A . Z protein levels increasing as the cells differentiate into extraembryonic endoderm 17 ., Whether H2A . Z has been associated with gene activation or repression in one study versus another may not represent a universal regulatory function for H2A . Z , but may instead be a reflection of the specific biological conditions under which the function of H2A . Z was observed in a given experiment , and the temporal resolution of the particular assays employed ., In this light and with a focus on development , we used Chromatin ImmunoPrecipitation on DNA microarrays ( ChIP-chip ) , genetic mutation , and RNAi to interrogate the function of HTZ-1 during embryogenesis in C . elegans ., HTZ-1 knockdown by RNAi has been previously shown to cause embryonic lethality 23 ., To further characterize the function of HTZ-1 ( R08C7 . 3 ) in C . elegans development , we analyzed animals harboring a deletion in the C . elegans htz-1 gene ., The mutant htz-1 ( tm2469 ) contains a deletion of 345 bp of the htz-1 gene , thereby eliminating 97 of the 140 predicted amino acids and making it a likely genetic null ., The majority of homozygous htz-1 ( tm2469 ) offspring from htz-1 ( tm2469 ) /+ heterozygotes ( denoted as maternal +; zygotic − , or M+Z− ) animals are rescued from embryonic lethality by a maternal contribution of HTZ-1 ., These rescued animals develop into worms exhibiting grossly normal morphology and germ cell proliferation until late adulthood ( Figure 1A–B ) ., Of the M+Z− animals that reach adulthood , 80% are sterile and do not generate any embryos , instead producing unfertilized oocytes that eventually fill the uterus ( Figure 1B ) ., In 20% of the rescued animals , M−Z− embryos are observed in the uterus ( Figure 1C ) ., None of the embryos produced by M+Z− mothers were expelled from the uterus onto plates , indicating that the M+Z− mothers have an egg-laying defect ( Egl ) ., Somewhat unexpectedly , 28% of the M−Z− embryos ( n\u200a=\u200a32 ) progressed through embryogenesis to produce a few hatched larvae ., All of these M−Z− escapers arrest at the first larval stage ( Figure 1D–E ) ., The M−Z− embryos that hatched tended to arise from the first few eggs produced by M+Z− mothers , suggesting that in these animals HTZ-1 were still maternally provided at very low levels , but subsequent divisions of the germ cell precursors diluted HTZ-1 such that later embryos received a level below that required for viability ., The viability and semi-fertility of the htz-1 ( tm2469 ) M+Z− offspring suggested that the maternal load of HTZ-1 received by an embryo is sufficient to allow it to reach adulthood with defects limited to germ cells and specification of cells in post-embryonic lineages , for example vulval development ., To test this , we targeted the maternal complement of htz-1 mRNA using RNAi ., Direct injection of dsRNA into the gonad of adult wild-type animals produced a more severe phenotype than was observed in M+Z− offspring ., Instead , the RNAi phenotypes are consistent with those observed in htz-1 ( tm2469 ) M−Z− embryos ., Specifically , embryonic lethality was observed for 70% of the embryos , with the remaining animals dying as larvae ( Figure 1F–G; Text S1 ) ., We verified that the htz-1 dsRNA injections did not cross-react with H2A mRNA by showing that expression of a GFP-tagged version of H2A was not affected ( Figure 2A–C ) ., We interpret the progression of phenotypes resulting from either RNAi treatment or genetic mutation to indicate that HTZ-1 is required for both embryogenesis and for post-embryonic development ., We propose that the occasional escape from lethality occurs due to perdurance of maternal HTZ-1 protein or RNA for as long as two generations , or in the case of RNAi , a failure to completely eliminate HTZ-1 protein or message in the offspring of injected mothers ( Discussion ) ., HTZ-1 RNA is abundant in the form of a maternal contribution , and remains abundant throughout the majority of embryogenesis , suggesting that the function of HTZ-1 in development is widespread 31 ., To investigate the distribution of HTZ-1 protein , we generated polyclonal antisera specific to a unique peptide sequence in the C-terminal region of HTZ-1 ( Methods ) ., The antibody recognized a single band of 15 kD on western blots of C . elegans protein extract , corresponding to the predicted molecular weight of HTZ-1 ( Figure S1 ) ., Using these antibodies , we stained whole embryos and adults and found that HTZ-1 protein is present in all cell types throughout all stages of development ., HTZ-1 protein levels are low in early embryos ( 1–12 cell ) , but increase as development progresses ( Figure 2D–F ) ., HTZ-1 protein becomes detectably incorporated into chromosomes by the four-cell stage , coincident with the onset of zygotic transcription ., This occurs in both wild-type and M+Z− embryos , demonstrating that zygotic transcription of htz-1 itself is not required for incorporation of HTZ-1 protein into chromatin ., In wild-type adults , HTZ-1 protein is observed in both somatic and germline precursor cells ( data not shown ) ., No HTZ-1 protein was observed by immunofluorescence in M+Z− adult gonads or their M−Z− embryos ( Figures G–O ) ., In addition , no protein staining was observed in the offspring of animals injected with HTZ-1 RNAi ( Figure 2P–R ) ., The low levels of HTZ-1 protein in young embryos , despite abundant htz-1 mRNA , suggests that much of the maternal contribution is RNA-based , with HTZ-1 protein levels controlled post-transcriptionally ( Figure 2D–F ) ., Another case in which HTZ-1 protein levels do not depend on zygotic transcription can be inferred from the presence of HTZ-1 protein in the germline precursors ( P lineage ) ., In these cells , HTZ-1 protein is present in chromatin at levels comparable to the surrounding somatic blastomeres , despite the repression of zygotic mRNA production in the P lineage ( Figure 2D–F ) 32 ., HTZ-1 protein is also observed in the chromatin of the primordial germ cells Z2 and Z3 ( data not shown ) , which undergoes a dramatic erasure of histone H3 modifications during development 33 , 34 ., To determine the genomic locations at which HTZ-1 functions , we performed ChIP-chip of HTZ-1 from extracts of wildtype N2 C . elegans embryos ( Methods ) ., For detection of ChIP-enriched loci , we used DNA microarrays consisting of 50-bp oligonucleotide probes that tile across the entire genome with 86-bp start-to-start spacing ( Methods ) ., Peaks of HTZ-1 binding were identified using ChIPOTle 35 ., Throughout the genome , 5163 sites of HTZ-1 incorporation were found , with 85% of the peaks occurring within intergenic regions ., Intergenic regions are defined as those that occur outside the boundaries defined by the translation start and stop sites of annotated transcripts or predicted genes ., Under this definition , intergenic regions comprise 58% of the bases in the genome ., Of the peaks within an intergenic region , 71% were within the 2-kb upstream of an annotated translation start site , 25% were within 2-kb of the translation stop , and only 4% were greater than 2-kb upstream of a translation start site ., Among the 15% of peaks found to occur within an annotated transcription unit , most occurred near the 5′ end ( median +545 bp downstream of the annotated translation start site ) ., Therefore , like yeast Htz1 , C . elegans HTZ-1 is preferentially incorporated into intergenic regions , specifically at promoters ( Figure 3A ) ., We next investigated whether HTZ-1 was incorporated specifically at sites of transcriptional initiation ., The majority of transcription initiation sites are not well-annotated in C . elegans , due in part to the prevalence of trans-splicing 36 ., Therefore as a proxy for transcription initiation sites , we plotted HTZ-1 binding relative to annotated translation start codons ., On average , the peak of HTZ-1 incorporation occurs just upstream of the translation start codon ( Figure 3B ) , which we interpreted to indicate incorporation at or near sites of transcription initiation ., To further test whether the observed signal represents sites of transcription initiation , we took advantage of a unique feature of the C . elegans genome ., Approximately 15% of C . elegans genes are predicted to reside in operons that are transcribed as a large polycistronic pre-mRNA , which is then trans-spliced into mRNAs for the individual genes 37 ., We plotted HTZ-1 incorporation relative to the first gene in operons , where transcription is expected to initiate , and also plotted incorporation relative to internal genes , where transcription is not expected to initiate ., Indeed , HTZ-1 incorporation is generally observed upstream of the first gene in an operon , and does not generally occur upstream of internal genes ( Figure 3C ) , indicating that C . elegans HTZ-1 is incorporated primarily at or near sites of transcription initiation ., We also observed some important exceptions to this general rule , which are discussed below ., Currently most C . elegans operons are identified primarily by two criteria: the appearance of two or more genes in close proximity that are transcribed on the same strand , and the isolation of a downstream RNA transcript with an SL2 trans-spliced leader 38 , 39 ., In this way , a total of 1118 putative operons have been identified ( genome release ws170 ) ., However , these criteria are imperfect , and do not provide information about genes that may be regulated both as part of an operon and by their own independent promoter ., Independent transcription events within operons have been difficult to detect because the 5′ ends of mRNAs produced by either trans-splicing of a poly-cistronic mRNA or an independent transcription event are not readily distinguishable ., To identify genes that are likely to be regulated both as part of an operon and individually , we examined incorporation of HTZ-1 at internally encoded genes of annotated operons ., Overall , 75% of operons contained at least one HTZ-1 peak ., A gene within an operon was more than twice as likely as a non-operon gene to have an HTZ-1 peak at its promoter ( Figure 4A–B ) ., Of operons containing at least one site of HTZ-1 incorporation , 85% contained a peak upstream of the first gene , as one might expect ., However , 49% of operons with HTZ-1 incorporation at the first gene also exhibited an internal peak of HTZ-1 incorporation ., This strongly suggests internal transcription start sites at 416 ( 37% ) of the currently annotated operons ( Figure 4B , Table S2 ) ., Because some operons contain multiple internal HTZ-1 peaks , this represents a total of 455 putative independently regulated genes within annotated operons ., This is likely to be an underestimate , since the HTZ-1 localization data is derived only from embryonic extracts , meaning that genes and operons regulated specifically in adults or germ cells are not represented ., The unexpectedly high number of individually regulated genes within operons may to some extent reflect a mis-annotation of operons based on traditional criteria ., To show that internal HTZ-1 incorporation can occur at verified operons , we examined CEOP1456 , one of the first characterized operons , supported by cistronic RNA evidence 37 , 40 ., In this well-characterized operon , both HTZ-1 and RNA Polymerase II occupy the chromatin immediately upstream of the internal kin-10 gene , strongly suggesting independent regulation ( Figure 4C ) ., Recently , differential regulation of genes driven by internal operon promoters was demonstrated using a GFP reporter assay 41 ., We find that one-third of these internal promoters are occupied by HTZ-1 in embryos ( Table S2 ) ., A time-course of the early embryonic transcription 31 provides evidence that genes within operons that contain multiple sites of HTZ-1 incorporation exhibit differential expression ( Figure S2 ) ., In contrast to yeast , in which Htz1 is incorporated into nearly every promoter 42 , our ChIP-chip data indicate that HTZ-1 is incorporated into the promoters of only 23% of C . elegans genes ( Methods ) ., To determine what might be held in common among the particular subset of genes that were occupied by HTZ-1 , peaks were annotated to gene promoters , assigned Gene Ontology ( GO ) terms according to the nearest downstream gene , and evaluated with GO::TermFinder 43 ., To avoid ambiguous gene assignments , only peaks annotated to unidirectional promoters or within coding regions were used in the input set ., We found that GO terms associated with metazoan development and positive regulation of growth were strongly over-represented among HTZ-1 bound genes , while no overrepresented GO term was associated with the non-HTZ-1 bound genes ( Table 1 , Table S1 ) ., This finding suggests that HTZ-1 functions preferentially at the promoters of genes essential for growth and development ., We next sought to examine the relationship between HTZ-1 occupancy at promoters and transcriptional activity during embryogenesis ., We found that , in general , transcript levels 31 were positively correlated with HTZ-1 promoter occupancy ( Figure 5A; Spearman rank-order correlation\u200a=\u200a0 . 35 ) ., A positive correlation was also observed between RNA levels reported by a completely independent study 44 and HTZ-1 occupancy ( Figure S3 ) ., Despite the positive overall correlation between occupancy and transcript levels , the relationship becomes negative at promoters of genes with very high transcript abundance ( Figure 5A ) ., This observation is consistent with a general loss of nucleosomes upstream of highly transcribed genes 45 , 46 ., We sought to establish a more direct link between HTZ-1 occupancy and transcription , so we determined the genome-wide occupancy of RNA polymerase II by ChIP-chip using an antibody specific to the C-terminal domain heptapeptide ( 8WG16 , Methods ) ., At gene promoters , HTZ-1 occupancy was strongly correlated with RNA Polymerase II occupancy ( Figure 5B ) ., In fact , the correlation was stronger than that observed between HTZ-1 occupancy and transcript levels ( Spearman rank-order correlation\u200a=\u200a0 . 57 ) ., Consistent with the correlation with transcript levels , at the promoters most highly occupied by RNA Polymerase II , the correlation with HTZ-1 occupancy was negative ., Again , this observation is likely due to general nucleosome loss at the promoters of highly transcribed genes , for example those that encode the histone and ribosomal proteins 45 , 46 ., Temporal regulation gene expression during embryogenesis may also affect this correlation and is considered in the Discussion ., To further illustrate the relationship between HTZ-1 localization and polymerase occupancy , the 4650 genes with HTZ-1 incorporated into their promoters were aligned according to their translation start site , and average RNA polymerase II occupancy relative to the start site was plotted ( Figure 5C ) ., HTZ-1-occupied promoters were on average occupied by RNA Polymerase II , whereas genes lacking HTZ-1 were not ( Figure 5D ) ., At promoters occupied by HTZ-1 , the average peak of HTZ-1 occupancy was at negative 12 bp relative to the translation start , while the average peak of RNA Polymerase II occupancy was slightly upstream at negative 98 bp ( Discussion ) ., An important consideration in interpreting these relationships is that our experiments were performed using extract derived from a mixed population of embryos composed of many cell types ., Therefore , our results are a projection of HTZ-1 occupancy throughout embryogenesis and represent a temporal and spatial average of the relationship between HTZ-1 , RNA Polymerase II , and transcription ( Discussion and Text S1 ) ., To examine if HTZ-1 occupied promoters direct a stereotypic pattern of gene expression , we compared HTZ-1 occupancy , RNA Polymerase II occupancy , and transcription at each gene using a published time-course of transcript abundance during embryonic development 31 ., Promoters occupied by HTZ-1 were clustered according to our RNA Polymerase II promoter occupancy data and the change in transcript abundance relative to the onset of zygotic transcription ., To avoid ambiguity , transcripts that were highly maternally loaded ( >100 parts per million ( ppm ) ) were removed from analysis ., Consistent with the aggregate analysis , RNA Polymerase II is abundant at most HTZ-1 occupied genes ( Figure 6A ) , while promoters at which HTZ-1 is not incorporated generally lack RNA Polymerase II ( Figure 6B ) ., However , a large proportion of genes downstream of promoters occupied by both HTZ-1 and RNA polymerase II produce low transcript levels ( Figure 6A ) , and conversely some genes produce high transcript levels despite low levels of HTZ-1 and RNA polymerase II at their promoters ( Figure 6B; Discussion ) ., Therefore , while HTZ-1 is strongly linked to RNA Polymerase II occupancy in aggregate , HTZ-1 bound promoters do not specify a stereotypic pattern of transcriptional regulation during development , suggesting that RNA polymerase occupancy and transcript levels are decoupled at some promoters ., The sex chromosomes are often sites of specialized chromatin , harboring unique histone variants and chromatin modifications ., To determine whether HTZ-1 was differentially localized to X , we co-stained embryos with anti-HTZ-1 in combination with either anti-DPY-27 , which marks the X chromosomes in embryos of greater than about 30 cells ( Figure 7A–D ) , or anti-MES-4 , which marks the autosomes but not X chromosomes in early embryos ( Figure 7E–H ) ., In embryos that had initiated somatic dosage compensation , HTZ-1 incorporation was noticeably reduced on the X chromosomes , which was marked by DPY-27 staining ( Figure 7A–D ) ., However , co-staining with MES-4 revealed HTZ-1 under-representation on X even before the onset of somatic dosage compensation ( Figure 7E–H ) ., These results indicate that in both early-stage embryos before the onset of dosage compensation and late-stage C . elegans embryos after dosage compensation is established , there is significantly less HTZ-1 associated with the X than with autosomes ., We next aimed to ensure that reduction of HTZ-1 we observed on the X chromosome by immunofluorescence was not due to epitope exclusion ., This concern was prompted by reports that mammalian H2A . Z on the inactive X chromosome is ubiquitylated , and that this modification can interfere with recognition by antibodies raised against a C-terminal peptide epitope 19 ., Our antisera were also raised against a C-terminal peptide ., To address this concern , we co-stained embryos expressing a HTZ-1:YFP transgene with anti-DPY-27 and anti-YFP antibodies ., We observed a similarly reduced YFP signal coincident with regions of DPY-27 signal ., This serves as independent evidence that within in the same nucleus , X chromatin has less HTZ-1 incorporation than autosomes ( Figure S4 ) ., Two explanations for the under-incorporation of HTZ-1 on X immediately come to mind ., One is that less HTZ-1 is incorporated on X as part of the C . elegans dosage compensation mechanism ., A second explanation , which we favor for the reasons presented below , is that genes important for development , whose promoters tend to be occupied by HTZ-1 , are under-represented on the X chromosome 47–50 ., To distinguish these possibilities , we examined at high resolution the sites of HTZ-1 incorporation on the X chromosomes relative to the autosomes ( Figure 7I ) ., One possible variation of the “dosage compensation” hypothesis predicts that sites of HTZ-1 incorporation are excluded or diminished on X as a consequence of the transcriptional repression imposed by the DCC ., In this case , one would expect HTZ-1 occupancy on X to be excluded from sites occupied by the dosage compensation machinery 51 ., Contrary to this prediction , we found strong co-localization of HTZ-1 incorporation and DCC binding , such that over 62% of HTZ-1 peaks are coincident with a DPY-27 peak ( Figure 7J–K , Figure S7 ) ., The highly concordant binding pattern of HTZ-1 and the DCC on X would appear to rule out a function for HTZ-1 as a direct negative regulator of DCC binding to autosomes ( Discussion ) ., We then considered the possibility that HTZ-1 incorporation is in fact a requirement for the loading of the DCC onto X . However , there are far more sites of DCC localization on X than HTZ-1 incorporation , meaning that most DCC-bound loci are not sites of HTZ-1 localization ., For example , while both HTZ-1 and DPY-27 are incorporated at the X-linked dpy-23 promoter , HTZ-1 is not incorporated at the well-characterized DCC recruitment site rex-1 , which occurs just 5 kb downstream of dpy-23 ( Figure 7L ) 51 , 52 ., We also examined in more detail apl-1 and lin-15 , two of the few genes known with some certainty to be dosage compensated 53 , 54 ., Although the DCC and RNA Polymerase II are present at both loci , HTZ-1 is present at lin-15 , but not at apl-1 ( Figure S5 ) , reinforcing the interpretation that HTZ-1 is not required for dosage compensation ., Conversely , the under-representation of HTZ-1 on X is not dependent on the dosage compensation process , because it is evident in the germline and before the onset of somatic dosage compensation ( Figure 7H ) ., The alternative “developmental gene” hypothesis for the under-incorporation of HTZ-1 on X is based on the observation that only about half as many essential genes occur on X as would be expected to occur on an autosome of the same size ( 201 vs . 562 expected , wormbase release ws170 ) 47–50 ., This hypothesis predicts that there would be fewer sites of HTZ-1 incorporation on X , but that those that do occur on X behave like those on autosomes ., The X harbored 495 HTZ-1 peaks , about half the number expected from a hypothetical autosome with the size and gene density of X ( p-value\u200a=\u200a2 . 05×10−43 and 8 . 09×10−93 respectively , Figure 7I ) ., There was no significant difference between the median height and width of HTZ-1 peaks on X ( z-score\u200a=\u200a2 . 28 and 774 bp , respectively ) as compared to the median height and width of HTZ-1 peaks on autosomes ( z-score\u200a=\u200a2 . 20 and 860 bp , respectively ) ( Figure S8 ) ., This indicates that while HTZ-1 incorporation occurs at fewer loci on X , where it does occur the degree of incorporation is the same as the autosomes ., The most parsimonious explanation for the under-representation of HTZ-1 on the X is that the types of genes that require HTZ-1 for proper regulation are themselves under-represented on the X chromosome ., Nonetheless , HTZ-1 is likely to have an indirect function in the dosage compensation process by affecting the regulation of genes required for dosage compensation ., Strong HTZ-1 incorporation is observed at the promoters of sdc-1 , sdc-2 , sdc-3 , dpy-27 , mix-1 , and dpy-30 , all of which are required for dosage compensation ., Although any number of complex scenarios involving a direct relationship between HTZ-1 and the canonical dosage compensation process remain possible , we interpret the under-representation of sites of HTZ-1 localization on X to be a simple consequence of the under-representation of germline and developmentally important genes on the X chromosome ( Discussion ) ., The C . elegans genome has been shaped by the developmental programs it must coordinately execute ., The general requirement of H2A . Z for development in metazoans suggests a function for H2A . Z in establishing or maintaining a specialized chromatin state at developmentally regulated promoters 27–30 , 55 ., In this study , we have established that HTZ-1 is incorporated upstream of genes critical for development , and that maternally provided HTZ-1 is sufficient for C . elegans embryogenesis ., We infer by the progressively deteriorating phenotype suffered by offspring lacking HTZ-1 that HTZ-1 is required for both embryogenesis and post-embryonic development ., The function of HTZ-1 in pharyngeal organogenesis may provide a model for the mechanism by which HTZ-1 is generally required for C . elegans development ., The development of the pharynx relies on precise temporal regulation of transcription activation , mediated in part by PHA-4 , a FoxA transcription factor 56 , 57 ., HTZ-1 depletion enhances defects in pharyngeal organogenesis associated with loss of PHA-4 , and activation of PHA-4-dependent promoters is delayed in the absence of HTZ-1 23 ., This is reminiscent of the delay of yeast GAL gene activation in the absence of Htz1 10 , and indicates a conserved role for H2A . Z in facilitating timely gene expression ., Previous genome-wide studies in yeast and other organisms have reached differing conclusions regarding the relationship between H2A . Z , RNA Polymerase II , and transcription 11 , 12 , 15 , 16 , 20–22 , 42 , 58 , 59 ., Functional divergence between yeast Htz1 and metazoan homologs are a possible source of the discrepancy ., Consistent with this , C . elegans HTZ-1 is only 61% identical to yeast Htz1 , but 77% identical to Drosophila H2Avd , and 83% identical to mouse or human H2A . Z ( Figure S6 ) ., In C . elegans , we found that HTZ-1 is incorporated specifically at promoters , where its occupancy is strongly correlated with RNA polymerase II occupancy , and to a lesser degree with transcript levels ( see Text S1 ) ., This suggests that RNA polymerase II is present at some HTZ-1 occupied promoters without being linked to a corresponding increase in transcripts ., One possible explanation is pausing of RNA polymerase II near initiation sites ., This phenomenon is common in human and Drosophila cells 15 , 60–62 but has not yet been established to occur in C . elegans ., The 8WG16 RNA polymerase II antibody we used is probably not the appropriate choice for making conclusions about RNA Pol II pausing , because the antibody recognizes primarily the unphosphorylated RNA Pol II CTD , but it is known to have some cross-reactivity with both CTD-Ser5P and CTD-Ser2P ., RNA Polymerase II pausing would be more appropriately examined with an independent , non-C-terminal domain RNA Pol II antibody or a CTD-Ser5P specific antibody ., Nonetheless , using the 8WG16 antibody , we observed a small number of genes ( about 300 , or ∼1 . 5% ) with promoter-restricted RNA Polymerase II ., A recent genome-wide study of the Drosophila H2A . Z homolog at mononucleosome resolution reported that an H2A . Z-containing nucleosome was often positioned just downstream of a paused RNA polymerase II 15 ., Although we did not observe any relationship , positive or negative , between HTZ-1 occupancy and this putative paused state , peak HTZ-1 occupancy occurs about 80 bp downstream of peak RNA Pol II occupancy at promoters ( Figure 5C ) ., Thus , the putative poised state may in some cases be facilitated by HTZ-1 , and could contribute to the efficient and timely activation of developmental promoters ., Indeed , our data does not formally exclude the possibility that H2A . Z functions to dampen transcription 63 ., In Drosophila and mammalian cells , H2A . Z plays a role in gene silencing by participating in the assembly of heterochromatin 64 , 65 ., While

Introduction, Results, Discussion, Materials and Methods

In all eukaryotes , histone variants are incorporated into a subset of nucleosomes to create functionally specialized regions of chromatin ., One such variant , H2A . Z , replaces histone H2A and is required for development and viability in all animals tested to date ., However , the function of H2A . Z in development remains unclear ., Here , we use ChIP-chip , genetic mutation , RNAi , and immunofluorescence microscopy to interrogate the function of H2A . Z ( HTZ-1 ) during embryogenesis in Caenorhabditis elegans , a key model of metazoan development ., We find that HTZ-1 is expressed in every cell of the developing embryo and is essential for normal development ., The sites of HTZ-1 incorporation during embryogenesis reveal a genome wrought by developmental processes ., HTZ-1 is incorporated upstream of 23% of C . elegans genes ., While these genes tend to be required for development and occupied by RNA polymerase II , HTZ-1 incorporation does not specify a stereotypic transcription program ., The data also provide evidence for unexpectedly widespread independent regulation of genes within operons during development; in 37% of operons , HTZ-1 is incorporated upstream of internally encoded genes ., Fewer sites of HTZ-1 incorporation occur on the X chromosome relative to autosomes , which our data suggest is due to a paucity of developmentally important genes on X , rather than a direct function for HTZ-1 in dosage compensation ., Our experiments indicate that HTZ-1 functions in establishing or maintaining an essential chromatin state at promoters regulated dynamically during C . elegans embryogenesis .

To fit within a cells nucleus , DNA is wrapped around protein spools composed of the histones H3 , H4 , H2A , and H2B ., One spool and the DNA wrapped around it are called a nucleosome , and all of the packaged DNA in a cells nucleus is collectively called “chromatin . ”, Chromatin is important because it modulates access to information encoded in the underlying DNA ., Spools with specialized functions can be created by replacing a typical histone component with a variant version of the histone protein ., Here , we examine the distribution and function of the C . elegans histone H2A variant H2A . Z ( called HTZ-1 ) during development ., We demonstrate that HTZ-1 is required for proper development , and that embryos are dependent on a contribution of HTZ-1 from their mothers for survival ., We mapped the location of HTZ-1 incorporation genome-wide and found that HTZ-1 binds upstream of 23% of genes , which tend to be genes that are essential for development and occupied by RNA polymerase ., Fewer sites of HTZ-1 incorporation were found on the X chromosome , probably due to an under-representation of essential genes on X rather than a direct role for HTZ-1 in X-chromosome dosage compensation ., Our study reveals how the genome is remodeled by HTZ-1 to allow the proper regulation of genes critical for development .

genetics and genomics/genomics, molecular biology/histone modification, developmental biology, molecular biology/transcription initiation and activation, genetics and genomics/gene expression, genetics and genomics/epigenetics, molecular biology, molecular biology/chromatin structure

journal.pcbi.1004071

Accurate Computation of Survival Statistics in Genome-Wide Studies

Next-generation DNA sequencing technologies are now enabling the measurement of exomes , genomes , and mRNA expression in many samples ., The next challenge is to interpret these large quantities of DNA and RNA sequence data ., In many human and cancer genomics studies , a major goal is to find associations between an observed phenotype and a particular variable ( e . g . , a single nucleotide polymorphism ( SNP ) , somatic mutation , or gene expression ) from genome-wide measurements of many such variables ., For example , many cancer sequencing studies aim to find somatic mutations that distinguish patients with fast-growing tumors that require aggressive treatment from patients with better prognosis ., Similarly , many human disease studies aim to find genetic alleles that distinguish patients who respond to particular treatments , i . e . live longer ., In both of these examples one tests the association between a DNA sequence variant and the survival time , or length of time that patients live following diagnosis or treatment ., The most widely approach to determine the statistical significance of an observed difference in survival time between two groups is the log-rank test 1 , 2 ., An important feature of this test , and related tests in survival analysis 3 , is their handling of censored data: in clinical studies , patients may leave the study prematurely or the study may end before the deaths of all patients ., Thus , a lower bound on the survival time of these patients is known ., Importantly , many studies are designed to test survival differences between two pre-selected populations that differ by one characteristic; e . g . a clinical trial of the effectiveness of a drug ., These populations are selected to be approximately equal in size with a suitable number of patients to achieve appropriate statistical power ( Fig 1A ) ., In this setting , the null distribution of the ( normalized ) log-rank statistic is asymptotically ( standard ) normal; i . e . follows the ( standard ) normal distribution in the limit of infinite sample size ., Thus , nearly every available implementation ( e . g . , the LIFETEST procedure in SAS , and the survdiff R function , and coin and exactRankTests packages in R and SPlus ) of the log-rank test computes p-values from the normal distribution , an approximation that is accurate asymptotically ( see S1 Text ) ., The design of a genomics study is typically very different from the traditional clinical trials setting ., In a genomics study , high-throughput measurement of many genomics features ( e . g . whole-genome sequence or gene expression ) in a cohort of patients is performed , and the goal is to discover those features that distinguish survival time ., Thus , the measured individuals are repeatedly partitioned into two populations determined by a genomic variable ( e . g . a SNP ) and the log-rank test , or related survival test , is performed ( Fig 1B ) ., Depending on the variable the sizes of the two populations may be very different: e . g . most somatic mutations identified in cancer sequencing studies , including those in driver genes , are present in < 20% of patients 4–9 ., Unfortunately , in the setting of unbalanced populations , the normal approximation of the log-rank statistic might give poor results ., While this fact has been noted in the statistics literature 10–12 , it is not widely known , and indeed the normal approximation to the log-rank test is routinely used to test the association of somatic mutations and survival time ( e . g . 13 , 14 and numerous other publications ) ., A second issue in genomics setting is that the repeated application of the log-rank test demands the accurate calculation of very small p-values , as the computed p-value for a single test must be corrected for the large number of tests; e . g . through a Bonferroni or other multiple-hypothesis correction ., An inaccurate approximation of p-values will result in an unacceptable number of false positives/negative associations of genomic features with survival ., These defining characteristics of genomics applications , unbalanced populations and necessity of highly-accurate p-values for multiple-hypothesis correction , indicate that standard implementations of the log-rank test are inadequate ., We propose to compute the p-value for the log-rank test using an exact distribution determined by the observed number of individuals in each population ., Perhaps the most famous use of an exact distribution is Fisher’s exact test for testing the independence of two categorical variables arranged in a 2×2 contingency table ., When the counts in the cells of the table are small , the exact test is preferred to the asymptotic approximation given by the χ2 test 15 ., Exact tests for comparing two survival distributions have received scant attention in the literature ., There are three major difficulties in developing such a test ., First , there are multiple observed features that determine the exact distribution including the number of patients in each population and the observed censoring times ., With so many combinations of parameters it is infeasible to pre-compute distribution tables for the test ., Thus , we need an efficient algorithm that computes the p-value for any given combination of observed parameters ., Second , we cannot apply a standard Monte-Carlo permutation test to this problem since we are interested in very small p-values that are expensive to accurately estimate with such an approach ., ( By the accurate estimation of a p-value using Monte-Carlo permutation test , we mean the calculation of a p-value and a confidence interval of the same order of magnitude of the p-value . ), Third , there are two possible null distributions for the log-rank test , the conditional and the permutational 2 , 16–18 ., While both of these distributions are asymptotically normal , the permutational distribution is more appropriate for genomics settings 2 , 19 , as we detail below ., Yet no efficient algorithm is known to compute the p-value of the log-rank test under the exact permutational distribution ., We introduce an efficient and mathematically sound algorithm , called ExaLT ( for Exact Log-rank Test ) , for computing the p-value under the exact permutational distribution ., ( ExaLT computes an estimate of the p-value under the exact permutational distribution; for this reason we denote the p-value obtained from ExaLT as an exact p-value , in contrast to the approximate p-value obtained from asymptotic distributions . ), The run-time of ExaLT is not function of the p-value , enabling the accurate calculation of small p-values ., For example , obtaining an accurate estimate of p ≈ 10−9 is required if one wants to test the association of 1% of the human genome ( e . g . , the exome ) with survival , and using a standard MC approach requires ( with the Clopper-Pearson confidence interval estimate ) the evaluation of ≥ 1011 samples , that for a population of 200 patients requires > 8 days; in contrast ExaLT is capable of estimating p ≈ 10−13 on 200 patients in < 2 hours ., In contrast to heuristic approaches ( see Materials and Methods ) ExaLT provides rigorous guarantees on the relation between the estimated p-value and the correct p-value; moreover , it returns a conservative estimate of the p-value , thus guaranteeing rigorous control on the number of false discoveries ., We test ExaLT on data from two published cancer studies 20 , 21 , finding substantial differences between the p-values obtained by our exact test and the approximate p-values obtained by standard tools in survival analysis ., In addition , we run ExaLT on somatic mutation and survival data from The Cancer Genome Atlas ( TCGA ) and find a number of mutations with significant association with survival time ., Some of these such as IDH1 mutations in glioblastoma are widely known; for others such as BRCA2 and NCOA3 mutations in ovarian cancer there is some evidence in the literature; while the remaining are genuinely novel ., Most of these are identified only using the exact permutational test of ExaLT ., In contrast , the genes reported as highly significant using standard implementations of the log-rank test are not supported by biological evidence; moreover , these methods report dozens-hundreds of such likely false positive associations as more significant than known genes associated with survival ., These results show that our algorithm is practical , efficient , and avoids a number of false positives , while allowing the identification of genes known to be associated with survival and the discovery of novel , potentially prognostic biomarkers ., We first assessed the accuracy of the asymptotic approximation for the log-rank test on simulated data from a cohort of 500 patients with a gene g mutated in 5% of these patients , a frequency that is not unusual for cancer genes in large-scale sequencing studies 4–7 ., We compared the survival times of the population 𝓟 ( g ) of patients with a mutation in g to the survival of the population 𝓟 ‾ ( g ) of patients with no mutation in g ., We computed p-values using R survdiff on multiple random instances ( in order to obtain a distribution for the p-value of g ) in which 𝓟 ( g ) and 𝓟 ‾ ( g ) have the same survival distribution ., S1 Fig shows that the p-values computed by the asymptotic approximation are much smaller than expected under the null hypothesis , with the smallest p-values showing the largest deviation from the expected uniform distribution ., The inaccuracy of the asymptotic log-rank test results in a large number of false discoveries: for example , considering a randomized version of a cancer mutation dataset ( S1 Table ) in which no mutation is associated with survival ( i . e . no true positives ) , the asymptotic log-rank test reports 110 false discoveries ( Bonferroni correction ) or 291 false discoveries ( False Discovery Rate ( FDR ) correction ) , with significance level α = 0 . 05 ( Fig 2 ) ., We found that for the number of patients of interest to current genomic studies , the inaccuracy of the asymptotic log-rank test results mostly from the imbalance in the sizes of the two populations , rather than the total number of patients or the number of patients in the smaller population ( see S1a Fig , S1b Fig , S1c Fig , S1d Fig , and S1 Text ) ., As noted above , there are two exact distributions for the log-rank test in the literature: the permutational distribution 2 and the conditional distribution 16 ., We developed an algorithm , Exact Log-rank Test ( ExaLT ) , to compute the p-value of the log-rank statistic under the exact permutational distribution ., On simulations of cancer data , we found that the p-values from the permutational exact test are significantly closer to the empirical p-values than the p-values obtained from the conditional exact test ( S2 Fig and S1 Text ) ., Thus , we derived a fully polynomial time approximation scheme ( FPTAS ) for computing the p-value under the permutational distribution ., In contrast to heuristic methods that do not provide any rigorous guarantee on the quality of the approximation of the p-value , our algorithm provides an approximation that is guaranteed to be within a user defined distance from the p-value , for any given sample size , in polynomial time ., Furthermore , the output of our scheme is always a conservative or valid p-value estimate ., The C++ implementation of ExaLT ( that can be called from R ) is available at https://github . com/fvandin/ExaLT ., To demonstrate the applicability of ExaLT we compared the p-values from the exact distribution to p-values from the asymptotic approximation reported in two recently published cancer genomics studies 20 , 21 ., Huang et al . 20 divides patients into groups defined by the number of risk alleles of five single nucleotide polymorphisms ( SNPs ) , and compares the survival distribution of the resulting populations ., In one comparison , the survival distribution of 2 patients ( 13% of total ) with at most 2 risk alleles was compared with the survival distribution of 14 patients with more than 2 risk alleles , and a p-value of 0 . 012 is reported ., Thus , this association is significant at the traditional significance level of α = 0 . 05 ., However , ExaLT computes an exact p-value of 0 . 17 , raising doubts about this association ., In another comparison patients at a different disease stage were considered , and the division of the patients into groups as above resulted in comparing the survival distribution of 8 patients ( 17% of total ) with the survival distribution of 40 patients , and a p-value of 6×10−6 is reported ., In contrast , ExaLT computes an exact p-value of 2×10−3 , a reduction of three orders of magnitude in the significance level ., Additional comparisons are shown in the S1 Text ., Therefore in these cases the asymptotic approximation underestimates the exact permutational p-values resulting in associations deemed more significant than what is supported by the data ., In 21 , the survival distribution of 14 glioblastoma patients ( 11% of total ) with somatic IDH1 or IDH2 mutations was compared to the survival distribution of 115 patients with wild-type IDH1 and IDH2 ., The reported p-value from the asymptotic approximation is 2×10−3 , while the exact permutational p-value is 5×10−4 , indicating a stronger association between somatic mutations in IDH1 or IDH2 and ( longer ) survival than reported ., Notably , this same association has been reported in three other glioblastoma studies 22–24 ., We analyzed somatic mutation and survival data from studies of six different cancer types ( S1 Table ) from The Cancer Genome Atlas ( TCGA ) ., For the range of parameters of these datasets our simulations show that the asymptotic approximation is not accurate for genes with mutation frequency ≤ 10%; we then did not considered genes with mutation frequency > 10% ., We also discarded genes mutated with frequency < 1% ., For each mutated gene , we first obtained an estimate p ˜ of the p-value using an MC approach , and if p ˜ ≤ 0 ., 01 we used ExaLT to compute a controlled approximation of the p-value ., We compared the p-value obtained in this way with the one obtained by using the asymptotic approximation as computed by the R package survdiff ( S2 Table , S3 Table , S4 Table , S5 Table ) ., Fig 3 shows the exact p-values and the R survdiff p-values for the glioblastoma multiforme ( GBM ) dataset and ovarian serous adenocarcinoma ( OV ) dataset ., The p-values for the other datasets are shown in S3 Fig . For most datasets the asymptotic p-values obtained from R survdiff are very different from the ones obtained with the exact p-values obtained by ExaLT , and the ranking of the genes by p-value is very different as well ( see S1 Text ) ., For example , in GBM none of the top 26 genes reported by R are in the list of the top 26 genes reported by the exact permutational test ., Since genomics studies are typically focused on the discovery of novel hypotheses that will be further validated , this striking difference in the ranking of genes by the two algorithms is important: a poor ranking of genes by their association with survival will lead to many false discoveries undergoing additional experimental validation ., While some of genes ranked in the top 10 by ExaLT are known to have mutations associated with survival ( e . g . , IDH1 in GBM and BRCA2 in OV ) , none of the top 10 genes reported by R survdiff ( S5 Table ) have mutations known to be associated with survival ., R survdiff ranks dozens-hundreds of presumably false positives associations as more significant than these known genes ., Moreover , R survdiff reports extremely strong association with survival for many of these higher ranked , but likely false positive , genes; e . g . , in uterine corpus endometrial carcinoma ( UCEC ) , 13 genes have p < 10−8 and an additional 19 genes have p < 10−5 , but none of these have a known association with survival ., The top 10 genes reported by ExaLT contain several novel associations that are supported by the literature and are not reported using R survdiff ., In GBM , ExaLT identifies IDH1 ( p ≤ 3×10−4 ) , VARS2 ( p ≤ 5×10−3 ) and GALR1 ( p ≤ 6×10−3 ) , among others ., As noted above , the association between mutations in IDH1 and survival has been previously reported in GBM 21–24 ., A germline variant in VARS2 has been reported to be a prognostic marker , associated with survival , in early breast cancer patients 25 ., The expression of GALR1 has been reported to be associated with survival in colorectal cancer 26 , and its inactivation by methylation has been associated with survival in head and neck cancer 27 , 28 ., In OV , ExaLT identifies BRCA2 ( p ≤ 4×10−3 ) and NCOA3 ( p ≤ 10−3 ) , and others ., Germline and somatic mutations in BRCA2 ( and BRCA1 ) have been associated with survival in two ovarian cancer studies 4 , 29 ., A polymorphism in NCOA3 has been associated with breast cancer 30 , and its amplification has been associated with survival in ER-positive tumors 31 ., Thus , the exact test implemented by ExaLT appears to have higher sensitivity and specificity in detecting mutations associated with survival on the sizes of cohorts analyzed in TCGA ., Finally , we note that the exact conditional test obtains results similar to R survdiff , confirming that the the exact permutational test implemented by ExaLT is a more appropriate exact test for genomics studies ., ( See S1 Text . ), We have also assessed the difference between the result obtained using ExaLT and the results obtained using the asymptotic permutational test ( S2 Table , S3 Table , S4 Table , S5 Table ) ., In this case the difference in the ranking of genes is reduced but still present; in particular , in COADREAD 4 of the top 10 genes identified by ExaLT are not among the top 10 genes found using the asymptotic permutational test ., Moreover , there are some genes for which there is a large difference in the p-value computed by ExaLT and the p-value from the asymptotic permutational test; for example , in UCEC data CTGF is ranked first by both ExaLT and the asymptotic permutational test , but it has p = 9 . 6×10−5 by ExaLT and p = 8 . 3×10−10 by the asymptotic permutational test ., In this work we focus on the problem of performing survival analysis in a genomics setting , where the populations being compared are not defined in advance , but rather are determined by a genomic measurement ., The two distinguishing features of such studies are that the populations are typically unbalanced and that many survival tests are performed for different measurements , requiring highly accurate p-values for multiple hypothesis testing corrections ., We show empirically that the asymptotic approximations used in available implementations of the log-rank test produce anti-conservative estimates of the true p-values when applied to unbalanced populations , resulting in a large number of false discoveries ., This is not purely a phenomenon of small population size: the approximation remains inaccurate even for a large number of samples ( e . g . , 100 ) in the small population ., This inaccuracy makes asymptotic approximations unsuitable for cancer genomic studies , where the vast majority of the genes are mutated in a small proportion of all samples 4–6 and also for genome-wide association studies ( GWAS ) where rare variants may be responsible for a difference in drug response or other phenotype , even if it is possible that for extremely large genomic studies ( e . g . , with 100000 patients ) the asymptotic approximations would provide results accurate enough even for imbalanced populations ( e . g . , when the small population is 1% ) ., The problem with the log-rank test for unbalanced populations has previously been reported 10 , 11 , but the implications for genomics studies have not received attention ., Note that the issue of unbalanced populations is further exacerbated by any further subdivision of the data: e . g . by considering mutations in specific locations or protein domains; by considering the impact of mutations on a specific therapeutic regimen; by testing the association of mutations with survival in a particular subtype of cancer; by grouping into more than two populations; or by correcting for additional covariates such as age , stage , grade , etc ., All of these situations occur in genomics studies ., We considered the two versions of the log-rank test , the conditional 16 and the permutational 2 , and we found that the exact permutational distribution is more accurate in genomics settings ., We introduce ExaLT , the first efficient algorithm to compute highly accurate p-values for the exact permutational distribution ., We implemented and tested our algorithm on data from two published cancer studies , showing that the exact permutational p-values are significantly different from the p-values obtained using the asymptotic approximations ., We also ran ExaLT on somatic mutation and survival data from six cancer types from The Cancer Genome Atlas ( TCGA ) , showing that our algorithm is practical , efficient , and allows the identification of genes known to be associated with survival in these cancer types as well as novel associations ., We note that ExaLT can be employed as part of permutation tests that require the computation of p-values for a large number of genomic features measured on the same set of patients 12 ., While the current implementation of ExaLT handles ties in survival times by breaking them arbitrarily , its extension to different tie breaking strategies and their impact is an important future direction ., The method we propose can be generalized to assess the difference in survival between more than two groups , by considering the exact permutational distribution for the appropriate test statistic ., For this reason , our method can be adapted to test the difference in survival between groups of patients that have homozygous or heterozygous mutations , or to test whether the presence of a group of genomic features has a different effect on survival compared to the presence of the single genomic features ., For the same reason , our method can incorporate categorical covariates , while it is unclear how methods based on the log-rank test , as ours , can incorporate continuous covariates or how they can be used to assess specific ( e . g . , additive ) models of interactions between genomic features and survival ., While our focus here was the log-rank test , our results are relevant to more general survival statistics ., First , in some survival analysis applications , samples are given different weights; our algorithm can be easily adapted to a number of these different weighting schemes ., Second , an alternative approach in survival analysis is to use the Cox Proportional-Hazards model 3 ., While in the Cox regression model one can easily adjust for categorical and continuous covariates , it is not clear how to incorporate continuos potential confounders in the log-rank test that we consider ., While this constitutes a limitation of our method , the Cox regression models is often used to compare two populations even when no adjustment for confounders is performed ., In this case , the significance of the resulting coefficients in the regression is typically done using a test that is equivalent to the log-rank test , and thus our results are relevant for this approach as well ., See S1 Text and S4 Fig . The challenges of extending multivariate regression models to the multiple-hypothesis setting of genome-wide measurements is not straightforward ., Direct application of such a multivariate Cox regression will often not give reasonable results as: there are a limited number of samples and a large number of genomic variants; and many variants are rare and not associated with survival ., Witten and Tibshirani ( 2010 ) 32 recently noted these difficulties for gene expression data stating that: “While there are a great number of methods in the literature for identification of significant genes in a microarray experiment with a two-class outcome … the topic of identification of significant genes with a survival outcome is still relatively unexplored . ”, We propose that exact tests such as the one provided here will be useful building blocks for more advanced models of survival analysis in the genomics setting ., We focus here on the two-sample log-rank test of comparing the survival distribution of two groups , P0 and P1 ., Let t1 < t2 < … < tk be the times of observed , uncensored events; in case of ties , we assume that they are broken arbitrarily ., Let Rj be the number of patients at risk at time tj ,, i . e ., the number of patients that survived ( and were not censored ) up to this time , and let Rj , 1 be the number of P1 patients at risk at that time ., Let Oj be the number of observed uncensored events in the interval ( tj−1 , tj , and let Oj , 1 be the number of these events in group P1 ., If the survival distributions of P0 and P1 are the same , then the expected value E O j , 1 = O j R j , 1 R, j . The log-rank statistic 1 , 2 measures the sum of the deviations of Oj , 1 from the expectation , V = ∑ j = 1 k ( O j , 1 − O j R j , 1 R j ) ., ( In some clinical applications one is more interested in either earlier or later events . In that case the statistic is a weighted sum of the deviations . Our results easily translate to the weighted version of the test . ), Under the null hypothesis of no difference in the survival distributions of the two groups , EV = 0 , and Pr ( ∣V∣ ≥ ∣v∣ ) is the p-value of an observed value v . Two possible null distributions are considered in the literature , the permutational distribution and the conditional distribution ( see S5 Fig ) ., In the permutational log-rank test 2 , the null distribution is obtained by assigning each patient to population P0 or P1 independently of the survival time ., Let n be the total number of patients , and n1 the number of patients in group P1 ., We consider the sample space of all ( n n 1 ) possible selections of survival times and censoring information from the observed data for the n1 patients of group P1 ., Each such selection is assigned equal probability ( n n 1 ) − 1 . In the conditional log-rank test 16 , the null distribution is defined by conditioning on Oj , Rj , and Rj , 1 for j = 1 , … , k ., If at time tj there are a total of Rj patients at risk , including Rj , 1 patients in P1 , then under the assumption of no difference in the survival of P0 and P1 the Oj events at time tj are split between P0 and P1 according to a hypergeometric distribution with parameters Rj , Rj , 1 , and Oj ., We considered the two versions of the log-rank test , the conditional 16 and the permutational 2 , that differ in the null distribution they consider ., The conditional log-rank test is preferred in clinical trials because it does not assume equal distribution of censoring in the two populations ., This is important in clinical trials when patients in the two groups are subject to different treatments that may affect their probability of leaving the trial ., However , unequal censoring is not a concern in genomic studies , since we do not expect a DNA sequence variant to have an impact of the censoring in the population ., Moreover , in the genomic studies of interest to this work , the patients are not assigned to the two populations at the beginning of the study , and the measured individuals are instead repeatedly partitioned into two populations determined by a genomic variable ., Therefore under the null hypothesis of no association between a genomic variable and survival time the two populations can be assumed to have the same distribution of potential follow-up times , and the correction for unequal follow-up , that is a concern in clinical trials 33 , is not required in our scenario ., In both versions of the test , under the null distribution the prefix sums of the log-rank statistic define a martingale , and by the martingale central limit theorem 3 , the normalized log-rank statistic has an asymptotic 𝓝 ( 0 , 1 ) distribution ., ( Sometimes the log-rank test is described using an asymptotic χ2 distribution; the two version of the tests are related , and our results hold for the version of the log-rank test based χ2 distribution as well ( S1 Fig ) ., ) The normalizing variance is different in the two null distributions and this may be reflected in differences between the p-values obtained from the two null distributions , but asymptotically the two variances are the same 17 , leading to the same p-values in the two versions of the test for large balanced populations ., Therefore , the distinction between the two versions of the test is largely ignored in practice , where most papers that use the log-rank test or software packages that implement the test do not document the specific version test they consider ., This can be explained , in part , by the widespread use of the log-rank in other scenarios , like clinical trials , where the issues specific to genomic settings ( e . g . , the imbalance between populations ) do not arise ., The differences between the tests are also rarely discussed in the literature , although there is some discussion 17 , 19 on which variance is the appropriate one to use to compute p-values from the asymptotic approximation ., In the case of small and unbalanced populations , the two null distributions yield different p-values , and the normal approximation gives poor estimates of both ( S1 Fig ) ., On simulations of cancer data , we found that the p-values from the permutational exact test are closer to the empirical p-values than the p-values obtained from the conditional exact test ( S2 Fig and S1 Text ) ., Moreover , we prefer the permutational null distribution because it better models the null hypothesis for mutation data ., While the exact computation of p-values in the conditional null distribution can be computed in polynomial time using a dynamic programming algorithm 33 , no polynomial time algorithm is known for the exact computation of the p-value in the permutational null distribution: current implementations are based on a complete enumeration algorithm , making its use impractical for large number of patients ( e . g . , the StatXact manual recommends using the enumeration algorithm only when the number of samples is at most 20 ) ., Several heuristics have been developed for related computations including: saddlepoint methods to approximate the mid-p-values 34 , methods based on the Fast Fourier Transform ( FFT ) 35–37 , and branch and bound methods 38 ., Such heuristics are shown to be asymptotically correct , converging to the correct p-value as the number of samples and computation time grows to infinity ., However , no explicit bounds are known for the accuracy of the computed p-value when these heuristics are applied to a fixed sample size and under a bounded computation time ., Therefore , when run on a specific input , these heuristics do not provide guarantees on the relation between the p-value and the approximation they report ., Given the systematic error we report below for the standard asymptotic implementation of the log-rank test , we argue that such guarantees are essential in this and many similar settings ., We developed an algorithm , Exact Log-rank Test ( ExaLT ) , to compute the p-value of the log-rank statistic under the exact permutational distribution ., In particular , we designed a fully polynomial time approximation scheme ( FPTAS ) for computing the p-value under the permutational distribution ., Our algorithm gives an explicit bound on the error in approximating the true p-value , for any given sample size , in polynomial time ., Furthermore , the output of our scheme is always a conservative or valid p-value estimate ., Conceptually , our algorithm is similar to the one presented in 39 ., Since the log-rank statistic depends only on the order of the events and not on their actual times , we can without loss of generality treat the survival data ( including the censored times ) as a

Introduction, Results, Discussion, Materials and Methods

A key challenge in genomics is to identify genetic variants that distinguish patients with different survival time following diagnosis or treatment ., While the log-rank test is widely used for this purpose , nearly all implementations of the log-rank test rely on an asymptotic approximation that is not appropriate in many genomics applications ., This is because: the two populations determined by a genetic variant may have very different sizes; and the evaluation of many possible variants demands highly accurate computation of very small p-values ., We demonstrate this problem for cancer genomics data where the standard log-rank test leads to many false positive associations between somatic mutations and survival time ., We develop and analyze a novel algorithm , Exact Log-rank Test ( ExaLT ) , that accurately computes the p-value of the log-rank statistic under an exact distribution that is appropriate for any size populations ., We demonstrate the advantages of ExaLT on data from published cancer genomics studies , finding significant differences from the reported p-values ., We analyze somatic mutations in six cancer types from The Cancer Genome Atlas ( TCGA ) , finding mutations with known association to survival as well as several novel associations ., In contrast , standard implementations of the log-rank test report dozens-hundreds of likely false positive associations as more significant than these known associations .

The identification of genetic variants associated with survival time is crucial in genomic studies ., To this end , a number of methods have been proposed to computing a p-value that summarized the difference in survival time of two or more population ., The most widely used method among these is the log-rank test ., Widely used implementations of the log-rank test present a systematic error that emerges in most genome-wide applications , where the two populations have very different sizes , and the accurate computation of very small p-values is required due to the evaluation of a number of candidate variants ., Considering cancer genomic applications , we show that the systematic error leads to many false positive associations of somatic variants and survival time ., We present and analyze a new algorithm , ExaLT that accurately computes the p-value for the log-rank test under a distribution that is appropriate for the parameters found in genomics ., Unlike previous approaches , ExaLT allows to control the accuracy of the computation ., We use ExaLT to analyze cancer genomics data from The Cancer Genome Atlas ( TCGA ) , identifying several novel associations in addition to well known associations ., In contrast , the standard implementations of the log-rank test report a huge number of presumably false positive associations .

journal.pcbi.1000137

Innate Visual Learning through Spontaneous Activity Patterns

The classic debates of nature vs . nurture , or innate vs . learned , are pervasive in the literature of early visual development ., A variety of studies have shown that the visual system requires external experience to mature ( e . g . , 1–3 ) ., On the other hand , many animals are able to see at birth , and have a functioning primary visual cortex even before eye opening ( e . g . , 4 , 5 ) ., It might seem straightforward to assign the properties found at birth to be innate and the properties dependent on visual experience to be learned ., However , a strict dichotomy may unnecessarily limit our integrated understanding of visual development ., In particular , we wish to focus on the issue of a form of learning that occurs before birth on patterns of activity that are generated internally ., It is well known that spontaneous endogenous activity is necessary , or permissive , for the proper development of the visual system ( see 6 for review ) ., The point of this paper is to discuss the statistical aspects of this activity that may be sufficient , or instructive , to guide development in much the same way that visual experience refines the mature visual system ., Essentially we propose an “innate learning” approach which prepares the system for later experienced-based refinement – a diplomatic balance between nature and nurture ., Several Studies have shown that in the early stages of rat visual development , retinal neurons are spontaneously active and correlated in their bursting patterns of activity 7 , 8 ., Later , these retinal wave patterns were recorded from many animals by calcium imaging in the developing retina\xa06 , 9 , with one example shown in figure 1A ., Experiments since then have manipulated these waves by abolishing them , over-stimulating them , or otherwise altering their properties and have shown how they are necessary for proper development 10–15 ., Several models have been proposed for the production of these waves 16–18 ., For the two most recent models , cholinergic amacrine cells mediate this activity with general agreement about the mechanism ., Neurons begin bursting spontaneously , while neighboring cells can be recruited if enough cells in the local area are also bursting ., With such rules for wave formation and propagation , biologically plausible models of retinal wave formation have been able to create complex images , such as those in figure 1B ., Although retinal spontaneous activity has been well studied , many areas beyond the retina exhibit patterned , spontaneous neural activity ., In the visual system , both the LGN 19 and V1 20–22 have patterned , spontaneous activity during development ., The effects on LGN and V1 connectivity have been analyzed functionally by layer segregation and orientation column formation 23 , 24 ., Patterned , spontaneous activity is also known to occur in the developing auditory system and is necessary for proper development 25 , 26 ., Similar developmental mechanisms are also found in hippocampus 27–29 and spinal cord 30 , 31 ., From a biophysical perspective it has been shown that spontaneous neural activity is necessary to mediate many mechanistic effects such as axon branching 32 , dendritic patterning 33 , and synaptic pruning 23 , 34 ., With the ubiquity of spontaneous activity in development and its ability to affect various aspects of neural connectivity , understanding the general role of spontaneous activity in early visual development is likely to have implications beyond vision ., In adult primary visual cortex , it has been known for nearly half a century that V1 cells respond strongly to bars and edges 35 with later experiments demonstrating that simple cells in V1 have a characteristic filter description much like a 2D gabor function 36 , 37 as shown in figure 2A ., The V1 cell has specific elongated subregions of visual space where relatively bright or dark parts in the visual image will stimulate the cell ., Note that this characterization is purely descriptive as a stimulus-response paradigm by answering “what” the neuron responds to instead of “why” the filters have that appearance ., According to the efficient coding hypothesis , the role of the early visual system is to remove statistical redundancy in the visual code 38 , 39 ., From this hypothesis , one way to understand the visual system is to develop and analyze a visual encoding scheme to remove the redundancy in images of natural scenes ., This was done using sparse coding 40 and independent components analysis ( ICA ) 41 on a set of natural images – pictures of rocks , trees , forest scenes , etc . ., ., The derived filters resemble the 2D gabor filters found in V1 simple cells – see figure 2B-C ., One can conclude from these results that V1 is developed and tuned to efficiently encode the visual world ., In this paper , we make the claim that there is a parsimonious computational reason for the existence of spontaneous patterns - a functional strategy that the early visual system can employ to guide this development both prior to and throughout experience ., In addition to molecular guidance cues we believe the visual code is refined from training on patterns of spontaneous activity during development in a similar manner to how the juvenile animals refine the visual code on statistical patterns found in natural images ., Many statistical structure models rely on the pairwise correlations between neighboring units ( also known as second-order statistics ) – an implicit assumption in other functional descriptions of spontaneous activity 42–47 ., However , many efficient coding models applied to natural images , such as sparse coding and ICA , rely on statistics beyond pairwise correlations ., In fact , often as a first step these correlations are removed in a process known as decorrelation or “whitening” ( e . g . , 40 , 41 , 48 ) ; a process that at least in part is considered a function of retinal ganglion cells 49 ( see 50 for a discussion ) ., Although the developmental activity patterns are known to have relevant pairwise correlations , we argue receptive field refinement may also rely on higher-order statistics – thus bridging the gap between models of sparse , efficient coding and spontaneous activity ., We will demonstrate that simple patterns of activity can be used as training images for refining the visual code ., The patterns we use resemble the only 2D imaged spontaneous activity available – retinal waves; this is demonstrated in figure 1 , with specific examples of our generated patterns in figure 1D ., Beyond a visual resemblance , our pattern generation technique also abstracts from the general properties and parameters of the current retinal wave models ., We strongly note , however , that this is strictly not a retinal wave model but an abstraction of what we believe are the essential features of the relevant endogenous activity ., We are more concerned in this paper with the statistical nature of the produced activity than its precise localization – including whether the activity originates in one particular area or is part of a larger , dynamical system ., For example , in comparison to retinal waves , LGN/V1 spontaneous activity has a more direct influence on cortical receptive field formation ., In ferrets the LGN remains spontaneously active at the beginning of V1 activation , while V1 activity and retinal wave production do not significantly overlap in time 6 ., LGN and V1 activity have been experimentally characterized 19 , 22 , but are far less understood than retinal activity , thus prompting our analogies to retinal waves in this paper ., Our patterns are generated using a variant of traditional site percolation models 51 - the analogy to retinal wave propagation and its relation to physiological models is detailed in the discussion section ., Models common to the study of critical phenomena in physics , such as percolation models or the Ising model , have been used in artificial neural networks and understanding adult retinal neurons and can be equally useful in understanding models of development ., Ising models , for example , have been adapted as artificial neural networks since Hopfields network 52 ., Recent work has also shown that Ising models are apt analogies for the maximal entropy and high-predictability neural firing in the retina upon natural stimulation 53 ., Although the pattern generation technique we use is quite abstract , similar networks have been shown to be relevant biologically and demonstrate desirable statistical properties ., The main goal of this paper is to show how the same adaptive , efficient algorithm can be applied for both natural inputs as well as spontaneous activity ., We show that certain wavefront-containing patterns possess the relevant statistics and a percolation network provides a useful abstraction for demonstrating this property ., These patterns , independent of how they were generated , can simply be used as an existence proof for the possible training role of spontaneous activity ., First , we will show our generated patterns qualitatively resemble known patterns of spontaneous activity ., We will then compare various methods of learning V1 receptive fields –showing how both natural images and spontaneous activity patterns can be used to produce V1-like gabor filters ., We will also demonstrate how significant variations of receptive field properties can occur even at the threshold for scale invariance – showing flexibility of learning even for this simplified model ., Finally , one of the main points of this paper , as expressed in the final figure , is that the relevant statistics for sensory coding go beyond simple correlations ., There are higher-order statistics which are still present after decorrelation ., Sparse and independent efficient coding algorithms rely on these statistics , which are found in natural scenes and are also present in the particular amorphous , wavefront-containing structure of spontaneous activity patterns ., We will present how this fact points to the conclusion that the same adaptive coding strategy may then be present both before and during visual experience ., We believe the relevant statistical properties for an efficient “innate learning” strategy are present in a wide class of amorphous , wavefront patterns in which current models of spontaneous activity belong ., We hope to demonstrate this generality , and avoid the pitfalls of selecting a particular physiological model , by using an abstract technique for pattern generation ., This technique , described in detail in the methods section , can be summarized as a simple , three parameter model – a threshold , site percolation network model ., Despite its abstract nature , this technique is analogous to known spontaneous activity patterns in generation and final pattern statistics , as mentioned in the discussion section ., We began by exploring the parameter space for a suitable training pattern by varying the proportion of nodes which are able to spread activity , p , and the threshold of active neighboring nodes needed to initiate activity , t ., For a fixed t , there is a clear phase transition , the critical percolation threshold , pc ., For p>pc activity would spread over the whole image , and in the extreme case only a few small areas would remain inactive ., At p<pc , active clusters would be finite in size , and in the extreme would be exceptionally small clusters – approaching random noise ., Although not strictly a property of physiological spontaneous activity patterns , we were interested in scale-invariant patterns in this model ., For this reason , sampling was done at p\u200a=\u200apc ( along the phase transition boundary ) as shown in figure 3A ., Approximate scale invariance is a property shared with natural images 54 ., In this case , it also allows neurons with limited dendritic fields to produce consistent , large-scale statistical effects ., We also chose this sampling as a mathematical convenience so results would not require a defined scale of analysis ., Note that known spontaneous patterns - such as retinal waves - are clearly not on this self-similar boundary , but may be considered close , with many species having limited wave sizes , and others - such as chick retina - covering large areas of retina and often terminating at the edges 55 ., The next step was to find if these patterns could be used to train an efficient coding system for natural vision ., Sparse coding and ICA have been used to find approximately independent codes for natural images with resulting filters resembling those found in primary visual cortex , as shown in figure 2A-C ., Figure 2C shows the filters derived from natural images given the parameters of image patch collection and coding as detailed in the methods section ., Following the main thesis of this paper , one might ask if an efficient coding of activity from more physiologically precise models is capable of producing similar V1-like filters ., To show this , we efficiently encoded thresholded , time-lapsed retinal wave images as in figure 2 of Godfrey and Swindale 18 ., These moderately resemble images of experimentally determined retinal wave extent as shown in figure 1C ( from figure 1 of 56 ) ., The resulting V1-like filters from this data are shown in figure 2D ., Although an efficient coding of this model qualitatively produces physiological filters , we would like to demonstrate that these images are embedded in a larger class of amorphous , wavefront-containing patterns capable of producing relevant filter properties ., We believe the question of whether or not the activity comes from a particular model - or even from the retina vs . the LGN/V1 - is important , but we would like to stress the necessary statistical properties independent of the particular source ., To demonstrate this we generated a set of images from our abstract pattern generation technique with the resulting filters shown in figure 2E for comparison ., Note how the general statistical structure of natural scenes , our abstract patterns , and more physiological models of spontaneous activity all produce filters resembling those found in V1 ., To further demonstrate the ability of these amorphous , wavefront patterns to generate physiological filters , we generate sets of images along the phase-transition boundary ., Filters derived from a representative sample are shown in figure 3C ., A qualitative difference between the gabor filters is visible , and we analyzed at least one aspect of these filters – the orientation bandwidth ., We chose orientation bandwidth because it is a well-defined , physiologically measured parameter ., We fit 7 parameter gabor filters to the 16×16 pixel derived filters ., After this fit , we used the parameters of the gabor fits to find the orientation bandwidth , with histograms of these fits shown in figure 3E , along with the primate physiological median 57 ., We also coarsely explored the area below the phase transition boundary for this parameter; the transparent color contour in figure 3A indicates how the median orientation bandwidth changes in this region ., Note for p<pc a manipulation of ‘p’ is more effective at changing the orientation bandwidth than ‘t’ – one indication of how models such as this one could lead to testable predictions through pharmacological manipulations ., However , we do not intend to stress a direct comparisons to physiological filters; we know that even within neurophysiological literature , orientation and spatial frequency bandwidth decreases as newborn macaques age 58 complicating direct comparison ., We show that even with this simple generation technique and imposed self-similarity constraint , a significant variation of filters can be produced ., This variation demonstrates one way a method like this may adjust local parameters to affect global pattern statistics and more closely resemble properties of adult physiological filters and natural scene efficient coding filters ., Current models proposed to explain pre-experience cortical receptive field development rely primarily on hebbian mechanisms and pairwise correlations ., These approaches do not address the relevant statistical structure for receptive field formation related to efficient coding ., Although hebbian models are capable of achieving arbitrary levels of complexity - and can even implement sparse coding strategies in specific configurations 59 - we note that the fundamental computational insight of hebbian models relies on pairwise correlations ., In figure 4 , we address the fundamental differences between these second order and higher order correlations with respect to relevant statistical structure and receptive fields ., Note that uncorrelated noise ( “white” noise ) has no second order or higher order statistics , so techniques that rely on pairwise correlations , as in PCA , or higher-order statistics , as in ICA , do not produce filters with discernable structure ., In patterns with only second order correlations , as in the random 1/f patterns ( “pink” noise ) , PCA can produce relevant filters ., However , in these 1/f noise patterns the sparse structure on which ICA relies is not present , and structured filters do not form ., For the natural and our patterned images in this figure , we have partially removed the second order correlations by a procedure to flatten a 1/f slope in the Fourier amplitude spectrum ., This removes the correlations in images that have an approximately 1/f slope , as was shown to be the case for natural images 54 ., When we whiten the images by removing the pairwise correlations , PCA bases resembling receptive fields are , by definition , unable to form , and we see that natural images as well as the wavefront patterns still retain important image structure ., This whitened structure , reminiscent of line drawings , is efficiently encoded using ICA ., Also note that for these image sets the ICA filters are inherently localized within the filter patches ., However , encoding using PCA will not produce localized filters without the use of additional constraints ., Whichever encoding scheme is used , it should be noted that the generated wavefront patterns have both pairwise correlation structure as well as sparse , edge-like structure used by ICA ., If only correlations were necessary to prepare the visual system , there are a number of even simpler ways to create these correlations without the additional wavefront , edge-like structure ., This additional , higher-order structure can be exploited by the visual system to guide receptive field formation and maintenance ., The fact that it exists in both spontaneous activity patterns and natural scenes suggests that both endogenous and external activity may use the same method of receptive field adaptation ., The point of this paper is to show how seemingly random patterns of activity can be used as training patterns for the visual system before eye opening ., We believe that real spontaneous activity patterns are part of a class of amorphous , wavefront-containing patterns with the relevant efficient coding statistics ., The patterns we create are also part of this class but abstract out the minimal , essential features while still retaining some biological plausibility ., This pattern generation technique is of interest for the following reasons:, 1 ) conceptual and analytical simplicity ,, 2 ) statistical properties – both self-similar/correlation-based and sparse coding/edge-like structure , and, 3 ) biological plausibility ., First , the technique is a simply stated three parameter model , collapsing to a one parameter model if you fix the neighborhood radius ( r\u200a=\u200a3 , here ) and require fractal self-similarity ( p\u200a=\u200apc ) ., Also , this technique is not only conceptually simple , but simple to implement given a biological substrate of dendritic fields , local activity pooling , and activation thresholds ., Second , the statistical properties have been discussed in detail – this pattern generation technique is capable of extremes from complete noise to clusters of activity to full activation; self-similar fractal patterns with similar statistics at all spatial scales; and edge statistics which vary the fractal dimension of the edges and consequently the sparse-coding structure of the resultant filters ., Third , this technique can be considered an abstraction of more biologically plausible models ., The retinal wave model of Butts and Feller 17 showed that wave propagation speed and termination were primarily determined by a 2-D map of one summary variable , f – the local fraction of recruitable amacrine cells – similar to our variable ‘p’ ., Their random variation of this parameter came from variations in cell refractory period , temporal dynamics from multiple waves , and influence of non-propagating spontaneous activity ., Although the more recent Godfrey and Swindale model 18 does not offer an equivalent summary variable , we believe a similar abstraction of local network excitability is equally possible ., In the Butts model a neuron would only fire if a threshold of neighboring cells fired , similar to our ‘t’ , while in the Godfrey model this threshold varied over time ., Both models also had a fixed dendritic field size , analogous to our ‘r’ ., Their parameters were chosen to match known physiological parameters such as wave size , speed , and frequency given neurophysiological constraints ., Our parameters choice , however , was more constrained by theoretical and computational concerns ., It may be useful to compare these models; for example , pharmacological manipulations of amacrine cell recruitability or neural firing threshold could move pattern generation along our p-t phase plane vertically or horizontally respectively , leading to potentially testable predictions ., We however consider this particular pattern generation technique better suited as a conceptual model to address a developmental paradigm and limitation of current statistical techniques , rather than a guide for directly verifiable experiments ., As stated in the introduction , we believe that the use of highly theoretical models such as percolation networks and Ising models have been of sufficient use in understanding neural phenomena 52 , 53 to warrant application in this domain ., This method provides an alternative approach to understanding the relation between spontaneous activity and V1 development by stressing the relation to image statistics and efficient coding in individual receptive fields ., There are a number of models that stress other physiological dimensions , such as cortical column map formation , which can provide more insights to development ., Linsker 42 demonstrated orientation column ( OR ) formation in a multi-layer model beginning with uncorrelated noise ., Grabska-Barwinska and von der Malsburg demonstrate orientation column formation using recent experimental evidence of patchy , spatially periodic cortical spontaneous activity 60 ., Miller 45 developed ocular dominance ( OD ) as well as orientation column ( OR ) formation ., More current models have become even more ambitious in the development of map features ., Bednar and Miikkulainen demonstrated direction selectivity ( DR ) to create a combined map ( OR/DR ) 61 ., A later model combined these features ( OC/OR/DR ) using translated natural images 44 ., Carreira-Perpinan et . al . 46 using the elastic net model 62 included a spatial-frequency map ( OC/OR/DR/SF ) , although the relation of their input to either natural stimulation or spontaneous activity is not clear ., In each of these models the goal was to synthesize a cortical map and receptive fields which mimic known neurophysiology ., Our use of functional , efficient coding methods precludes any relation to a particular topography , but with this we generate individual receptive fields with properties more relevant to physiological filters – more spatially bandpassed and localized ., Our technique also directly addresses how the resulting code reflects its function during natural vision - by similarly efficiently encoding natural inputs using the same adaptive algorithm ., Although our model clearly lacks a columnar organization , it does uniquely address the relation of spontaneous activity to current statistical methods of efficient coding ., Although this paper stresses the effects and theoretical justifications of spontaneous activity , there are clearly limitations to this method for preparing V1 ., Crowley and Katz 24 stated that ocular dominance columns initially form through molecular guidance mechanisms , and subsequent activity was needed for maintenance and plasticity during the critical period ., Also , Ringachs connectivity model 63 , 64 shows how V1 receptive fields and functional topography could form based on the quasi-regularity of the ON/OFF center retinal ganglion cells in the retinal mosaic; with closely located ON and OFF-center cells forming simple receptive fields ., Certainly a number of molecular-guidance mechanisms are necessary for proper development , and even if rudimentary receptive fields can form through simple axon guidance mechanisms , we still believe the simplicity and functional benefits of endogenous activity suggest a plausible role in development ., The visual system will eventually refine based on the statistical structure in the experienced natural signals , and the pre-experience receptive fields can refine using the same mechanism on simple patterns ., This conceptual model is able to address general properties of this process; however , it is more difficult to address the precise nature of the receptive fields between molecular guidance cues and the onset of natural experience ., In addition to physiological details , such as optical and retinal maturity , the goals of this handoff between development processes need to be specified for a given animal ., Some precocial animals may require a well functioning visual system from the onset , implying a goal of immediate efficient coding at the expense of later adaptability ., On the other hand , altricial animals , such as monkeys and humans , may trade off immediate optimality for a certain amount of environmental adaptability; this may be one functional justification for the large spatial frequency and orientation bandwidths in neonatal monkeys ., Although a more detailed , species-specific analysis may require additional assumptions , the general strategy may be universally beneficial ., The functional benefits are an increased refinement beyond rough molecular cues using techniques which are relatively simple given the existence of a separate , adaptive learning system ., In summary , our pattern generation technique resembles known patterns of spontaneous activity in both appearance and how they are generated ., We have demonstrated that simply-generated , sparse , wavefront-containing patterns have the statistics to produce a sparse , efficient code with filters resembling those found in primary visual cortex and those produced by an efficient coding of natural scenes ., Also , this work demonstrates the critical importance of statistics beyond simple pairwise correlations ( figure, 4 ) which exist in wavefront-containing patterns ., Efficient coding models relating natural scene statistics to V1 activity have relied on higher-order statistics for over a decade ., Previous spontaneous activity models that try to explain V1 formation rely only on lower-order statistics that may not be as relevant to early visual processing from a functional perspective ., The combination of a simplified abstraction of physiological methods of spontaneous activity and the demonstration that it provides a richer theoretical and computational understanding of why these patterns emerge is clearly attractive as it gives us a better , deeper understanding of the nature of spontaneous activity in development ., With spontaneous activity present in sensory systems , the hippocampus , and motor systems 65 , any additional methods of understanding this activity may lead to insights of value in many other areas of brain development ., We believe it is useful to add a computational perspective to the mechanistic interpretation of this activity - in addition to the role of spontaneous activity in axon branching , dendritic patterning , and synaptic pruning ., Clearly these implementation-level goals are necessary for function , but do not address the general , functional purpose for this connectivity ., A statistical , computational perspective is more likely to address the universal and ubiquitous nature of these patterns during development ., In this paper , we have given a parsimonious explanation of both why this activity has particular sparse , edge-like statistics beyond simple correlations and how this allows the same adaptive learning system to use both endogenous spontaneous activity and natural inputs to refine the visual code ., But more generally , we believe that by examining spontaneous activity in this way , we bring about a conceptual shift in the way people interpret developmental strategies ., In the context of the visual system , it appears that the system both learns from patterns extrinsic to its functionality , but strictly internal to the animal; a bridging point between both learned and innate interpretations of mental phenomena ., The pattern generation is a variation of site percolation 51 ., We use a simple , three parameter ( p , r , t ) model involving initiation and complete propagation of wave activity – thus the patterns have no temporal component ., On a square array of points , mark a random fraction ‘p’ of the points on the grid as potentially active ., To initiate an active cluster , we randomly select a location and activate all available points in a neighborhood radius ‘r’ ., Neighboring potentially active points near the wave can only become active if there are at least ‘t’ active points within a distance ‘r’ ., The wave is allowed to propagate until no more cells can become active ., This completely explains the method of pattern generation , but not the interesting aspects of the behavior ., Introductory percolation theory involves networks with t\u200a=\u200a1 , often with r\u200a=\u200a1 as a typical example ., When ‘p’ approaches a value known as the percolation threshold , ‘pc’ , the pattern of activity is known to be fractal – the image statistics appear similar at all scales ., For example , when p<pc , the activation terminates forming small clusters , when p>pc the activation spreads without bound leaving small holes without activity , but when p\u200a=\u200apc , both the activity and holes are nearly infinite in extent – leading to a fractal interpretation ., Images at increasingly larger scales become indistinguishable ., For examples , the following ( p , r , t ) triplets are known empirically , and in some cases analytically , to produce fractal images – ( ∼0 . 592 , 1 , 1 ) , ( ∼0 . 407 , 1 . 8 , 1 ) , ( ∼0 . 288 , 2 ,, 1 ) 66 ., For our measurements , for a given ‘r’ and ‘t’ pair , we found pc by finding the maximum derivative in the function of cluster size to ‘p’ value ., To obtain enough edge statistics on these waves , we randomly chose points to begin wave propagation until more than 20% of the available points were activated – stopping when the last wave was allowed to fully propagate ., Only the spatial statistics of these final patterns were explored ., All encoding was done by downsampling the image by 2 to minimize any local edge effects due to aliasing ., Unless otherwise noted , we set r\u200a=\u200a3 for simplicity ., The method used to analyze these patterns is demonstrated in figure 3 ., 1 ) Generate a series of patterns from a given set of parameters ,, 2 ) extract image patches from that set ,, 3 ) preprocess ( “whiten” ) the data , and find the optimal code for the data using independent component analysis ( details below ) ,, 4 ) fit 2D gabor functions to the resul

Introduction, Results, Discussion, Methods

Patterns of spontaneous activity in the developing retina , LGN , and cortex are necessary for the proper development of visual cortex ., With these patterns intact , the primary visual cortices of many newborn animals develop properties similar to those of the adult cortex but without the training benefit of visual experience ., Previous models have demonstrated how V1 responses can be initialized through mechanisms specific to development and prior to visual experience , such as using axonal guidance cues or relying on simple , pairwise correlations on spontaneous activity with additional developmental constraints ., We argue that these spontaneous patterns may be better understood as part of an “innate learning” strategy , which learns similarly on activity both before and during visual experience ., With an abstraction of spontaneous activity models , we show how the visual system may be able to bootstrap an efficient code for its natural environment prior to external visual experience , and we continue the same refinement strategy upon natural experience ., The patterns are generated through simple , local interactions and contain the same relevant statistical properties of retinal waves and hypothesized waves in the LGN and V1 ., An efficient encoding of these patterns resembles a sparse coding of natural images by producing neurons with localized , oriented , bandpass structure—the same code found in early visual cortical cells ., We address the relevance of higher-order statistical properties of spontaneous activity , how this relates to a system that may adapt similarly on activity prior to and during natural experience , and how these concepts ultimately relate to an efficient coding of our natural world .

Before many animals first open their eyes , neurons in the retina , thalamus , and visual cortex fire spontaneously in highly structured , patterned ways ., Experimental manipulations have demonstrated that this activity is necessary for proper function , but it is difficult to answer certain fundamental questions about the role of this activity by using experimental manipulations alone ., We know that the early visual system can adapt to better encode statistical regularities in the environment ., Can the same learning system that adapts to natural input be applied to this patterned activity to learn the visual code before birth ?, What qualities would we want in an instructional pattern of activity in the developing visual system ?, We answer these questions by presenting an abstract model of spontaneous activity in the early visual system—with direct relations to more physiological models ., We demonstrate that instructive statistical properties can exist in spontaneously generated patterns based on very simple , local interactions ., Also , we demonstrate that these patterns not only have the necessary pairwise correlations , which previous models have relied upon , but also additional sparse , edge-like structure ., This higher-order statistical structure is universal to natural visual scenes and is necessary to understand neural responses as an efficient coding of our natural world ., Most importantly , this additional structure would allow the visual system to use the same adaptive efficient coding strategy in two cases previously treated as separate—learning from natural visual experience as well as through innately generated patterns before visual experience .

neuroscience/theoretical neuroscience, neuroscience/sensory systems

journal.pntd.0004059

Mitochondrial Genome Analyses Suggest Multiple Trichuris Species in Humans, Baboons, and Pigs from Different Geographical Regions

Neglected tropical diseases , including helminthiases , have a devastating effect on human health ., It is estimated that about one billion people are infected with soil-transmitted helminths ( STHs ) , including the common roundworm ( Ascaris ) , hookworms ( Necator and Ancylostoma spp . ) , and whipworm ( Trichuris ) , mostly in underprivileged regions of the world 1 ., Approximately 0 . 5 billion people are infected with T . trichiura , resulting in the loss of 0 . 64 million disability-adjusted life years 2 ., Compared with adults , children are more prone to developing clinical symptoms such as dysentery , bloody diarrhea , rectal prolapse , and cognitive impairment in cases of chronic infection 3 , 4 ., Although whipworm infections in non-human primates are usually called T . trichiura , recent studies suggest that primates may host multiple species of Trichuris ., Ravasi et al . 5 found evidence of two Trichuris species in both baboons and humans based on the sequences of the internal transcribed spacers ( ITS ) of nuclear ribosomal DNA ., Another study by Hansen et al . 6 based on studies of the beta-tubulin gene and ITS-2 sequencing suggested that humans and baboons host shared Trichuris species ., On the other hand , Liu et al . 7 identified a potentially novel species of Trichuris in a non-human primate ( françois’ leaf monkey ) based on complete mitochondrial genome analysis and the ITS-1 and -2 regions ., Recently , Ghai et al . 8 suggested that Trichuris spp ., in human and non-human primates represent several species that differ in host specificity ., Therefore , there is a need to further explore which species of Trichuris that infect primates and investigate potential ( zoonotic ) routes of transmission between host species ., The whipworm of pigs , Trichuris suis is associated with production losses due to reduced growth rates and lower feed conversion efficiency 9 ., Although morphologically indistinguishable from T . trichiura , several studies identified extensive genetic diversity between T . trichiura and T . suis based on nuclear and mitochondrial DNA analysis 10–12 ., However , molecular characterization of Trichuris from sympatric pigs and humans indicated that T . suis can cause zoonotic infection in humans , emphasizing the public health importance of this pig parasite 12 ., The circular mitochondrial ( mt ) genomes are relatively small in size ( 13–26 kb ) and encode enzymes required for oxidative phosphorylation ., Mitochondrial DNA has a number of advantages for delimiting closely related species due to its high substitution rate coupled with its low effective population size , which leads to rapid lineage sorting following speciation 13 ., Comparative mitochondrial DNA analysis is therefore useful for identifying cryptic ( “hidden” ) species , i . e . , those that cannot be differentiated by traditional methods , including morphological analysis ., On the other hand , mitochondrial pseudogenes ( numts ) in the nuclear genome may lead to incorrect phylogenetic inferences , which is why caution is warranted whenever mt genes are used in phylogenetic analyses 14 ., Moreover , sole dependence on mtDNA for delineating the taxonomic status might also lead to ambiguous phylogeny and misidentification of individuals due to incomplete lineage sorting or mitochondrial introgression 15 ., Parasites with a wide geographical distribution or multiple host species may comprise cryptic species 13 ., For instance , Hypodontus macropi , an intestinal parasite in macropodid marsupials , was found to consist of several cryptic species based on mt genome analysis 16 ., In the study by Blouin 17 , the genetic difference between sibling nematode species typically ranges between 10%–20% using cox1 and nad4 mt genes , whereas intra-species variation is usually below 2% ., In the present study , we logically extend previous investigations to investigate levels of genetic variation among specimens of Trichuris from a human from Uganda , two baboons and pigs from Denmark and Uganda ., To do this , we, ( i ) sequenced and characterized complete mt genomes from individual adult worms from these three host species and, ( ii ) compared them ( at the amino acid sequence level ) with those of Trichuris spp ., of human , françois’ leaf-monkey and pig determined in previous studies , in order to assess levels of genetic variation within Trichuris among host species and geographical regions ., The human Trichuris was recovered from the feces of a child after anthelmintic treatment as part of an efficacy study as described previously 12 ., Permission was obtained from the Ministry of Health and the National Council of Science and Technology in Uganda , and the Danish Central Medical Ethics Committee approved the study ., The parents and children were informed about the study and received a consent form in both English and the local language ., Written informed consent was received for each individual participating in the study ., Worms from baboons in the Southwest National Primate Research Center , Texas , USA and the Copenhagen Zoo , Denmark were recovered during post mortem examination , which is performed both places on all animals culled on a routine basis ., T . suis was obtained from an experimentally infected pig in Denmark ., The Animal Experiments Inspectorate , Ministry of Justice , Denmark , approved the animal study protocol , which was carried out according to stipulated guidelines ( License no . 2005/561-1060 ) ., T . suis was obtained from a naturally infected pig in Uganda raised on a private farm , slaughtered , and used for local consumption ., Permission to recover worms from the animal was obtained from the owner ., Adult Trichuris worms were recovered from an olive baboon , Papio anubis , at Southwest National Primate Research Center , Texas , USA , and a hamadryas baboon , Papio hamadryas , in Copenhagen Zoo , Denmark , both during post mortem examination ., Adult Trichuris were collected from domesticated pigs post mortem from Denmark and Uganda and recovered from a human stool sample from Uganda upon anthelmintic treatment as described 12 ., Worms were rinsed with tap water and stored in 70% ethanol at 5°C until DNA extraction ., The MasterPure DNA Purification Kit ( Epicenter Biotechnologies ) was used to extract total genomic DNA from the anterior thin part of the worms according to manufacturers protocol ., Worm material was homogenized in lysis solution in an Eppendorf tube using a matching plastic pestle followed by incubation at 56°C for at least six hours ., PCR-linked restriction fragment length polymorphism analysis ( PCR-RFLP ) of the internal transcribed spacer-2 ( ITS-2 ) region was used to genotype worms , since Trichuris from primates , including humans , and pigs are morphologically indistinguishable 8 , 12 ., PCR products and digested fragments were resolved using 1 . 5% agarose gels , stained with GelRedTM ( Biotium ) , and detected using UV light ., Worms from humans and baboons showed banding patterns characteristic of T . trichiura ( ~130 , 220 and 340 bp ) and worms from pigs showed banding patterns characteristic of T . suis ( ~130 and 490 bp ) 12 ., Different primate- and pig-derived Trichuris were chosen for long-range PCR amplification and next generation sequencing ( NGS ) ., Two baboon worms , P . hamadryas ( TTB1 ) and P . anubis ( TTB2 ) , one Uganda human worm ( TTHUG ) , and two pig worms ( TSDK and TSUG ) from Denmark and Uganda , respectively were chosen based on their distinct haplotypes identified as part of another study when sequencing the rrnL gene of 140 Trichuris worms ., The two mt genomes of T . trichiura and T . suis ( Accession nos . GU385218 and GU070737 , respectively ) were aligned to identify conserved regions relevant to primer design ., However , no suitable conserved regions were identified , which precluded the design of general primers applicable to all worm samples ., Hence , different sets of primers were designed for each genome ., Primers were designed based on the genome of T . trichiura ( GU385218 ) to amplify the mt genomes of the baboon- and human-derived Trichuris in three overlapping fragments ( ~5 kbp each ) and for pig-derived Trichuris TSDK and TSUG in three overlapping fragments ( ~6 , 5 , and 3 kbp ) based on the genome of T . suis ( GU070737 ) ( Table 1 ) ., However , several obstacles were encountered in the amplification and sequencing processes ., First , only TTB2 was amplified , and other sets of primers were therefore designed to amplify the TTB1 and TTHUG genomes in two overlapping fragments ( ~8 and ~6 kbp ) ( Table 1 ) ., However , due to the presence of non-specific bands , amplified DNA from the band representing the fragment nad1–rrnL was extracted from the agarose gel using spin columns ( Millipore ) as stipulated by the manufacturer’s protocol ., Second , the library construction ( see below ) of the TTHUG genome failed , and the genome was amplified in 15 fragments of ~1 , 000 bp each , using 15 overlapping primer pairs designed based on the TTB1 mt genome ( S1 Table ) and sequenced by Sanger dideoxy-sequencing ( Macrogen Inc . , Seoul , South Korea ) ., Long-range PCR was conducted in a total volume of 20 μL containing 2 μL 10X PCR buffer , 0 . 2 mM of each dNTP , 0 . 4 mM of each primer pair , 2 . 0 mM MgCl2 , and 2 . 5 U of Long PCR Enzyme Mix ( Thermo Scientific ) ., PCR cycling conditions included initial denaturation at 92°C for 4 min , followed by 35 cycles of denaturation at 92°C ( 20 s ) , annealing at 50°C ( 30 s ) , extension at 62–67°C ( 7 min ) , and a final extension at 60–67°C for 10 min ., PCR gradient and MgCl2 titration was used to optimize the PCR for each primer pair ., PCR products were stained using GelRedTM ( Biotium ) and visualized after gel electrophoresis ( 0 . 8% agarose ) under UV light ., PCR products were cleaned enzymatically using 1 μL Exonuclease I ( Fermentas ) and 2 μL FastAP Thermosensitive Alkaline Phosphatase ( 1 U/μL ) ( Fermentas ) for each 5μL of amplicons and incubated for 15 min at 37°C , followed by 15 min at 85°C ., Finally , DNA concentration was measured using a NanoDrop 1000 spectrophotometer ( Thermo Fischer Scientific ) , and equal amounts of fragments of each genome were pooled ., Library construction , including tagging ( indexing ) of samples and NGS using the Illumina HiSeq 2000 platform , was performed by Macrogen Inc . ( Seoul , South Korea ) ., Reads ( ~100 bp ) of each genome were assembled using CLC Genomics Workbench v . 6 . 0 . 4 ( CLC Inc , Aarhus , Denmark ) de novo except for sample TSDK that was assembled using TSUG and GenBank entry GU070737 ( TSCH ) ., The files of the NGS raw data can be provided upon request ., For TTHUG , sequences were manually checked , edited , and trimmed using Vector NTI 18 and BioEdit 19 and aligned to TTB1 ., After assembly , genome annotation was performed using the pipeline MITOS 20 and BLAST search tools available through NCBI ( http://blast . ncbi . nlm . nih . gov/Blast . cgi ) ., Secondary structures for all tRNAs were predicted using tRNAScan-SE 21 and ARWEN 22 ., The genomes were compared with T . trichiura from a human in China ( TTHCH ) ( GU385218 ) ; Trichuris sp ., GHL from francois leaf monkey ( T . GHL ) from China ( KC461179 ) , and T . suis from China ( TSCH ) ( GU070737 ) ., Protein-coding genes ( PCGs ) and ribosomal DNA genes were individually extracted and aligned by ClustalW using default settings ., Another data set was generated by concatenating all PCG and rDNA sequences ., Genetic distances were estimated for these data sets using MEGA v . 6 . 1 23 ., Nucleotide diversity ( π ) was calculated across the genomes of Trichuris from humans and non-human primates and Trichuris from pigs using a sliding window of 100 bp with 25 bp steps implemented in DnaSP v . 5 24 ., Three different methods were used for phylogenetic analysis , namely Neighbor Joining ( NJ ) , Maximum Likelihood ( ML ) , and Bayesian Inferences ( BI ) ., Two different data sets ( DNA and amino acid sequences ) were generated for the phylogenetic analyses ., Amino acid sequences for the 13 PCGs were aligned using ClustalW for 10 Trichuris spp ., , namely Trichuris from baboons ( TTB1 , TTB2 ) , from humans ( TTHUG and TTHCH ) , from francois leaf monkey ( KC461179 ) , from pigs ( TSUG , TSDK and TSCH ) , and from T . ovis and T . discolor ( JQ996232 and JQ996231 , respectively ) ., Similarly , the DNA sequences representing the PCGs and rDNA genes were aligned using ClustalW ., Trichinella spiralis ( AF293969 ) was used as an outgroup in the phylogenetic analyses ., ML and NJ trees were generated using MEGA v . 6 . 1 23 ., The best-to-fit substitution model was identified using jModelTest0 . 1 . 1 25 under Akaike information criterion ( AIC ) 26 for each dataset ., BEAST v . 1 . 6 . 1 27 was used for the BI on the two data sets ., Uncorrelated log normal was used as prior for the mutation rate with mtRev as the substitution model for protein sequences and the General Time Reversible ( GTR ) model for DNA sequences , with gamma distribution and invariant sites assumed in both substitution models ., A random starting tree with Yule prior was assumed as well ., Three independent runs with 10 million steps each with a burn-in of 10 , 000 steps were carried out ., Tracer v . 1 . 6 27 was used to analyze log files of the MCMC chains , and the reliability of parameters was checked by recording effective sample size values above 200 ., Tree Annotater v . 1 . 6 . 1 27 was used to summarize the tree data with a posterior probability limit of 0 . 5 ., In order to investigate the phylogenetic relationship between the mt genome haplotypes identified in this study with other Trichuris haplotypes from primates and pigs in different geographical regions , partial ( 372bp ) cox1 sequences from GenBank were obtained ( Table 2 ) for phylogenetic analyses ., ML and NJ trees were generated using MEGA v . 6 . 1 23 , and the best-to-fit model was identified using jModelTest0 . 1 . 1 25 under Akaik information criterion ( AIC ) 26 ., Ascaris lumbricoides ( AB591799 ) was used as an outgroup ., The complete mt genomes of the primate worms TTB1 , TTB2 , and TTHUG comprised 13 , 984 , 14 , 009 , and 14 , 079 bp , respectively , whereas the two pig worms , TSDK and TSUG comprised 14 , 521 and 14 , 410 bp ( GenBank accession nos . KT449822-KT449826 ) ., The genomes contain 13 PCGs , 22 tRNAs , and two ribosomal RNA genes ( Tables 3 and 4 ) ., The general mt features , including gene synteny , is the same as previously described for Trichuris spp ., 7 , 11 , 28 as all genes are transcribed from the heavy strand , except 4 PCGs ( nad2 , nad5 , nad4 , and nad4L ) and 10 tRNA motifs ( tRNA-Met , tRNA-Phe , tRNA-His , tRNA-Arg , tRNA-Pro , tRNA-Trp , tRNA-Ile , tRNA-Gly , tRNA-Cys , and tRNA-Tyr ) , which are transcribed from the light strand ., The starting and termination codons for some PCGs differed between Trichuris spp ., recovered from identical host species ., For instance , the starting codon for TSDK is ATA for the nad4 gene , while it reads ATG in the TSUG genome; for the atp6 gene , the starting codon is GTA in TSDK , while being GTG in TSUG ., Moreover , the termination codon in the cox1 gene is TAG for TSDK and TAA for TSUG , and in the nad4 gene , TAA is the termination codon in TSDK , while it reads TAG in TSUG ., Similar observations were found in the Trichuris genomes from baboons ., For nad2 and atp6 , the starting codons were ATA and ATG , respectively , in TTB1 , while they read GTA and GTG , respectively , in TTB2 ., Likewise , the termination codons read TAG for the cox2 , nad1 , and nad5 genes in TTB1 , while they read TAA for the same genes in the TTB2 genome ., Finally , the length of the open reading frame ( ORF ) for some of the genes differed between the genomes ., nad4 and nad4L showed different ORF lengths between TSDK and TSUG ., For the TTB1 and TTB2 genomes , nad1 , nad2 , nad5 , nad4 , nad4L , atp6 , and atp8 also varied in terms of respective ORF lengths ., However , TTB1 and TTHUG were identical in terms of all initiation and termination codons and gene lengths , except for the nad4 gene , which was one amino acid ( 3 nucleotides ) shorter in TTHUG ., Genetic distances between each PCG and rDNA gene of the different genomes of Trichuris spp ., in primates and pigs are listed in Table 5 , together with differences in amino acid sequences , based on all encoded proteins ., The genetic distances between worms for individual PCGs and rDNA genes are given in S2 Table ., The highest genetic variation was found in the atp8 gene , and the most conserved gene was rrnS ., Among all the PCGs , cox1 and atp8 were found to be the most and least conserved gene , respectively ., The overall differences in nucleotide and amino acid sequences between the genomes of TTHCH and TTHUG were high ( 18 . 8% and 14 . 6% , respectively ) , whereas the baboon Trichuris TTB1 was genetically nearly identical to the human TTHUG ., TTB2 was most closely related to TTHCH with an overall nucleotide difference of 6 . 5% ., Among the primate-derived Trichuris , T . GHL from francois leaf-monkey was most distinct , with a nucleotide difference of 27%–28% compared with worms from humans and baboons ., Nucleotide differences between TSUG and TSCH ( 3 . 1% ) were much lower compared with TSDK ( ~9% ) ., Nucleotide diversity among the Trichuris genomes was analyzed using the sliding window approach for all the PCGs and rDNA genes ., The variation estimated for all primate- and pig-derived Trichuris is given in two separate windows ( Fig 1 ) ., The overall variation within the primate-derived Trichuris was higher compared with that of pig-derived Trichuris ., The rDNA and cox1 genes were found to have the lowest nucleotide diversity among pig- and primate-derived Trichuris ., Amino acid and nucleotide data sets gave similar tree topologies by all the methods applied ( NJ , ML and BI ) ., The best-to-fit model was mtREV+G+I+F for the amino acid sequences and General Time Reversible with gamma distribution and invariant sites ( GTR+G+I ) for the nucleotide sequences ., Three major groups were identified in the phylogeny based on the mt genomes ( amino acid sequences ) , namely primate- , pig- , and ruminant-derived Trichuris ( Fig 2 ) ., The nucleotide sequence-based phylogeny is provided in S1 Fig and depicts similar tree topology ., For the cox1 sequences , the best data fit was obtained with the Tamura 3-parameter model with gamma distribution ., Both NJ and ML depicted similar tree topologies; hence , the NJ tree is depicted in Fig 3 ., TSDK clustered with Spanish T . suis and forms the T . suis Europe clade , and TSCH is found in the T . suis China clade ., TSUG is most closely related to T . suis from China , which is concordant with the mt genome phylogeny ., The cox1 phylogeny also supports the presence of a Trichuris species complex infecting primates , identifying five distinct clades , which were named after the Trichuris spp ., recovered from these hosts , namely T . colobae recovered from Colobus guereza kikuyensis 29 and Trichuris sp ., GHL from francois leaf monkey 7 ., Human-derived Trichuris TTHUG and TTHCH were found in two separate clades named after the country of origin , Human Trichuris Uganda and Human Trichuris China , respectively , while the last clade comprises Trichuris sp ., from different non-human primates ( Theropithecus gelada , Macaca fascicularis , and P . anubis ) and here named Trichuris sp . non-human primates ., The baboon Trichuris TTB1 clusters with the Trichuris from baboon ( P . hamadryas ) in the Human Trichuris Uganda clade , while TTB2 clusters with the Human Trichuris China according to the mt genome phylogeny ( Fig 2 ) ., The human Trichuris from Czech Republic is found clustering with the human Trichuris from China ., Remarkably , Trichuris sp ., from P . anubis from the Czech Republic is genetically very distinct and clusters in a clade ( Trichuris sp . non-human primates ) distant to that of TTB2 , although both are Trichuris isolated from the same host species ., We sequenced the complete mt genomes of Trichuris spp ., recovered from a human , baboons , and pigs and evaluated their genetic and evolutionary relationships ., Several major haplotypes with clear genetic distinctiveness were observed , suggesting that multiple Trichuris species infect these host species and supporting the hypothesis that whipworms in primates comprise a species complex , which may also be the case for whipworms in pigs ( S3 Fig ) ., The two human Trichuris from Uganda and China were genetically distinct , and the difference in amino acid and nucleotide sequences was found to be around 14 . 6% and 18 . 8% , which is in the range of previously reported differences between different parasitic nematode species , suggesting the presence of at least two Trichuris species infecting humans ., For instance , the difference in amino acids for mt protein sequences between T . ovis and T . discolor adds up to 15 . 4% 28 , 11 . 7% between Wuchereria bancrofti and Brugia malayi 30 , 10 . 3% between Chabertia ovina and C . erschowi 31 , and ranges from 4% to 18% between different species of Trichinella 32 ., The baboon Trichuris , TTB1 , was nearly identical to the human Trichuris from Uganda , which is in accordance with a previous study analyzing beta tubulin genes and the ITS-2 region 6 , while the other baboon Trichuris , TTB2 , was genetically more related to the human Trichuris from China , suggesting that baboons—similar to humans—may also host at least two Trichuris spp ., In accordance with our study , but based on ITS-1 and -2 sequence analyses , Ravasi et al . 5 identified two different Trichuris species in humans from Cameroon and China , which were also found in chacma baboons in South Africa ., On the other hand , Trichuris from the leaf monkey was very distinct from baboon worms , suggesting different Trichuris species in non-human primates as proposed by Liu et al . 7 ., Indeed , Ghai et al . 8 recently suggested that primates may be infected with several Trichuris species , with some species only found in humans and others only found in non-human primates , while others again are shared , suggesting various degrees of host specificity of the different Trichuris spp ., in primates ., Amino acid sequence distances between TSDK compared with TSCH and TSUG were considerable ( around 6 . 1% and 5 . 5% , respectively ) , while TSCH and TSUG were genetically more closely related ( 2 . 2% ) ., Although the distances between TSDK , TSUG , and TSCH were not notably high , similar amino acid sequence distances between different parasitic nematode species have been reported , such as bewteen Ancylostoma duodenale and A . caninum ( 4% ) 33 , 34 and different Toxocara spp ., ( 5 . 6%–7 . 2% ) 35 , suggesting that pigs may also harbor different Trichuris species ., The cox1 phylogeny also supports that whipworms in primates and pigs make up a cryptic species complex ., The human Trichuris TTHCH and TTHUG cluster in two distinct clades , here designated Human Trichuris China and Human Trichuris Uganda ., The previously described Trichuris species from different non-human primates ( T . colobae and Trichuris sp . GHL in black-and-white colobus and francois leaf monkey , respectively ) 7 , 29 were also found in distinct clades ( Fig 2 ) ., Moreover , one of the clades included whipworms from other non-human primates ( olive , gelada baboons , and long-tailed macaque ) , which could represent a different Trichuris species in non-human primates ( ‘Trichuris sp . non-human primates’ clade ) ., Hence , the cox1 phylogeny suggested at least five potential Trichuris spp ., infecting primates ., Likewise , Ghai et al . 8 identified a distinct group of worms found only in non-human primates , and these might be related to the ‘Trichuris sp . non-human primates’ clade in our study ., However , the Trichuris from a black-and-white colobus was not identified as a separate species by Ghai et al . 8 , suggesting that this host can also be infected with different Trichuris spp ., , or it may reflect the use of different genetic markers between studies 36 ., For pig Trichuris , the cox1 phylogeny identified the T . suis from Spain to be genetically closely related to TSDK ( ‘T . suis Europe’ clade ) but distinct from T . suis from China , supporting the possibility that different T . suis species can be found in various geographical regions ., In addition to obvious transmission issues for Trichuris species that are shared between humans and non-human primates , the presence of different cryptic species might also be very important for implementation of appropriate control strategies ., For instance , different cryptic species of the human trematode Opisthorchis viverrini in different localities ( Laos and Thailand ) were found to have significantly different fecundity as measured by eggs/g/worm 37 ., Moreover , benzimidazole resistance has been associated with single nucleotide polymorphisms ( SNPs ) in the beta tubulin gene and has been detected in T . trichiura 38 , but the presence and frequencies of these SNPs may vary with geography 6 , 38 and between whipworms within the species complex ., Hence , control and treatment in different areas may not be equally effective , and therefore , there is a need to further explore the species diversity and compare the pathology , epidemiology , and drug susceptibility of different Trichuris species 13 ., In conclusion , based on complete mt genome analyses , we suggest the existence of a Trichuris species complex in primates and pigs ., Moreover , a rich source of genetic markers is provided that can be used to inform further investigation into the genetic variation among Trichuris spp ., infecting these hosts ., There is an urgent need to further elucidate the Trichuris species infecting primates in order to illuminate transmission routes and to identify and implement appropriate control measures ., Consequently , differences in pathology and treatment efficacy between species should be investigated ., This study also suggests that Trichuris in pigs may consist of a cryptic species complex with similar implications ., However , this hypothesis needs further testing including samples from various geographical regions and including nuclear DNA markers as well .

Introduction, Methods, Results, Discussion

The whipworms Trichuris trichiura and Trichuris suis are two parasitic nematodes of humans and pigs , respectively ., Although whipworms in human and non-human primates historically have been referred to as T . trichiura , recent reports suggest that several Trichuris spp ., are found in primates ., We sequenced and annotated complete mitochondrial genomes of Trichuris recovered from a human in Uganda , an olive baboon in the US , a hamadryas baboon in Denmark , and two pigs from Denmark and Uganda ., Comparative analyses using other published mitochondrial genomes of Trichuris recovered from a human and a porcine host in China and from a françois’ leaf-monkey ( China ) were performed , including phylogenetic analyses and pairwise genetic and amino acid distances ., Genetic and protein distances between human Trichuris in Uganda and China were high ( ~19% and 15% , respectively ) suggesting that they represented different species ., Trichuris from the olive baboon in US was genetically related to human Trichuris in China , while the other from the hamadryas baboon in Denmark was nearly identical to human Trichuris from Uganda ., Baboon-derived Trichuris was genetically distinct from Trichuris from françois’ leaf monkey , suggesting multiple whipworm species circulating among non-human primates ., The genetic and protein distances between pig Trichuris from Denmark and other regions were roughly 9% and 6% , respectively , while Chinese and Ugandan whipworms were more closely related ., Our results indicate that Trichuris species infecting humans and pigs are phylogenetically distinct across geographical regions , which might have important implications for the implementation of suitable and effective control strategies in different regions ., Moreover , we provide support for the hypothesis that Trichuris infecting primates represents a complex of cryptic species with some species being able to infect both humans and non-human primates .

Trichuris trichiura and Trichuris suis are whipworms found in humans and pigs , respectively , causing morbidity in humans and being associated with production losses in pigs ., Although Trichuris from non-human primates is attributed to T . trichiura , hence considered the same species as the one infecting humans , several recent reports question this assumption ., Morphologically similar parasites that have a wide global distribution and/or those capable of infecting several host species may comprise several ‘hidden’ species ., In this study , we sequenced , annotated , and compared the mitochondrial genomes ( including published genomes ) of Trichuris obtained from different hosts in different geographical regions , including humans ( Uganda and China ) , pigs ( China , Uganda , and Denmark ) and two types of non-human primates ( baboons and françois’ leaf monkey ) ., We found high genetic distinctiveness between human Trichuris from China and Uganda ., Likewise , pig Trichuris from Denmark and other regions also showed considerable , although lower , genetic diversity ., This suggests that both pig- and human-derived Trichuris may represent different species with potential differences in endemicity , which may have important implications for implementing effective control strategies ., Our data also suggests that Trichuris infecting primates comprises several species and may be transmitted from non-human primates to humans .

journal.ppat.1001076

Distinct Merkel Cell Polyomavirus Molecular Features in Tumour and Non Tumour Specimens from Patients with Merkel Cell Carcinoma

Polyomaviruses are small , non enveloped double stranded DNA viruses which infect many species with a restricted host range ., The initial discovery of Murine polyomavirus ( MPyV ) and Simian vacuolating 40 ( SV40 ) was closely linked to the demonstration of their experimental tumorigenic properties 1 ., Infections by the human polyomaviruses BK ( BKPyV ) , JC ( JCPyV ) , and the recently identified KI ( KIPyV ) and WU ( WUPyV ) are highly prevalent in most populations 2 , 3 ., Polyomaviruses persist latently in the host and may reactivate , causing disease in the immunocompromised 4 , 5 , but have not been firmly associated with cancer in humans 6 ., Therefore , the discovery in a rare but aggressive skin cancer , Merkel Cell Carcinoma ( MCC ) , of a fifth human Polyomavirus , named Merkel Cell Polyomavirus ( MCPyV ) has raised new interest in the oncogenic potential of human Polyomaviruses 7 ., MCPyV DNA was shown to be monoclonally integrated into most MCC , and tumour cells were found to express the major viral oncoprotein , large T antigen ( LT ) 8 ., Remarkably , MCPyV present in MCC tissue exhibited a molecular signature , consisting of mutations which truncate LT and suppress its helicase domain required for viral replication 9 ., These features are similar to molecular defects observed in MPyV 10 and bring strong evidence for a causative role of the virus in MCC ., MCC is a carcinoma of neuroendocrine cells which affects principally elderly and immunocompromised patients ., The close association between MCPyV and MCC has been confirmed in several case series worldwide 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 ., However , MCPyV has been detected in normal skin 23 , benign skin lesions 24 and non MCC skin cancer 25 , and also found in various tissues such as the oral cavity 26 , gastrointestinal and urogenital tracts 23 , tonsils 27 , respiratory tract 28 , blood 9 , and rarely in non skin cancers 19 , 25 , 29 , 30 ., Seroepidemiologic studies have revealed that most healthy individuals are infected with MCPyV , as with other human Polyomaviruses 31 , 32 , 33 , 34 , 35 ., Therefore , numerous questions related to the persistence and replication of the virus in the host and mechanisms of oncogenesis remain unanswered ., We investigated whether MCC patients clinical data were related to the presence of MCPyV and specific molecular features in tumour and non tumour tissues ., In a cohort of 33 MCC patients , we collected MCC and non MCC samples , and compared the frequencies of MCPyV DNA detection , viral load and nucleotide sequences ., We looked for integration of MCPyV in MCC cells and for the presence of integrated/mutated forms of MCPyV in non tumour samples ., This study was conducted according to the principles expressed in the Declaration of Helsinki ., According to French regulation , information was delivered to the patients on research performed on biological samples , and written informed consent was obtained for participation in the study , which was approved by the institutional review board of the Comité de protection des personnes Ile de France 3 ., All patients with Merkel cell carcinoma who attended the Dermatology Departments of Cochin and Bichat hospitals from March 2008 to April 2010 were prospectively included in the study ., Histopathologic confirmation of MCC diagnosis relied on tumour morphology consistent with MCC on hematoxylin-eosin-stained tissue sections , paranuclear dot immunostaining pattern for CK20 or positive immunostaining for synaptophysin and chromogranin A . Clinical data were retrieved from hospital case records , and included sex , age at diagnosis , site and size of primary MCC , and stage of the disease at diagnosis , according to Allens classification 36 based on primary tumour diameter ( <2 cm =\u200aI , >2 cm = II ) and the presence of metastasis ( regional lymph node metastasis = III , distant = IV ) ., Samples from primary and/or metastatic MCC lesions were recovered and included retrospectively retrieved formalin-fixed paraffin-embedded ( FFPE ) sections and/or fresh-frozen specimens for patients newly diagnosed or who relapsed ., FFPE sections from non MCC cancer tissue were retrieved if possible ., In addition , blood and/or nasal swab and/or urine were sampled in patients at inclusion and follow-up visits when possible , and disease stage and status were recorded ., Status was defined according to the presence or absence of tumour ( primary and/or metastatic ) as alive with disease ( AWD ) or in complete remission ( CR ) respectively ., Date and cause of death were recorded ., For FFPE samples , two consecutive 10 µ-thick sections were retrieved ., Frozen fragments were cut and mechanically dissociated ., DNA was extracted using the QIAmp DNA tissue extraction kit ( Qiagen ) according to the manufacturer instructions ., Nasal swabs discharged in 200 µl sterile saline buffer and urine were immediately frozen and kept at –20°C until analysis ., PBMC were obtained by Ficollpaque centrifugation ( Eurobio ) of 3 mL EDTA blood , suspended in 200 µl of PBS , immediately frozen and kept at –20°C until analysis ., DNA was extracted using the QIAmp blood DNA extraction kit ( Qiagen ) ., For tissue material , lysis was extended to 24 h ., DNA was eluted in 200 µl elution buffer ( Macheret ) , and concentration was measured by UV-spectrometry ( NanoDrop Technology , Willmington ) ., To detect MCPyV sequences , a sensitive real time PCR assay was designed using primers which encompass a 91 bp fragment of the LT oncoprotein gene , located upstream of the Rb-binding encoding sequence ( Table S1 ) 7 ., Amplification was performed on 100 ng DNA with 300 nM each primer and 100 nM probe in 50 µl Taqman Master Mix ( Applied Biosystems , Courtaboeuf ) ., After a 10 min denaturation step at 95°C , cycling conditions consisted of 50 cycles of 15 sec at 95°C and 1 min at 60°C on an ABI 7500 platform ( Applied Biosystems ) ., To check for carryover between the samples , mock samples were included in each series ., Confirmation LT3 and VP1 PCR assays were performed as described 7 ., Absolute quantification of MCPyV viral load in patients samples was obtained by establishing a calibration curve with ten-fold serial dilutions ( from 10 to 106 copies ) of known concentrations of plasmid MCCIC13 which contains 1 copy of the MCPyV genome inserted in the pCR-XL-TOPO vector 18 ., To quantify MCPyV DNA in copy number per cell in MCC and non MCC cancer samples , the housekeeping apolipoprotein B gene ( hapb ) 37 was amplified by real time PCR and the delta-delta Ct method was used ., To detect BKPyV and JCPyV sequences , real time PCR assays were designed to amplify a 79 and a 110 bp fragment of the LT antigen coding sequence in BKPyV and JCPyV genome respectively ( Table S1 ) ., Amplification was performed using the same cycling and control conditions as above ., To characterize integration of MCPyV , we used the DIPS-PCR technique which amplifies junctions between the host and MCPyV genomes as previously described 18 , 38 ., Briefly , DNA extracted from MCC tumours was digested overnight with Taq I ( Invitrogen ) , an enzyme that does not cleave the MCV350 sequence ( GenBank accession number EU375803 ) , and ligated to adapters ., Ligated fragments were subjected to 40 cycles of linear amplification with forward primers sequentially designed along the MCPyV LT gene ( Table S1 ) , followed by a 30 cycle exponential amplification using internal primers coupled to a reverse adapter-specific primer ., PCR products were visualized and cut from agarose gel , purified with QIAmp gel extraction kit ( Qiagen ) , and directly sequenced using Big-Dye terminator DNA-Sequencing technology ( Applied Biosystems ) ., To discriminate integrated versus non integrated forms of the MCPyV genome in patients samples , we amplified short fragments bracketing each of the characterized integration sites ., DNA ( 100 ng ) was subjected to PCR using a forward MCPyV-specific primer located 100 to 300 bp upstream of the integration site , and a reverse downstream primer , specific either to the human integration locus or to the MCPyV non-integrated genome ( Table S1 ) ., PCR was performed in 50 µl Ampli Taq Gold Master Mix ( Applied Biosystems ) for 50 cycles with conditions specific to each primer set ., Overlapping fragments of the MCPyV LT encoding gene were amplified using four primer pairs in order to cover the whole second exon of LT ( base pair number 861 to 3080 according to the MCV 350 genome ( Table S1 ) ., PCR was performed on 100 ng DNA by using HotStartTaq Master Mix kit ( Qiagen ) containing 0 . 5 µM primers in a final volume of 50 µL ., PCR products were then directly sequenced as above ., The cumulative rates of survival in complete remission relative to MCPyV load in primary MCC ( <1 or ≥1 copy per cell ) , and overall survival relative to the presence or absence of MCPyV DNA in PBMC were estimated by the Kaplan-Meier method ., Patients with follow-up under 3 months were excluded from analysis ., Survival in complete remission was calculated from the date of diagnosis to the date of first tumour recurrence in patients AWD or died of disease ( DOD ) at last follow-up ., Patients in complete remission were censored at their last follow-up visit ., Analyses were performed using the XLStat software ( Addinsoft , Paris , France ) ., Multivariable analysis of survival in complete remission was done by using a stepwise Cox proportional hazards model that used forward covariate entry to the model ., The proportions of patients either CR or AWD , or at stages I or II or III versus IV who had either a MCPyV negative or MCPyV positive PBMC sample were compared using the exact Fischer test ., All P values were two-sided , and P values less than . 05 were considered statistically significant ., Thirty nine patients with MCC attended the Dermatology Departments of Bichat and Cochin hospitals ., Six patients without retrieved MCC material were excluded from the study ., The remaining 33 patients included 14 males and 19 females ( sex ratio =\u200a0 . 6 ) ., Their median age at diagnosis was 77 years ( range 39–88 ) ., Four patients were immunocompromised , because of corticoid therapy for rheumatoid arthritis , hepatic transplantation , lymphopenia and recurring hairy cell leukaemia ., Thirteen ( 39% ) patients had a history of cancer other than MCC ( non MCC skin cancer and/or non skin cancer ) ( Table S2 ) ., Primary MCC was localized to the limbs , head , and trunk in 21 ( 64% ) , 11 ( 33% ) and 1 ( 3% ) cases respectively ., MCC median diameter was 25 mm ( range 7–70 mm ) ., At diagnosis , patients were at Allens stages I , II , III and IV in 9 ( 27% ) , 16 ( 48% ) , 7 ( 21% ) and 1 ( 3% ) cases respectively 36 ., The median delays from diagnosis until inclusion and last follow-up were 7 months ( up to 112 months ) and 16 months ( up to 134 months ) respectively ., At last follow-up , 18 ( 54% ) patients were in CR , 8 ( 24% ) patients were AWD and 7 ( 21% ) patients had died of disease ( DOD ) ( Table 1 ) ., We analyzed 43 MCC samples consisting of 26 primary MCC ( 15 fresh-frozen and 11 formalin fixed paraffin-embedded ( FFPE ) specimens ) , 14 skin metastasis ( 12 fresh-frozen and 2 FFPE specimens ) and 3 fresh-frozen regional lymph node metastasis samples ., Viral DNA was detected in 41/43 samples from 31/33 patients , with a median viral load quantified in 37 samples of 3 copies per cell ( range 3 . 10−3 to 3 . 103 ) ., Negative results observed in one FFPE section and one fresh frozen sample from primary MCC were confirmed using the previously described LT3 and VP1 PCR assays 7 ., In twenty four patients with follow up greater than three months , viral load in the primary tumour was analyzed with respect to survival ., Median survival in complete remission was longer in patients who had ≥1 copy per cell ( n\u200a=\u200a15 ) , than in patients who had no detectable viral DNA ( n\u200a=\u200a2 ) or <1 copy per cell ( n\u200a=\u200a7 ) ( 34 months , 95% CI =\u200a26 to 42 vs 10 months , 95% CI =\u200a7 to 14 , Kaplan Meier log-rank P\u200a=\u200a0 . 037 ) ( Figure 1 ) ., Among clinical parameters analyzed ( sex , age , limb site and size of primary tumour , and presence or absence of lymph node metastases at diagnosis ) , which didnt differ in the two groups , only female sex was associated with a better outcome ., Adjusted for sex , the relative hazard for survival in complete remission was 4 . 8 ( 95% confidence interval 0 . 90–26 , P\u200a=\u200a0 . 066 ) with primary tumour containing more than 1 copy per cell ., We then asked if , in MCC patients , MCPyV was restricted to MCC tissue ., Viral DNA was detected in 27/28 ( 96% ) nasal swabs from 21/21 patients , with a median load of 3 . 103 copies per sample ( range 5–2 . 106 ) ., MCPyV DNA was also found in 22/38 ( 58% ) urine samples from 18/28 ( 64% ) MCC patients , with a median load of 6 . 102 copies/ml ( range 100–4 . 105 ) ., In addition , MCPyV was amplified from 20/49 ( 41% ) PBMC samples from 15/30 ( 50% ) patients with a median load of 102 copies per ml whole blood ( range 10–5 . 104 ) ., MCPyV DNA was detected in 1/6 FFPE non MCC cancer samples from 5 patients ( Table S2 ) ., We wondered whether the high rate of detection of MCPyV was common to other human Polyomaviruses ., BKPyV and JCPyV DNA were amplified from 9% each of nasal swabs , 31% and 7% of urine samples , and 3% and 6% of PBMC respectively ., Since MCPyV DNA was detected infrequently in urine and in the PBMC of about half of the patients , we asked whether MCPyV DNAuria and/or DNAemia were linked to the stage and/or evolution of the disease ., No correlation between MCPyV detection in urine was found with any of these parameters ., In contrast , detection of MCPyV in PBMC was more frequent in patients sampled alive with disease ( AWD ) than in patients in complete remission ( CR ) ( 18/30 or 60% versus 2/19 or 11% , P\u200a=\u200a0 . 00083 ) ., Moreover , among AWD patients , MCPyV tended to be more frequently detected in patients with distant metastasis than at less advanced stages of the disease ( 6/6 MCPyV positive PBMC when sampled at stage IV versus 12/24 or 50% when sampled at stages I , II or III , P\u200a=\u200a0 . 057 ) ., In twenty eight patients with follow up greater than three months , the detection of MCPyV in PBMC was associated with poorer outcome , since patients with at least one positive sample had a median survival of 28 months ( 95% IC =\u200a19 to 36 ) whereas all patients with no detectable MCPyV survived after a median follow-up of 71 months ( Kaplan Meier log-rank P\u200a=\u200a0 . 003 ) ( Figure 2 ) ., All clinical parameters analyzed were comparable in the PBMC-positive and PBMC-negative groups except sex , since significantly more male patients had MCPyV-positive PBMC ( P<0 . 009 ) ., Among these parameters , only primary tumour size above 2 cm was associated with higher risk of death ., We then looked at PBMC results according to viral load in primary MCC ., Among 15 patients with ≥1 copy per cell , 4/4 patients who relapsed had positive PBMC , compared with 3/11 disease-free patients ., Among 9 patients with <1 copy per cell , 8 had a PBMC sample tested ., Two of four patients who relapsed tested positive while two with MCPyV-negative tumours tested negative ., The four patients who were disease free at last follow-up had negative PBMC ., Using the DIPS-PCR method , we demonstrated the integration of MCPyV DNA in six MCC cases ., We first confirmed viral integration in metastatic tissue from two patients , and found virus-host genome junctions identical to those previously reported in their primary tumours 18 ., In three new cases , integration of MCPyV was found to interrupt the second exon of LT downstream from the Rb binding coding sequence , whereas it interrupted the 3′ end of the VP1 gene in another one ., Integration was located on four distinct chromosomal loci , next to or into known human genes ( Table 2 ) ., Two of them , PARVA and DENND1A genes , encode proteins involved in cell junctions and in formation of clathrin coated vesicles or cell adhesion and cytoskeleton organization respectively ., A third gene , TEAD1 , encodes a transcriptional activator reported to be used by the SV40 enhancer to activate expression of the early T oncoprotein gene 39 ., Finally , in one case , integration resulted in fusion of the MCPyV LT gene and successive truncated fragments of the seventh and the tenth introns of the GMDS gene , two regions separated by approximately 200 kb in the human genome , demonstrating that large rearrangements occurred ., We sequenced the whole second exon of LT gene in MCC and non MCC tissues ., Fifty-two sequences from 26 patients displayed >99% homology with prototypes MCC350 , MCC339 , MKL-1 and TKS published sequences 7 , 9 ., To characterize strain-specific and/or tumour-specific markers , a local consensus reference was constructed by alignment of all sequences , and variations were indicated as silent , non synonymous or stop mutations , deletions or insertions ., The total number of silent and non synonymous mutations in MCC and non MCC samples with respect to the consensus differed by three and sixteen fold respectively ., Single nucleotide polymorphisms ( SNPs ) characterized strain specific signatures in eight patients , and a common single silent mutation was identified in three other patients ( Figure 3A ) ., Fourteen MCC sequences ( 13 complete and 1 partial ) were obtained from 12 patients ., In two patients , we verified that sequences from distinct metastasis exhibited 100% homology ., Mutations which truncate LT were characterized in nine cases , consisting in stop mutations ( 5 cases ) , or deletions causing frameshifts which generate stop codons ( 3 cases ) ( Figure 3A ) ., In the remaining case , a 250 bp insertion of the third intron of the DENND1A gene was lying inside LT , upstream the VP1-host genome junction identified in the second intron of the same human gene ., Overall , mutations preserved the Rb fixation domain but inactivated the helicase domain of the oncoprotein ( Figure 3B ) ., In one additional patient , we failed to amplify the 3′ end 600 last bp of LT , despite a high viral load and repeated attempts using different sets of primers , suggesting a truncation of this sequence ., Finally , in the last two cases , full length sequences were obtained and encoded a non truncated protein ., Interestingly , in 5 MCC cases where the DIPS-PCR characterized integrated-truncated LT , a complete LT sequence was amplified downstream the truncation site , suggesting the coexistence of integrated concatemers or latent episomes of the MCPyV genome and truncated-integrated viral sequences ., Then , we analyzed 16 nasal and 8 urine sequences ., All sequences were complete except three obtained from weak positive samples , probably reflecting sensitivity limits of the method used rather than truncations ., Two sequential nasal sequences from the same case showed 100% homology ., All full-length sequences were wild-type ., None of the 9 non MCC samples from 7 patients who displayed a truncated LT in MCC harboured the tumour-specific molecular signature ( Figure 1A ) ., These results suggest that nasal swabs or urine are likely to contain MCPyV DNA from either excreted or episomal virus ., PBMC sequences were obtained in 14 cases , including partial sequences obtained from 7 weak positive samples ., Complete PBMC and MCC sequences from 4 patients were compared ( Figure 1A ) ., In one case , both PBMC and MCC sequences encoded intact LT antigen open reading frame ( ORF ) ., In another case , the premature stop codon observed in MCC LT was absent from MCPyV recovered from PBMC ., In contrast , in two other cases , the specific MCC signatures ( a 5 and a 25 bp deletion respectively ) were recovered from the two patients PBMC samples ., Notably , in one of these patients , sequences from nasal swabs and urine were also analyzed and didnt harbour the tumour signature ., Since the two patients presented disseminated metastatic lesions at the time of sampling and were both DOD at the end of the study , we assume that the presence of mutated MCPyV DNA in PBMC reflected circulation of metastatic MCC cells ., To further confirm this last hypothesis , we amplified a portion of the LT gene bracketing the predicted integration site and the viral-host junction sequence in tumour and non tumour samples of five patients ., Virus-host junction sequences were amplified from MCC tissue in all cases , from PBMC in two cases , but from neither urine nor nasal swabs ., Both integrated and non integrated products were amplified from MCC samples , further suggesting the coexistence of integrated-truncated viral sequences and integrated concatemers or episomes in MCC samples ., Altogether , these results suggest that MCPyV sequences recovered at peripheral urine and respiratory sites merely correspond to free excreted virus or episomal DNA , whereas the presence in the patients PBMC of tumour-like sequences argued for the presence of circulating tumour cells ., Since the discovery by Feng et al who identified MCPyV in 6/8 MCC 7 , several large studies have demonstrated that MCPyV is associated with most cases of MCC except in Australia 11 , 12 , 13 , 17 , 19 , 22 , 24 , 26 , 33 ., We detected MCPyV in MCC from 31/33 patients ., Negative results from FFPE specimens could be due to poor conservation of DNA ., Only one fresh frozen tumour tested negative ., We estimated MCC median viral load at 3 copies per cell , with a 6 log variation between samples , consistent with other reports 13 , 18 , 22 , 40 , 41 , 42 , 43 ., Variations may be due to tissue quality , the proportion of non tumour cells in samples , or mutations in the target viral sequence ., However , variations in rates of LT expressing MCC cells were also reported 8 , 43 ., The fact that some MCC cases do not contain MCPyV DNA nor express LT suggests that MCC is a heterogeneous disease with at least two etiologies , despite a lack of phenotypic markers able to distinguish between MCPyV positive and MCPyV negative cases 8 , 44 ., Interestingly , patients with MCC containing at least 1 copy of viral genome per cell had better outcome than patients with lower values of MCPyV ., Although the low number of patients studied impairs definite conclusions , it is striking that two previous studies also reported poorer survival rate in patients with the lowest viral DNA load and LT expression in MCC 22 , 40 ., Although the mechanisms of MCC pathogenesis are unknown , variations in MCPyV load and in patients outcome further argue for heterogeneity and variable implication of the virus in the disease 11 , 12 , 16 , 21 , 45 , as previously described in HPV related and unrelated carcinoma 46 ., Several observations support the causal role of MCPyV in most MCC ., In particular , cell transformation by MCPyV was shown to depend on LT , as in other polyomaviruses-induced oncogenesis 1 ., First , MCPyV LT is able to bind and sequester the tumor-suppressor protein Rb through a conserved LxCxE motif 9 ., Second , the transformed phenotype of MCPyV-positive MCC cell lines depends on LT expression , since cells undergo growth arrest and/or death upon LT silencing 47 ., However , in two models of adenovirus and polyomavirus-induced oncogenesis , the dependence of transformed cells on viral oncoproteins was reversed upon time , since cells conserved an oncogenic phenotype while viral expression was shut-down in one case and viral sequences were lost in the other 48 , 49 ., These findings suggest that transformed cells acquire subsequent genetic alterations which circumvent their need for a continued expression of viral oncoproteins ., Therefore , more cellular genetic alterations may be necessary in virus-unrelated than in virus-related oncogenesis ., In this respect , the number of chromosomal alterations and amplifications was significantly higher in HPV-unrelated than in HPV-related carcinomas , and correlated with unfavourable prognosis 46 ., Recurrent genomic changes have been described in MCC 16 , 18 , but their link with MCPyV has not yet been extensively investigated ., Monoclonal integration of MCPyV is viewed as a key element in oncogenesis ., We identified in metastatic lesions from two patients the same virus-host genome integration characteristics previously described in their primary tumours , sustaining the hypothesis that viral integration constitutes an early event in MCC oncogenesis 7 , 18 ., We also showed the integration of truncated LT in four cases , and in one of these this led to a complex rearrangement between LT , VP1 and the target human gene ., All chromosomal integration sites identified so far differ from each other 7 , 18 ., We are currently verifying whether expression of putative target human genes , located in the vicinity or at the site of integration , is modified in tumour cells , notably TEAD1 which was reported to be used by another Polyomavirus , SV40 , as a transcriptional enhancer factor ., In addition , MCC LT sequences revealed various point or frameshift mutations which preserve the Rb binding domain but truncate the oncoprotein before the helicase domain , as in the tumour-specific molecular signatures previously described 9 , 50 ., Such truncations preserve the transformation ability of LT through Rb sequestration , but prevent viral DNA replication ., A similar loss of full-length LT has been observed in vitro and in vivo in models of SV40 and MPyV-induced carcinogenesis 51 , 52 In addition , replication-defective polyomaviruses with loss of LT binding to the origin of replication showed enhanced transforming properties 53 ., Our results extend previous observations and reinforce the hypothesis that acquisition of mutations within LT is a common feature and may be a prerequisite for carcinogenesis induced by polyomaviruses ., However , in three cases of this series and in two previously reported cases , mutations truncated LT upstream an identified nuclear localization signal , which could prevent nuclear expression of the protein 9 ., Lastly , mutations in LT were not observed in all cases in this nor in other studies 43 , 54 ., We cant exclude that these cases display mutations at other sites critical for MCPyV replication ., A point mutation in a pentanucleotide sequence of the replication origin was observed in a MCC strain and prevented replication 55 ., Finally , the fact that the full length second exon of LT was sequenced in five MCC samples although integration interrupted LT suggests that , as previously observed with Southern Blot analysis 9 , truncated/integrated and probably whole genomic copies of MCPyV coexist in tumour cells , as confirmed by PCR assay which discriminates integrated versus non integrated MCPyV genomes ., The lifecycle of MCPyV in the host is unknown ., Serological studies showed that infection is common in the general population and occurs before the third decade 33 , long before development of MCC ., Routes of transmission and sites of excretion are not completely known ., We showed presence of MCPyV in the respiratory tract of most MCC patients , in serial samples drawn at a several-month interval , in contrasts with low detection rate ( below 17% ) in non MCC patients reported in the literature and observed with our own detection method ( data not shown ) 4 , 27 , 28 , 50 , 56 , 57 , 58 ., MCPyV DNA excretion in urine , which was previously reported in one MCC case 59 , was observed in almost half of patients , above rates ( below 25% ) reported in control subjects 23 , 26 ., Comparative LT sequencing from MCC and non MCC samples revealed strain-specific SNPs ., Whereas most MCC sequences displayed tumour-specific molecular signatures , all nasal swabs and urine sequences were wild-type , suggesting that the latter correspond to excreted or episomal virus , whereas the former belong to integrated genomes ., Thus , high rates of MCPyV excretion both in the respiratory tract and urine may be a hallmark of MCC patients ., Urine excretion of BKPyV or JCPyV is frequent in immune competent subjects and increases with age , during pregnancy or immune suppression 60 , 61 ., Since excretion rates of BKPyV and JCPyV were comparable in MCC patients to rates of non MCC patients 61 , 62 , we hypothesize that patients present a specific failure to control latency of MCPyV but not all Polyomaviruses ., This hypothesis is supported by the fact that high levels of antibodies directed towards the major viral capsid protein VP1 of MCPyV but not other human Polyomaviruses were more frequently observed in MCC patients than in the general population 31 , 34 ., Indeed , in the case of BKPyV and JCPyV infection , reactivation and active shedding were positively correlated with serum antibody responses to VP1 63 , 64 ., Lastly , our results show evidence of high rates of MCPyV DNA in MCC patients PBMC , in contrast with low rates ( 0–8% ) reported in the serum , whole blood or PBMC of non MCC subjects 17 , 19 , 27 , 29 , 50 ., Moreover , MCPyV DNA detection in PBMC was significantly correlated with the disease stage and outcome since patients with at least one PBMC positive sample had shorter survival in remission that patients in whom MCPyV had not been detected in any PBMC sample ., In one patient , MCPyV recovered from PBMC had a wild-type genotype whereas the viral genome recovered from MCC had a LT truncating mutation ., We hypothesize that MCPyV DNAemia may correspond to active viral replication following reactivation , as observed with other human polyomaviruses 2 , 61 ., Indeed , MCPyV DNA detection was reported in activated circulating monocytes of one MCC patient and one control 59 ., In two patients in our study however , viral sequences recovered from PBMC displayed the patients MCC-specific molecular signature ., As both patients were sampled at a metastatic stage and subsequently died of their disease , we believe that PBMC viral DNA revealed metastatic circulating cells , since MCC cells were previously identified in the peripheral blood of one MCC patient 65 ., Altogether , our results provide new insights in the life cycle of MCPyV during MCC pathogenesis ., The low number of cases studied might weaken the statistical power of our results ., However , we suggest that quantitative and qualitative molecular analysis of MCPyV in tumour and non tumour sites of MCC patients may be a useful tool to characterize their disease stage and manage their follow-up ., We are currently designing a prospective study to confirm these results in large series of patients .

Introduction, Materials and Methods, Results, Discussion

Merkel Cell Polyomavirus ( MCPyV ) is associated with Merkel Cell carcinoma ( MCC ) , a rare , aggressive skin cancer with neuroendocrine features ., The causal role of MCPyV is highly suggested by monoclonal integration of its genome and expression of the viral large T ( LT ) antigen in MCC cells ., We investigated and characterized MCPyV molecular features in MCC , respiratory , urine and blood samples from 33 patients by quantitative PCR , sequencing and detection of integrated viral DNA ., We examined associations between either MCPyV viral load in primary MCC or MCPyV DNAemia and survival ., Results were interpreted with respect to the viral molecular signature in each compartment ., Patients with MCC containing more than 1 viral genome copy per cell had a longer period in complete remission than patients with less than 1 copy per cell ( 34 vs 10 months , P\u200a=\u200a0 . 037 ) ., Peripheral blood mononuclear cells ( PBMC ) contained MCPyV more frequently in patients sampled with disease than in patients in complete remission ( 60% vs 11% , P\u200a=\u200a0 . 00083 ) ., Moreover , the detection of MCPyV in at least one PBMC sample during follow-up was associated with a shorter overall survival ( P\u200a=\u200a0 . 003 ) ., Sequencing of viral DNA from MCC and non MCC samples characterized common single nucleotide polymorphisms defining 8 patient specific strains ., However , specific molecular signatures truncating MCPyV LT were observed in 8/12 MCC cases but not in respiratory and urinary samples from 15 patients ., New integration sites were identified in 4 MCC cases ., Finally , mutated-integrated forms of MCPyV were detected in PBMC of two patients with disseminated MCC disease , indicating circulation of metastatic cells ., We conclude that MCPyV molecular features in primary MCC tumour and PBMC may help to predict the course of the disease .

Merkel cell polyomavirus ( MCPyV ) is a recently discovered virus highly associated with a rare skin cancer , Merkel cell carcinoma ( MCC ) ., The causal role of MCPyV in cancer is suggested by integration of viral sequences into the cell genome and by a specific molecular signature ., We looked for and compared molecular species of MCPyV in tumour and non tumour samples of 33 MCC patients ., We showed that a tumour viral load greater than 1 copy per cell was associated with a better outcome , and that detection of the virus in blood but not in urine correlated with a shorter overall survival ., A tumour–specific molecular signature was found in the blood of two patients with metastatic disease , but did not occur in their respiratory nor urine samples ., We propose that molecular analysis of MCPyV in tumour and blood be used as a biomarker of infection and cancer progression in MCC patients .

dermatology/skin cancers, including melanoma and lymphoma, oncology/skin cancers, virology/viruses and cancer, virology/emerging viral diseases

journal.pcbi.1006564

A computational framework to assess genome-wide distribution of polymorphic human endogenous retrovirus-K In human populations

Endogenous retroviruses ( ERVs ) are derived from infectious retroviruses that integrated into a host germ cell at some time in the evolutionary history of a species 1–5 ., ERVs in humans ( HERVs ) comprise up to 8% of the genome and have contributed important functions to their host 6–8 ., The infection events that resulted in the contemporary profile of HERVs occurred prior to emergence of modern humans so most HERVs are fixed in human populations and those of closely related primates ., However some HERVs are still transcriptionally active and capable of causing new germline insertions so that individuals differ in the number and genomic location occupied by an ERV , a situation termed insertional polymorphism 9–11 ., Among all families of HERVs , HERV-K is the only one known to be insertionally polymorphic in humans ., However , HERV-K genomes are closely related and as with many repetitive elements , they are difficult to accurately assign to a genomic location using standard mapping approaches 12 , 13 ., The DNA form of a retrovirus is called a provirus and minimally encodes the structural gag and env gene , and genes for a protease and polymerase , termed pol ., Viral genes are flanked by long terminal repeats ( 5’ or 3’ LTR ) ., While there are several HERV-K that are full length , none are infectious and most contain mutations or deletions that affect the open reading frames or truncate the virus ., Further , the LTRs are substrates for homologous recombination , which deletes virus genes while retaining a single , or solo , LTR at the integration site 14–16 ., Insertional polymorphism typically refers to the presence or absence of a retrovirus at a specific locus 17 , 18 ., However an occupied site can contain a provirus in some individuals and a solo LTR in others and hence still display polymorphism ., Thus HERV-K and other HERVs have contributed to genomic diversity in the global human population in several ways 19 ., The presence of antibodies to HERV proteins or HERV transcripts has spurred a quest to determine if HERVs from multiple families have a role in either proliferative or degenerative diseases in humans 20–26 ., Although there are known mechanisms by which a HERV can cause disease; for example , by inducing genome structural variation through recombination 27–31 , affecting host gene expression 32 , and inappropriate activation of an immune response by viral RNA or proteins 23 , it has been difficult to establish an etiological role of a HERV in any disease ., HERV-K specifically has been associated with breast and other cancers 3 , 33–37 , and autoimmune diseases , such as rheumatoid arthritis 38 , 39 , multiple sclerosis 22 , 40 and systemic lupus erythematosus 8 , 22 , 41 without definitive evidence of causality or of specific loci involved ., Recently , a HERV-K envelope protein was shown to recapitulate the clinical and histological lesions characterizing Amyotropic Lateral Sclerosis 42 , 43 , providing an important mechanistic advance of a role for a HERV-K protein in a disease ., Despite growing evidence for a contribution of HERV-K transcripts or proteins to the pathogenesis of human disease , it is difficult to distinguish among HERV-K loci to investigate potential roles and , in particular , to determine if a loci that is polymorphic for presence or absence of a provirus could be involved ., In this paper , we focus on characterizing the genomic distribution of known insertionally polymorphic HERV-K proviruses in the 1000 Genomes Project ( KGP ) data ., We present a data-mining tool and a statistical framework that accommodates low depth whole genome sequence data characteristic of the KGP—and often patient—data to estimate the presence or absence of a provirus at all loci currently known to contain a HERV-K provirus ., Using these data , we determine the number of known polymorphic HERV-K proviruses per genome because HERV-Ks can affect genomic stability 44 contributing to the pathogenesis of a disease ., We also provide a tool to visualize HERV-K co-occurrence in global populations to facilitate exploration of synergy that might exist among specific polymorphic HERV-K in disease 45 ., Our results provide a reference of global population diversity in HERV-K proviruses at all currently known polymorphic loci in the human genome and demonstrate that there are notable differences in the prevalence of HERV-Ks in different global populations and in the total number of HERV-Ks currently known to be polymorphic within a person’s genome ., The goal of this research was to develop a computationally efficient and easy to use tool that could accurately report the status of all reported insertionally polymorphic HERV-Ks with coding potential ( provirus ) from whole genome sequence ( WGS ) data ., We use the KGP database , which represents individuals in five super-populations and 26 populations , to establish the diversity in global populations at each known polymorphic HERV-K proviral locus and the total number of these polymorphic HERV-K in individual genomes to provide a foundation to study the role of HERV-K in human disease ., Our reference set consists of all HERV-K sequences that are available in public databases and that can be unambiguously assigned a location in hg19 ., Sequences of HERV-K that are not present in hg19 but that were generated by PCR primers to the host flanking regions are included in the reference HERV-K set ., From these HERV-K reference sequences , we generate a set of k-mers ( see S2 Fig for optimizing k ) that are unique to all HERV-Ks at each locus ., The analysis of subject data starts with a data mining step that recovers all whole genome sequence reads that map to identified HERV-K elements in hg19 ., The rationale here is that polymorphic HERV-K that are not present in hg19 are greater than 80% homologous to those in the human reference genome and will map on existing elements ., The recovered reads from a query WGS data set are then reduced to k-mers and mapped , requiring 100% match , to the reference set of k-mers ( T ) , which represents all unique sites for HERV-K at each locus ., The output is a ratio ( n/T ) of subject k-mers ( n ) that are 100% match to the reference k-mers ( T ) ( see Methods for full details; the value of T for each HERV-K is in S1 Dataset:virus ) ., Our preliminary analysis of the KGP data demonstrated that our k-mer-based approach is sensitive to sequence depth; some HERV-K loci are represented by an almost continuous range of n/T values from 0–1 ( S1 Fig ) , making presence/absence classification difficult ., However , the majority of the KGP data is approximately 6x depth and thus to make use of this important resource , we developed a mixture model to statistically assign the n/T values from genomes to a cluster considering the sequence depth ., K was optimized to 50 because this value improved our model computational efficiency and output ( Fig 1B , S1 Text , S2 Fig ) ., The affect of sequence depth on n/T can be seen by comparing the sequence data of 28 individuals in the KGP data that have both low and high sequence depth data ( Fig 1 shows a subset of eight individuals for clarity ) ., If read depth is greater than 20 , there is less dispersion of n/T values , most likely because more reads from the query WGS data are recovered from the mapped intervals ., The states , ‘provirus’ , ‘solo LTR’ , and ‘absent’ are preliminarily assigned to each cluster based on the high depth data ( data in Fig 1B used for description below ) ., Individuals with n/T = 1 have the reference allele ( represented by the yellow cluster of low depth data ) and n/T = 0 ( red cluster ) indicates that the HERV-K is absent ( no k-mers to unique sites in the HERV-K at this locus were recovered from mapped sequence reads ) ., The k-mers derived from persons with low ( green ) and intermediate ( blue ) n/T values were mapped to the HERV-K reference for this locus to determine whether they localized only in the LTR ( assign ‘solo LTR’ to green cluster ) or in the coding region ( assign ‘provirus’ to blue cluster ) ( S3 Fig ) ., The WGS data of each individual in the KGP dataset were evaluated using our optimized analysis workflow ., HERV-Ks on chrY were not considered ., Twenty sites , omitting one at chr1:73594980 see Methods that have been reported to be polymorphic for presence/absence 10 , 11 , 34 , 46 were identified as polymorphic for a HERV-K provirus by our analysis ( S1 Dataset:virus ) ., Polymorphic HERV-Ks greater than 6 kbp in length cluster together in a phylogenetic analysis indicating that they are closely related ( S4 Fig ) ., The prevalence ( proportion of individuals in a given population with a provirus present at a given locus ) of the 20 polymorphic HERV-K proviruses varied from 0 . 9% to 99 . 5% when averaged across the entire KGP dataset ( Table 1 ) ., However , there were notable differences in prevalence at each HERV-K site among the five super-populations ( AFR , EAS , AMR , EUR , SAS; see Methods for key to abbreviations ) ., Of the 20 , the prevalence of seven polymorphic HERV-Ks was greater than 90% and the difference between populations with the lowest and highest prevalence was less than 6 . 5% ( Table 1 ) ., There was 100% occupancy for six of the seven high prevalence polymorphic HERV-Ks ( 98 . 8% for the seventh ) , indicating that the rate of conversion to solo LTR is low for viruses at these sites ( see S1 Text for occupancy and S2 Dataset:KGP ( absence , solo , presence ) for model prediction of solo LTR prevalence ) ., Two polymorphic HERV-Ks had an overall prevalence of less than 10% in any population ( Table 1 ) and were found in individuals of AFR origin; we found no evidence of a solo LTR at these two sites ., Nine of the remaining 11 HERV-Ks are of interest because the difference between super-populations with the highest and lowest prevalence is between 28 and 80 percentage points ( Table 1 ) ., Of note , for the three HERV-Ks with the largest difference among super-populations , the prevalence is lowest in EAS populations ., Individuals from African populations differ significantly from the other four super-populations in the prevalence of ten of the polymorphic HERV-K , three of which occur in close proximity on chr19 ., ( Table 1 , S2 Dataset:compare_prevalence ) ., EUR and AFR super-populations are significantly different in the prevalence at all but one of the 20 polymorphic HERV-K based on adjusted p-values ( S2 Dataset:compare_prevalence ) ., The HERV-K genome is close to 10 kbp ., As there are 20 known HERV-K loci with the potential to encode a provirus that are polymorphic in human populations , we asked if there is a difference in the burden of these repetitive , and potentially functional , viral elements among individuals ., This was indeed the case ., Of the 20 polymorphic HERV-K proviruses assessed , the number per person’s genome ranges from 7–18 ( Fig 2 , S2 Dataset:HERV-K per person ) ., More than 63% of individuals from all super-populations except EAS carry 12 to 14 proviruses in their genome ., Individuals from EAS have a lower burden with 69% of individuals carrying 9–11 of the 20 polymorphic HERV-K proviruses ., 7% of AFR individuals have 16 or 17 proviruses compared to a maximum of 2% in other groups ( S2 Dataset:HERV-K per person ) ., These data suggest that a comprehensive investigation of polymorphic HERV-Ks may be a more productive means to advance studies of their potential disease impact ., Our data provide a comprehensive picture of sites occupied by HERV-K provirus in each genome ., Although most previous studies investigating a role of HERV-K in human disease assessed the prevalence of the HERV-K at a given locus , it is possible that , for example , two HERV-Ks each at 40% prevalence in a population rarely co-occur in an individual genome ., By providing the status of all known polymorphic HERV-K in the genome , our tools facilitate such assessment and can advance investigation of HERV-K and human disease ., We assessed combinations of three , four and five polymorphic HERV-Ks in KGP data and found that there are many combinations of co-occurring viruses that are population-specific ( S3 Dataset ) ., To facilitate exploration of HERV-K combinations among KGP populations , we developed a D3 . j visualization tool ( see Methods ) that allows a user to choose any combination of the 20 polymorphic HERV-K proviruses and display the co-occurrence prevalence among the 26 populations represented in the KGP data ., As an example , we show a combination of four HERV-Ks to represent the variation that occurs in KGP individuals , which in this case ranges from 3% in EAS to 59% in EUR ( Fig 3A ) ., We also determine that the three polymorphic HERV-Ks found on chr19 co-occur only from three AFR populations and in less than 2% of individuals ( Fig 3B ) ., Because there are clearly population-specific differences in both individual HERV-K prevalence and in the prevalence of HERV-K co-occurrence , we explored whether the presence or absence of these 20 documented polymorphic HERV-Ks is sufficient to distinguish populations using Fisher’s linear discriminant analysis ( LDA ) 47 ., Based on the status ‘provirus’ , ‘solo LTR’ , or ‘absence’ , there is little resolution of AFR , EUR , and EAS super-populations ( Fig 4A ) ., However , there is sufficient signature to separate AFR , EUR , and EAS if we utilize the n/T ratio of the 20 polymorphic HERV-Ks ( S5 Fig ) and we further improve population separation if we use the n/T ratio for all 96 HERV-Ks ( Fig 4B ) ., This indicates that we are losing information by reducing the data to three states and that fixed HERV-K also contain signal for population of origin ., An n/T = 1 indicates that the query set contains all k-mers that map to the reference set T for a specific HERV-K ., If there is a HERV-K allele that has not been reported in any database but that is common in a population , we expect n/T <1 because we require 100% match to reference set T and k-mers covering allelic sites will be excluded ( see Fig 1B , blue cluster for an example ) ., We assessed the density distributions of n/T plots for each of the 96 HERV-Ks for evidence of population-specific alleles ( S1 Text , S7 Fig ) ., Five HERV-Ks have some indication of population specific distributions ( S1 Dataset:virus ) ., The HERV-K at chr1:155596457–155605636 , which we report as fixed , is notable because the reference allele ( n/T = 1 ) is only found in AFR ( Fig 5A , S7 Fig ) ., Individuals from other populations have n/T near 0 . 5 ., We mapped k-mers from individuals with n/T near 0 . 5 to the reference HERV-K sequences and confirmed that there is a loss of k-mers at several sites covered by the unique reference k-mers for this virus ( S8 Fig ) ., There are also cases where the reference allele is found in all populations except AFR ( Fig 5B and see S7 Fig for additional examples ) ., Our research provides a tool to mine whole genome sequence data to collectively evaluate the status of HERV-K provirus at known polymorphic and fixed sites in the human genome ., The tool incorporates a statistical clustering algorithm to accommodate low depth sequence data and a visualization tool to explore the co-occurrence of known polymorphic HERV-K in the global populations represented in the KGP data ., There are numerous significant differences in the prevalence of individual and co-occurring known polymorphic HERV-K among the five KGP super-populations ., It is notable that individuals from EAS carry a lower total burden of the 20 polymorphic HERV-K than other represented populations ., These data provide a comprehensive framework of genomic diversity among 20 documented polymorphic HERV-K proviruses to advance studies on potential roles for HERV-K in human disease , which have been alluring yet difficult to establish 21 , 22 , 24 ., Tools developed to interrogate ERV insertional polymorphism typically exploit the unique signature created by the host-virus junction 11 , 48 , 49 ., These approaches indicate that a site is occupied by an ERV but not whether there is a provirus associated with the site , which is more difficult to accomplish with short read sequence data ., Our analysis tool provides an efficient means to detect occupancy and provirus status in one step ., We decrease computational time by analyzing only the set of reads that map to existing HERV-K loci in the reference genome ., This approach is justified because the known polymorphic HERV-K that are missing from the human reference are closely related to those in the reference genome assembly ( see S4 Fig ) and hence reads derived from them map to a related HERV-K in the reference ., We employ k-mer counting methods , which also increase computational efficiency ., A reference set of k-mers that is unique to each HERV-K is generated for each location in the genome and the proportion of reads ( n/T ) from the query set that maps to the k-mer reference set is reported as a continuous variable; there is no threshold of read count or coverage imposed for classification ., Instead we utilize a mixture model to statistically cluster values based on n/T and sequence depth and assign the same HERV-K status to all individuals in a cluster ., Clusters representing n/T of 1 consist of individuals from whom all the unique k-mers identified in the HERV-K reference set were recovered from their mapped WGS data ., We classify other clusters by determining if k-mers mapped on the reference allele are distributed at sites in the coding portion of the genome or only in the LTR; reads mapping only in the LTRs are classified as solo LTR ., This approach demonstrated that the k-mers derived from some individuals only covered a subset of the unique sites and led to the interesting finding that several HERV-K loci could have population specific alleles ., Wildschutte et al 11 have conducted the most comprehensive study of HERV-K prevalence in the KGP data to date ., The goal of that paper was to identify new polymorphic insertions , either provirus or solo LTR , based on detecting reads containing the host virus junction ., However , they implemented an additional step to detect provirus and provide the prevalence of some polymorphic HERV-K provirus for comparison with our results ( see S1 Dataset:virus for comparison of prevalence values reported in Wildschutte et al 11 ) ., There are five HERV-K previously reported in Subramanian et al 2011 10 that were not included in Wildschutte et al 11; all are polymorphic in our analysis ( range 43–99% , see Table 1 and S1 Dataset:virus-column N ) ., Seven polymorphic HERV-K , which Wildschutte et al 11 indicate occur in greater than 98% of KGP individuals , are fixed in our study ., Our estimated prevalence for 14 HERV-K differs from that reported in Wildschutte et al 11 by 5% or more ., Of these 14 , the prevalence estimates at chr1:155596457–155605636 are most divergent ., Our data show this site is fixed for provirus and Wildschutte et al 11 report that only 14% of the KGP data , all from AFR , have a HERV-K provirus integration ., Our plots for chr1:155596457–155605636 show that AFR individuals carry the reference allele at this site ( n/T near 1 , Fig 5A ) and all other individuals have n/T near 0 . 5 ., The k-mers from individuals with low n/T values for chr1:155596457–155605636 map to only a subset of sites marked by unique k-mers in the coding region ( S8 Fig ) , which is consistent with sequence polymorphism or a deletion at these positions ., The reference set T is small for this HERV-K and therefore overall coverage of the genome is low ., Because Wildschutte et al 11 used a minimum coverage threshold for their k-mer mapping method , it is possible that alleles present in non-AFR populations do not meet their inclusion criteria ., There is a similar signal for alleles , represented by lower n/T values , at the other 13 HERV-K sites although the differences between our prevalence estimates and those of Wildschutte et al 11 are small ( S1 Dataset:virus ) ., In most cases these putative alleles are found in all populations at different frequencies but in five there is some degree of population specificity ( Fig 5 , S7 Fig , S1 Dataset:virus ) ., Our results indicate that there could be considerably more sequence variation in HERV-K among human populations than previously appreciated ., These data also suggest that using a HERV-K consensus sequence to study pathogenic potential could miss important features of HERV-K proviral polymorphism , which can be characterized by both the site occupancy status ( presence/absence ) and , when present , by sequence differences among individuals ., HERV-Ks are the youngest family of endogenous retroviruses in humans and consequently they share considerable sequence identity ., This has the effect of limiting the number of unique sites associated with some HERV-K , which decreases the size of the reference set T ( S1 Dataset:virus ) ., The set T is small for near identical HERV-K such as HERV-Ks involved in a duplication event ., The HERV-Ks at chr1:13458305–13467826 and chr1:13678850–13688242 are identical and cannot be distinguished ., We report n/T for only one of these HERV-K ( see S1 Dataset:virus , column M ) ., We treat the two HERV-K proviruses spanning chr7:4622057–4640031 as a single virus with n/T = 1 reflecting the tandem arrangement found in the hg19 ., In this case , n/T<1 can mean either that both proviruses are present but with substitutions at a unique k-mer site or that one provirus converted to a solo LTR ., Thus although an n/T ratio of 0 or 1 reliably indicates absence and presence of reference HERV-Ks , respectively , when T is small , sequence polymorphism and a deletion event can be difficult to distinguish from a solo LTR ., However , because our mixture model statistically clusters similar n/T values based on sequence depth , all individuals in a cluster have the same status ( e . g allele or solo LTR ) even if we do not know what that state is ., The ability of our tools to resolve the status of closely related HERV-K provirus sequences will improve as more empirical sequence data becomes available ., Our approach provides researchers with a rapid means to determine if the prevalence , and overall burden of the 96 HERV-K proviruses evaluated differ between a patient data set and the population represented in KGP to which they trace ancestry ., The visualization tool will facilitate investigation of combinations of HERV-Ks in certain clinical conditions ., The potential that HERV-K has multiple allelic forms in different populations is worthy of further analysis because a sequence allele could also contribute to a disease condition ., The 96 HERV-K proviruses previously reported 10 , 11 , 34 , 46 were supplemented with HERV-K alleles present in the NCBI nt database ( November 2016 release ) ( 92 in hg19 , and 4 from the NCBI nt database ) ., We required that any allele of a HERV-K from the nt database have at least 2kb of hg19 reference-matching host flanking sequence to confirm genome location ., In total , 234 alleles were collected at the 96 known HERV-K loci ., The location information and virus features are summarized in S1 Dataset: virus ., We identified the k-mers that correspond to unique sequence characterizing each HERV-K ., K-mers are substrings ( subsequences ) of length k that exist in a string ( DNA sequence ) ., The length k is determined empirically ( S1 Text ) ., Each k-mer is labeled with the corresponding viruses in which it is observed ., Only those k-mers referring to a single virus locus , unique k-mers , are selected for the set T . Where multiple alleles of a HERV-K are available , k-mers unique to all alleles at that location comprise T . Multiple 2bps different k-mers ( such as SNPs ) corresponding to the same location on the virus , are merged into a single entry for the purposes of computing T . We map unique k-mers back to the corresponding alleles to determine coverage of the HERV-K and whether k-mers are located in LTRs ( S3 Fig; S1 Dataset: virus ) ., To develop a method to recover sequences containing information on HERV-K we leverage the fact that HERV-Ks are closely related ., Thus , most sequence reads obtained from an individual with a polymorphic HERV-K that is absent in the human reference , hg19 , will map to the location of a closely related HERV-K that is present the human genome reference ., ( As we show in S4 Fig , the known polymorphic HERV-K proviruses are closely related . ), A file with the coordinates for all reported HERV-K insertions is used to extract mapped reads from a genome sequence file ( S1 Dataset:bed , which provides the coordinates for both hg19 and hg38 ) ., Note that the KGP data were mapped to GRCh37 , which includes the decoy sequence hs37d5 ., This decoy contains the HERV-K at chr1:73594980_73595948 , which is not present in hg19 ., Thus , we did not recover any reads for this HERV-K , which is polymorphic but reportedly at high prevalence in most populations 11 ., The KGP data were downloaded in aligned Binary Alignment/Map ( BAM ) format ( ftp://ftp . ncbi . nlm . nih . gov/1000genomes/ftp/data/ ) ., It contains data for 2 , 535 individuals ( S1 Dataset:KGP ) sequenced via low-depth whole-genome sequencing ( mean depth = 6 . 98X ) ., The individuals represent 26 populations , derived from 5 super-populations , including African ( AFR ) , Admixed America ( AMR ) , East Asian ( EAS ) , European ( EUR ) , and South Asian ( SAS ) 50 , 51 ., Of 2 , 535 individuals , 28 also have high-depth DNA sequences ( mean depth = 48 . 06X ) , which we use as a pilot dataset to develop the mixture model , described below and in Supplementary Text ., Our computational framework to indicate the status of each known HERV-K provirus is based on the n/T ratio , which is the proportion of k-mers in the data mined from WGS of each individual that are identical to the reference set T for each HERV-K provirus ., Sequence reads are extracted from a mapped file of whole human genome sequence data based on coordinates corresponding to each annotated HERV-K ., The reads are k-merized and mapped to the set T , which represents all unique k-mers assigned to each HERV-K in the reference set ., We use exact match to map the k-mer data set to the unique k-mer references ., The n/T ratio is an indicator of the presence of each HERV-K; n/T = 1 indicates that the individual has the HERV-K in our reference dataset documented to be at that locus while n/T = 0 indicates that no k-mers unique to a HERV-K locus were recovered ( see Fig 1 for more explanation ) ., Using a hash table ( S1 Text ) , it takes 15 minutes to generate the n/T matrix for 100 files ., The source code for the entire process is at https://github . com/lwl1112/polymorphicHERV We utilized a statistical model to account for the dependency of the number of k-mers obtained from a person’s sequence data ( denoted by nik for the ith subject and kth HERV-K , with i = 1 , … , I , k = 1 , … , 96 ) that maps to the reference set T for each HERV-K on sequencing depth ., Thus for each HERV-K we could statistically cluster those nik/T values for i = 1 , … , I based on the sequence depth of the WGS data for each individual for subsequent biological classification ( provirus , solo LTR , absence , see Fig 1 ) ., More specifically in our analysis , for each k HERV-K , k = 1 , … , 96 , consider a sample of size I measurements xi ( i = 1:I ) , where each xi is a vector of length 2 xi = ( xi1 , xi2 ) with xi1 being the nik/T measurement and xi2 the log function of depth ., Here , for notation simplification , we use xi instead of xik ., To perform clustering analysis , we utilize the mixture model approach , which is arguably the most widely used statistical method for clustering ., Specifically , we follow the work proposed by Lin et al . 52 that employs a Gaussian Mixture Model ( GMM ) with density function given by, f ( xi|θ ) =∑j=1MπjN ( μj , Σj ) , fori=1:I, ( 1 ), where all relevant and needed ( unknown ) parameters are represented by θ = ( π{1:M} , μ{1:M} , Σ{1:M} ) ., N ( μj , Σj ) is the Gaussian density for the jth component parameterized by the 2-dimensional mean vector μj and 2x2 covariance matrix Σj ., π{1:M} are the mixture components prior probabilities summing to 1 ., To allow a flexible modeling approach , we employ the standard Bayesian ( truncated ) Dirichlet Process prior for the parameters θ = ( πj , μj , Σj , j = 1:M ) 53 , 54 ., The idea is that some of the mixture probabilities ( πj ) can be zero , hence the actual number of mixture components needed may be smaller than the upper bound M . This mechanism allows automatic determination of the number of mixture components needed by the data set at hand ., For model estimation , a latent indicator Zi ∈{1 , 2 , … , M} with P ( Zi = j ) = πj is used , for i = 1:I ., Specifically , Zi = j if , and only if , xi comes from component j ., Given a fitted model via the Bayesian expectation–maximization algorithm , in terms of estimates of all parameters θ , instead of interpreting the fitted Gaussian mixture components as clusters , we identify clusters by aggregating Gaussian components so that non-Gaussian type of clusters can be flexibly represented ., Merging components into clusters can be done by associating each of the Gaussian components to the closest mode of f ( x1:I|θ ) = ∏i = 1:If ( xi|θ ) ., Hence , the number of modes identified is the realized number of clusters ., S1 Text for additional detail We consider that both the individual prevalence of a HERV-K and the co-occurrence of multiple HERV-Ks could differ among populations ., The time of a brute-force approach for finding all combinations Cm of size m from p polymorphic HERV-K is ( ∑m=1p ( pm ) =2p−1 ) , which is not efficient and is redundant ., We employed the Apriori algorithm 55 , which is commonly used for finding frequent pattern sets; in our case indicating which of the known polymorphic HERV-K frequently appear together ., It first generates combinations Cm ( initialized to 1 ) ., In the optimization , frequent combinations Fm are returned from candidates Cm when prevalence exceeds the minimum threshold of co-occurrence ., Fm are then self-joined to generate combinations Cm+1 of size m +1 and out of which Fm+1 satisfy the minimum co-occurrence ., In each pass , candidate combinations are pruned so as to avoid generating all combinations , which reduces running time significantly ., We made statistical comparisons across 5 super-populations for the following three problems ., For each problem , there are ( 52 ) = 10 families of 1-to-1 comparisons conducted ., The ‘prop-test’ function in R is used to test whether the proportions for two super-populations are the same ., Therefore , multiple hypotheses would be conducted on frequencies F across super-populations P1…5 as follows: Null hypothesis , H0:FPi=FPj , where i≠j; Alternative hypothesis , HA:FPi≠FPj , where i≠j ., A separate P-value is computed for each test and the Benjamini-Hochberg procedure 56 is used to account for multiple comparisons ., We utilized D3 . js ( Data Driven Documents ) 57 , an open-source java script library to create an interactive visualization to display co-occurrence of polymorphic HERV-Ks in human populations ., Our visualization system includes two modules , a welcome page and a result page ., Input JSON data include locations of polymorphic HERV-K , population information , and the 0/1 ( absence / presence ) matrix ., ( See S1 Text ) ., Source code is available at: https://github . com/lwl1112/polymorphicHERV/tree/master/visualization and a searchable tool with the data reported here is at: http://pages . iu . edu/~wli6/visualization/

Introduction, Results, Discussion, Materials and methods

Human Endogenous Retrovirus type K ( HERV-K ) is the only HERV known to be insertionally polymorphic; not all individuals have a retrovirus at a specific genomic location ., It is possible that HERV-Ks contribute to human disease because people differ in both number and genomic location of these retroviruses ., Indeed viral transcripts , proteins , and antibody against HERV-K are detected in cancers , auto-immune , and neurodegenerative diseases ., However , attempts to link a polymorphic HERV-K with any disease have been frustrated in part because population prevalence of HERV-K provirus at each polymorphic site is lacking and it is challenging to identify closely related elements such as HERV-K from short read sequence data ., We present an integrated and computationally robust approach that uses whole genome short read data to determine the occupation status at all sites reported to contain a HERV-K provirus ., Our method estimates the proportion of fixed length genomic sequence ( k-mers ) from whole genome sequence data matching a reference set of k-mers unique to each HERV-K locus and applies mixture model-based clustering of these values to account for low depth sequence data ., Our analysis of 1000 Genomes Project Data ( KGP ) reveals numerous differences among the five KGP super-populations in the prevalence of individual and co-occurring HERV-K proviruses; we provide a visualization tool to easily depict the proportion of the KGP populations with any combination of polymorphic HERV-K provirus ., Further , because HERV-K is insertionally polymorphic , the genome burden of known polymorphic HERV-K is variable in humans; this burden is lowest in East Asian ( EAS ) individuals ., Our study identifies population-specific sequence variation for HERV-K proviruses at several loci ., We expect these resources will advance research on HERV-K contributions to human diseases .

Human Endogenous Retrovirus type K ( HERV-K ) is the youngest of retrovirus families in the human genome and is the only group of endogenous retroviruses that has polymorphic members; a locus containing a HERV-K can be occupied in one individual but empty in others ., HERV-Ks could contribute to disease risk or pathogenesis but linking one of the known polymorphic HERV-K to a specific disease has been difficult ., We develop an easy to use method that reveals the considerable variation existing among global populations in the prevalence of individual and co-occurring polymorphic HERV-K , and in the number of HERV-K that any individual has in their genome ., Our study provides a reference of diversity for the currently known polymorphic HERV-K in global populations and tools needed to determine the profile of all known polymorphic HERV-K in the genome of any patient population .

linear discriminant analysis, statistics, human genomics, data mining, genomic databases, mathematics, genome analysis, mammalian genomics, information technology, research and analysis methods, sequence analysis, computer and information sciences, bioinformatics, mathematical and statistical techniques, biological databases, statistical methods, animal genomics, sequence databases, data visualization, database and informatics methods, genetics, biology and life sciences, physical sciences, genomics, computational biology

journal.pgen.1005486

A Large-Scale Functional Analysis of Putative Target Genes of Mating-Type Loci Provides Insight into the Regulation of Sexual Development of the Cereal Pathogen Fusarium graminearum

Fusarium graminearum , a homothallic ( self-fertile ) ascomycetous fungus , causes serious diseases ( e . g . , Fusarium head blight ) in major cereal crops , and produces several mycotoxins in diseased cereals 1 ., Recently , this species was defined as a member of the F . graminearum species complex , which consists of more than 16 phylogenetically distinct species found worldwide 2–7 ., To complete the recurrent cycle of cereal diseases , F . graminearum produces sexual progeny ( ascospores ) on cereal debris as overwintering propagules 8 ., The sexual reproduction of F . graminearum is controlled by master regulators called mating-type ( MAT ) loci 9 , 10 ., Unlike their heterothallic relatives , F . graminearum carries two closely linked MAT loci ( MAT1-1 , MAT1-2 ) ., A single nucleus contains individual MAT genes in a structural organization ( MAT1-1-1 , MAT1-1-2 , MAT1-1-3 at the MAT1-1 locus; MAT1-2-1 at the MAT1-2 locus ) similar to that of other Sordariomycetes ( e . g . , Neurospora crassa , Podospora anserina , Sordaria macrospora ) 9 , 10 ., All of the MAT genes encode transcription factors that carry conserved DNA-binding motifs called an alpha box ( MAT1-1-1 ) , an HMG-box domain ( MAT1-1-3 , MAT1-2-1 ) , and a PHP domain ( MAT1-1-2 ) 9 , 10 ., The importance of individual MAT transcripts and MAT loci for sexual development has been intensively studied in F . graminearum , but their functional requirement is not conserved among other fungal species ., All of the four individual MAT genes at the MAT loci are essential for sexual development in F . graminearum 11–14 , whereas SmtA-1 and SmtA-3 ( comparable to MAT1-1-1 and MAT1-1-3 , respectively ) are dispensable for fruiting body ( perithecium ) formation in homothallic S . macrospora 15 ., In heterothallic species , MAT1-1-2 is essential for perithecium formation in P . anserina , but has a redundant function together with MAT1-1-3 in N . crassa 15 ., The phenotypic changes caused by MAT deletions and gene expression patterns in F . graminearum strongly suggest that MAT genes are involved in both the early and late stages of sexual development 13 , 14 ., In contrast , the prominent roles of MAT genes in heterothallic species are to maintain the sexual identity of cells that express the opposite MAT gene for mating ( i . e . , controlling sexual compatibility ) and to regulate pheromone-mediated signaling pathways ., Recently , an additional transcript ( MAT1-2-3 ) with no DNA-binding motif was identified in the MAT1-2 locus 16 , but its role ( s ) in sexual development is not essential in F . graminearum 14 ., MAT transcriptional factors may control the transcriptional expression of downstream genes that are necessary for sexual development in filamentous fungi ., Several transcriptional profiling analyses have been performed to identify MAT loci target genes that are differentially expressed in fungal strains lacking MAT genes during sexual development 17–20 ., However , the function and related regulatory pathways of MAT-target genes have not been sufficiently elucidated to allow a comprehensive understanding of sexual development under the control of the MAT loci; only homology-based functional categorization and limited information regarding gene function ( such as pheromone/receptor genes ) are available ., Very recently , putative target genes of a fungal mating-type gene ( MAT1-1-1 ) carrying a DNA binding alpha box domain were identified by a genome-wide search using chromatin immunoprecipitation combined with next-generation sequencing ( ChIP-seq ) in Penicillium chrysogenum , but only a limited number of target genes were functionally characterized 21 ., In F . graminearum , genome-wide transcriptional analyses during perithecium development have been also performed using microarrays 22 and RNA-sequencing ( RNA-seq ) technology 23 , but the functions of most of the highly expressed genes remain unclear ., Despite intensive investigation of MAT genes in filamentous fungi , many questions regarding sexual developmental processes regulated by MAT genes remain unanswered ., Various cellular and developmental events occur during sexual reproduction in ascomycetes: ascogonium formation , fertilization , nuclear migration and proliferation in ascogonium , nuclear recognition and fusion in dikaryotic hyphae , meiosis , and ascus/ascospore formation ., However , little is known about the specific roles of MAT genes in these sexual stages , particularly those after fertilization , although pheromone/receptor-mediated fertilization under control of MAT is well-established in heterothallic species ., In homothallic species , the function of MAT loci that are present within a single nucleus are less known compared to those in heterothallic species; even the mechanism by which MAT controls the mating process remains unclear ., Homothallic F . graminearum is an ideal species for exploring these unanswered questions for several reasons described below ., The presence of both MAT1-1 and MAT1-2 loci in a single nucleus provides a good model system for investigating the roles of both loci after fertilization ( e . g . , nuclear fusion , meiosis , perithecium maturation ) , which requires two parental strains of the opposite mating types in heterothallic species ., The capacity of F . graminearum to outcross and self-cross 11 suggests that it has gene regulatory mechanisms for sexual development that are identical to those of heterothallic ascomycetes ., Unlike S . macrospora , F . graminearum , requires all of the transcripts at both MAT loci for sexual development 14 , which makes the effects of MAT deletions on the expression and function of MAT target genes more evident ., In addition , F . graminearum can be molecularly manipulated to allow high-throughput gene deletions 24 and genetic analyses 11 ., Finally , the production of sexual progeny is ecologically important for disease development by F . graminearum , because its sexual cycle predominates in the field , which makes the current study significant both practically and fundamentally ., To explore the regulatory mechanisms controlled by MAT genes in F . graminearum , we performed a large-scale study of the target genes of two MAT loci using several strategies including genome-wide transcriptional profiling in various genetic backgrounds , protein binding microarray analysis , in-depth quantitative real-time PCR , and high-throughput gene deletions ., The results of this study combined with previous reports provide an insight that allows a comprehensive understanding of the sexual developmental processes under the control of the MAT loci in F . graminearum ., For the microarrays , we used four transgenic strains that were derived from the self-fertile wild-type ( WT ) strain ( Z3643 ) ., Three strains , designated ΔMAT1-1 , ΔMAT1-2 , and ΔMAT1-1;ΔMAT1-2 , contained different deletions of the two MAT loci ( MAT1-1 , MAT1-2 ) , and one strain ( OM2 ) overexpressed the MAT1-2-1 allele ( for details , see S1 Text and S1–S3 Figs ) ., To identify genes that were regulated by the MAT loci during sexual development , genome-wide microarray analysis was performed using total RNA extracted from mycelia and/or perithecial initials of three MAT-deletion strains , OM2 , and their WT progenitor ( Z3643 ) ., Analysis of the transcriptional profiles revealed a total of 1 , 245 genes that were differentially regulated by ≥ 2-fold in all of the transgenic strains compared to Z3643 ., Among these , 1 , 106 ( 647 downregulated , 459 upregulated ) were differentially regulated in the three MAT-deletion strains ( ΔMAT1-1 , ΔMAT1-2 , ΔMAT1-1;ΔMAT1-2 ) , and 187 ( 177 downregulated , 10 upregulated ) were in OM2 ( Fig 1 , S1 Table ) ., All of the DEGs identified in the three MAT-deletion strains could be categorized into 14 groups according to their expression patterns in each MAT-deletion background ( Fig 1 , S1 Table ) ., Of the 647 genes that were downregulated compared to WT , 522 ( 80 . 7% ) were in either or both ΔMAT1-1 and ΔMAT1-2 , but not in ΔMAT1-1;ΔMAT1-2 ., Among these , 337 were downregulated in both ΔMAT1-1 and ΔMAT1-2 , but were not significantly changed in ΔMAT1-1;ΔMAT1-2 ( designated DDN , where the first D means downregulated in ΔMAT1-1 , the second D is for the downregulation in ΔMAT1-2 , and N means no change in ΔMAT1-1;ΔMAT1-2 ) , 117 were downregulated in only ΔMAT1-1 ( DNN ) , and 68 were downregulated in only ΔMAT1-2 ( NDN ) ., The remaining 125 genes were downregulated in ΔMAT1-1;ΔMAT1-2 , regardless of the differential regulation in either ΔMAT1-1 or ΔMAT1-2 , among which 98 ( 78 . 4% ) were DDD-type ( Fig 1 ) ., Among the upregulated genes , most ( 97 . 2% ) were present in either or both ΔMAT1-1 and ΔMAT1-2 , but not in ΔMAT1-1;ΔMAT1-2; 220 , 105 , and 121 genes were UUN- , UNN- , and NUN-type , respectively ( Fig 1 ) ., Most of the DEGs identified in OM2 were downregulated , among which 15 were also downregulated and 30 were also upregulated in the MAT-deletion strains ( S1 Table ) ., Four of the ten genes upregulated in OM2 were also downregulated in the MAT-deletion strains ( S1 Table ) ., MAT1-2-1 was NDD-type and was upregulated in OM2 , and two MAT1-1 transcripts ( MAT1-1-2 , MAT1-1-3 ) were the DND-type , and were unchanged in OM2 ., In addition , 729 DEGs ( 325 downregulated , 404 upregulated ) were identified in the two MAT deletion strains ( ΔMAT1-1 and ΔMAT1-2 ) compared to the MAT null strain ( ΔMAT1-1;ΔMAT1-2 ) ( S4 Fig , S2 Table ) ., In total , 87 of the 101 NNU-type genes ( unchanged in ΔMAT1-1 or ΔMAT1-2 , but upregulated in WT compared to ΔMAT1-1;ΔMAT1-2 ) overlapped with the DDD-type genes that were identified in comparison with WT ( S2 Table ) ., The GO analysis revealed that several Biological Process categories were enriched among the DDN- , DNN- , and NDN-type genes , including various types of metabolism , and developmental processes ., Among the genes ( UNN , UUN , NUN ) upregulated in the MAT-deletion strains were enriched the categories of carbohydrate metabolism , developmental processes involved in sporulation , cellular response to chemical stimuli , and cell wall organization ., Most of the cellular components categories enriched among these groups were fungal cell wall , plasma membranes , and extracellular ( S3 Table ) ., The DDD-type genes were poorly matched to the GO-terms; only those involved in developmental processes ( e . g . , regulation of cell morphogenesis and response to stimuli ) , and organic hydroxyl compound metabolism ( including the polyketide biosynthesis for perithecial pigment ) categories were enriched in this group ( S3 Table ) ., Among the genes that were downregulated in OM2 , the categories of metabolism for lipid and nitrogen compounds were enriched ., ( S3 Table ) ., In addition to the genes enriched for GO terms , 21 genes that might be involved in the cellular processes ( e . g . , cell fusion , nuclear fusion , cell division , chromosome partitioning ) required for sexual development were identified among the DEGs ( S4 Table ) ; the expression of most of these was unchanged in ΔMAT1-1;ΔMAT1-2 ( XXN ) ., A total of 58 DEGs identified in this study were analyzed using quantitative real-time PCR in MAT-deletion strains at the perithecial induction stage to confirm that their differential expression was caused by each MAT deletion ., The expression patterns of 42 of the 58 genes compared in the three MAT-deletion strains and WT were consistent with the microarray results ( S5 Table ) ., The other 16 genes were also differentially expressed , but their patterns in one of three MAT-deletion strains were not consistent with the qPCR data ., Among these , eight genes that were identified as DDN-type in the microarrays , exhibited expression that was downregulated almost two-fold in ΔMAT1-1;ΔMAT1-2 compared to WT , and were confirmed as DDD-type using qPCR ., Previous studies confirmed that an additional eight genes were downregulated in either ΔMAT1-2 or both ΔMAT1-1 and ΔMAT1-2 strains by Northern blot analysis 17 , 25 ., A total of 50 transcription factor ( TF ) genes ( 6 . 9% of the total TFs in the F . graminearum genome ) 24 were identified among the DEGs in the current study ( Fig 2 ) ., TF genes with a specific and essential function for sexual development in F . graminearum , which were identified based on phenotypic changes after gene deletions 24 , were only enriched among gene groups that were downregulated in ΔMAT1-1;ΔMAT1-2 ( DND , NDD , NND , DDD ) ., All of the sexual development-specific TFs , other than the MAT genes themselves , were DDD-type ., The TFs belonging to other gene groups , whose expression levels were unchanged in ΔMAT1-1;ΔMAT1-2 ( DDN , DNN , NDN ) , were either dispensable for sexual development ( ten TFs ) , involved in pleiotropic phenotypes ( i . e . involved in sexual development and other traits; four TFs ) , or involved in traits other than sexual development ( two TFs; Fig 2 ) ., In contrast , none of the TFs that were specific to sexual development were identified among the 20 TFs upregulated in the MAT-deletion strains; only 1 TF gene deletion proved lethal ., Among the four TFs downregulated in OM2 , only one TF ( FGSG_00404 ) was sexual-specific in the Z3639 strain in a previous study 24 , but was not in Z3643 in the current study ., The remaining TFs were dispensable or were involved in sexual development along with other trait ( zearalenone production ) ( FGSG_07368 ) ( Fig 2 ) ., Based on the phenotypic changes by gene deletions 23 , the F . graminearum locus IDs ( FGSG_ ) on a gray background indicate the genes specific in function to sexual development , IDs in bold and underline are for those involved in sexual development and other traits , IDs in bold are for those are involved in the traits other than sexual development , and underlined IDs are for those probably lethal ., We focused on the expression profiling of genes involved in secondary metabolism since it has been known that secondary metabolism and sexual development are linked in filamentous fungi ., To identify SM genes among the DEGs , they were compared against members of the 67 tentative SM gene clusters in F . graminearum 26 ., Gene member ( s ) belonging to 22 SM clusters were identified in the DEGs from the MAT-deletion strains or OM2 ( S6 Table ) ; however , only 6 SM clusters included DEGs that encoded key ( signature ) enzymes ., Among these , members of two polyketide synthase ( PKS ) gene clusters were downregulated in the MAT-deletion strains: PKS3 ( along with four additional genes ) , which is responsible for the biosynthesis of dark perithecial pigment , and PKS7 ( with one tailoring gene ) , whose chemical product has not yet been identified; these were DDD- and DDN-type , respectively ., Two non-ribosomal peptide synthetase genes ( NPS10 , a NPS-like gene ) for unknown metabolites were also identified ., In addition to these key enzyme genes , those that encoded either tailoring enzymes or transporters belonging to other PKS clusters ( PKS2 , PKS14 , PKS15 , PKS17 for unknown polyketide compounds ) , an NPS1 cluster for a siderophore ( malonichrome ) 27 , and a butenolide cluster were identified among the DEGs ( S6 Table ) ., A total of 169 DEGs ( 13 . 6% ) identified in this study overlapped with the 2 , 064 genes identified previously as sexual development-specific in the F . graminearum PH-1 strain , and whose transcripts were only detected during perithecium formation 22 ., The genes downregulated in ΔMAT1-1;ΔMAT1-2 ( i . e . NND- , NDD- , DND- , DDD-type ) overlapped at a higher frequency ( 43 of 125; 34 . 4% ) than those downregulated in both or either ΔMAT1-1 and ΔMAT1-2 ( DDN , DNN , NDN ) ( 76 of 522; 14 . 6% ) ( S7 Table ) ., The DEGs identified in this study were also compared to those from the F . graminearum strains lacking FgVelB or GzGPA1 , both of which are self-sterile 28 , 29 ., More than half of the genes ( 378 of 647; 58 . 4% ) that were downregulated in the MAT-deletion strains overlapped with those in F . graminearum ΔFgVelB ( Fig 3 , S8 Table ) ., In addition , 121 genes ( 18 . 7% ) downregulated in the MAT-deletion strains overlapped with those in the ΔGzGPA1 strain , among which 100 ( 82 . 6% ) were also DEGs in the ΔFgVelB strain ( Fig 3 , S8 Table ) ., Interestingly three MAT genes MAT1-1-3 ( FGSG_08890 ) , MAT1-1-2 ( FGSG_08891 ) , and MAT1-2-1 ( FGSG_08893 ) , and one MAT gene MAT1-1-3 ( FGSG_08890 ) were downregulated in the ΔFgVelB 28 and ΔGzGPA1 29 strains , respectively ., Surprisingly , only a small number of DEGs overlapped with DEGs in S . macrospora strains lacking MAT genes ., Specifically , 19 DEGs overlapped with 311 genes regulated exclusively in ΔSmtA-2 ( ΔMAT1-1-2 ) , 26 DEGs corresponded to 520 genes regulated in both ΔSmtA-1 and ΔSmtA-2 15 , and 6 DEGs overlapped with 80 genes from ΔSmta-1 ( ΔMAT1-2 ) 19 ( S9 Table ) ., To assess how the MAT loci regulate the expression of genes encoding pheromones ( GzPPG1 , GzPPG2 ) and their cognate receptors ( GzPRE1 , GzPRE2 ) during the early stage of perithecial induction ( 3 days after the removal of the aerial mycelia on carrot agar ) , qPCR was used to compare the transcript levels of each gene in the fungal strains used in microarray analysis , as well as in those lacking individual genes ( MAT1-1-1 , MAT1-1-2 , MAT1-1-3 ) in the MAT1-1 locus ( Table 1 ) ., Because northern blotting previously confirmed that all four genes were only expressed in the WT strain during sexual development 30 , we used the transcript level of each gene in WT , or the weakest expression level in GzPRE1 as references to evaluate the effects of MAT deletion or overexpression ( Table 1 ) ., The expression of GzPPG1 was significantly reduced in all of the MAT-deletion strains examined compared to WT , with the exception of ΔMAT1-1-2 ., Because of the relatively high abundance of the GzPPG1 transcript compared to other genes in WT , this suggests that GzPPG1 is highly expressed only in the WT and ΔMAT1-1-2 strains , but not in a MAT-locus-specific manner ., In contrast , GzPPG2 expression was reduced in the ΔMAT1-2 and MAT-null strains , but increased dramatically in the strain lacking the entire MAT1-1 locus ( ΔMAT1-1 ) , and that lacking only the MAT1-1-1 gene at the MAT1-1 locus ., This suggests that GzPPG2 is only expressed in the WT and ΔMAT1-1 strains , and therefore exhibits a MAT1-2-locus-specific expression pattern , consistent with our previous study 30 ., GzPRE1 was downregulated in both the ΔMAT1-1 and MAT-null strains , but was expressed at comparable levels in the ΔMAT1-2 and WT strains , suggesting a MAT1-1-locus-specific expression ., However , the upregulation of GzPRE1 in strains lacking individual MAT1-1 transcripts was surprising , although most of the upregulated transcript were expressed at levels that were weaker than or similar to GzPRE2 in WT ., In contrast , GzPRE2 was upregulated in all of the MAT-deletion strains except for ΔMAT1-2 , suggesting that GzPRE2 was constitutively expressed in all of the strains examined ., The effects of ΔMAT1-1-1 on the expression of genes encoding pheromones and their receptors were more dramatic than those of other MAT1-1 gene deletions ( ΔMAT1-1-2 , ΔMAT1-1-3 ) , with the exception of GzPRE1 , suggesting that MAT1-1-1 is the major regulator of the pheromone/receptor system among the three transcripts at the MAT1-1 locus ., In addition , the expression of GzPPG2 , GzPRE1 , and GzPRE2 was upregulated in the OM2 strain ( Table 1 ) ., Recently , similar gene expression data for pheromone/receptor genes in the MAT-deletion strains were reported by Zheng et al 13 ., However , those data cannot be directly compared with those in the current study because they were obtained using RNA samples from aerial hyphae on carrot agar before perithecial induction 13 ., To determine the functional requirement of the DEGs identified in the current study during sexual development , we selected 106 DEGs based on their expression patterns and putative functional roles ., Then , we deleted each DEG from the F . graminearum Z3643 genome using a targeted gene replacement strategy ., Including the 32 DEGs that overlapped with those previously identified as being functionally required for or transcriptionally specific for sexual development and/or other traits ( e . g . , hyphal growth , toxin production , virulence ) in F . graminearum , the results of the functional analysis of a total of 127 DEGs were reported here ( S10 Table , S11 Table , Table 2 ) ., Based on the phenotypes of the gene deletion strains , 40 genes were responsible for phenotypic changes ., Among these , 37 were involved in sexual development alone or together with other traits , and the remaining three were required for phenotypes other than sexual development ( S10 Table ) ., Of the 37 genes involved in sexual development , 25 genes were specific to sexual development ( Table 2 ) ., The phenotypic changes caused by the deletion of these genes were restricted to only sexual developmental processes ranging from the formation of perithecium initials to ascospore discharge; no changes in other traits such as hyphal growth and pigmentation , conidiation , mycotoxin production , and/or virulence were observed ., The transgenic strains in which five genes had been individually deleted ( FGSG_00404 , 04480 , 05239 , 13708 , and 03916 ) produced no perithecium initials on carrot agar , and those in which each of nine genes were deleted ( FGSG_01366 , 08320 , 11826 , 13162 , 10742 , 08890 MAT1-1-3 , 08891 MAT1-1-2 , 08892 MAT1-1-1 , 08893 MAT1-2-1 ) produced barren perithecia that were smaller in size and/or number than WT and contained no asci/ascospores ( S11 Table , Table 2 , S5 Fig ) ., By contrast , the remaining 11 genes were not absolutely required for the production of fertile perithecia , but instead , were specifically involved in sexual development ., The mutants in which six genes ( FGSG_03673 , 05151 , 07578 , 06059 , 11962 , and 02655 GzPRE2 ) were deleted individually produced lower numbers of mature perithecia , whereas those lacking FGSG_06966 and FGSG_01862 produced larger perithecia , or showed delayed perithecia formation , respectively , compared to WT ., The deletion mutants of FGSG_00348 , FGSG_02052 , and FGSG_09182 ( PKS3 ) produced perithecia that looked similar to those in WT , but exhibited defects at different stages of perithecia maturation ( Table 2 , S5 Fig ) ., The targeted deletion of FGSG_00348 ( designated FgSMS-2 ) from Z3643 , which exhibited sequence similarity to a gene encoding an Argonaute protein known to participate in the RNA interference ( RNAi ) pathway in Drosophila melanogaster 31 and N . crassa 32 , caused no dramatic changes in major traits such as hyphal growth , conidiation , pigmentation , virulence , and perithecia formation in F . graminearum ., Unlike the perithecia produced in WT , those in ΔFgSMS-2 produced no cirrhi ( ascospores oozing from the perithecia ) at the ostiole 10 days after perithecial induction , and contained fewer numbers of asci that were formed at least 2 days later than WT ( Fig 4 ) , as previously reported in the Z3639 strain 33 ., The germination rate of ascospores was not significantly different from WT ., Interestingly , outcrossing the ΔFgSMS-2 strain as a male to the ΔMAT1-2 strain as a female produced incomplete tetrads that mainly carried four ascospores rather than eight , in which the GFP marker did not segregate equally ( Fig 5 ) ., Furthermore , the outcross between the ΔFgSMS-2 strain ( female ) and ΔMAT1-2 strain ( male ) produced asci similar to those in the self-cross of ΔFgSMS-2 strain ( Fig 5 ) ., qPCR confirmed that FgSMS-2 was specifically expressed at a later stage ( 7 days after perithecial induction ) of sexual development , and was regulated transcriptionally by both MAT loci ( S6 Fig ) ., In addition , the expression of three DEGs ( FGSG_02877 and belonging to UUN , and 05906 to DDN ) was upregulated during sexual development in the ΔFgSMS-2 strain compared to the WT strain , suggesting that FgSMS-2 is involved in the degradation of the mRNAs of these DEGs ( S6 Fig ) ., The deletion of another Argonaute-like gene ( FGSG_08752 ) in the F . graminearum genome , which was not differentially regulated by the MAT loci , had no effect on sexual development; even the double deletion of FgSMS-2 and FGSG_08752 yielded an identical phenotype as the ΔFgSMS-2 strain ( Fig 4 ) ., Unlike ΔFgSMS-2 , the strain lacking FGSG_02052 produced cirrhi at least 2 days earlier than WT , but it produced as many normal-looking ascospores as WT ( Fig 6 ) ., Among the 25 genes with sexual development-specific functions , ten ( 40%; FGSG_04480 , 00404 , 01366 , 11826 , 05151 , 06966 , 08890 , 08891 , 08892 , 08893 ) , including four MAT genes , encode transcription factors , two ( FGSG_08320 and 09182 ) encode SM gene cluster members , and the others are involved in metabolism ( FGSG_13708 , FGSG_03673 , FGSG_07578 , FGSG_06059 ) , chromatin silencing ( FGSG_13162 ) , cell adhesion ( FGSG_03916 ) , signaling ( FGSG_05239 ) , cytoskeleton dynamics ( FGSG_01862 ) , RNA inference ( FGSG_00348 ) , or unknown functions ( FGSG_11962 and FGSG_02052 ) ( Table 2 ) ., Interestingly , 21 ( 84% ) of the 25 sexual-specific genes were downregulated in ΔMAT1-1;ΔMAT1-2 ., Only 1 gene ( FGSG_05239 ) among the 31 DDN-type genes examined had a sexual-specific function ( Table 2 ) ., In addition , 12 genes ( FGSG_00532 , 02572 , 06039 , 06228 , 07368 , 07546 MYT2 , 07869 , 08795 , 09019 , 09834 , 09896 GzICL1 , 10825 were responsible for pleiotropic phenotypes , including defects in sexual development ., Three genes , which were not essential for sexual development , were involved in hyphal growth ( FGSG_04946 ) or virulence toward the host plant ( FGSG_05906 FGL1 , FGSG_10396 ) ( Table 2 ) ., A total of 37 DEGs identified in the current study were analyzed in the WT strain grown on carrot agar using qPCR to determine the time course of the transcriptional profiles during both the vegetative and sexual stages ., Most DEGs examined in the current study ( 33 of 38 ) were confirmed as sexual development-specific at the transcriptional level , because they were expressed at higher or lower ( for FGSG_05906 only ) levels in the WT Z3643 strain under perithecial induction conditions compared to vegetative conditions ( S12 Table ) ., Among these , four genes ( FGSG_ 05246 , FGSG_05847 , FGSG_06549 , FGSG_01763 ) could be assumed to be involved in the early stage of perithecium formation , because their expression peaked 3 days after perithecial induction ., In contrast , the remaining 28 genes are probably specific to a later stage of sexual development ( S12 Table ) ., However , the functional assignment of these genes relies on only a correlation , and needs further confirmation ., In our previous study 28 , 42 genes that were identified as DEGs in the current study were confirmed to be sexual development-specific in the WT Z3639 strain ., Among these , 50% ( 21 genes ) and 28 . 6% were DDD- and DDN-type , respectively ( S8 Table ) ., When the DEGs in the current study were searched against the transcriptome data obtained from the fungal PH-1 strain at 6 developmental stages during perithecium formation , data revealed that 1 , 152 DEGs were expressed at these time points 23 ( S13 Table ) ., In particular , the transcript accumulation of 72% ( 87/121 ) of the XXD-type genes ( including DDD , DND , NDD , NND ) peaked 96 h after perithecial induction , whereas the expression of 70% of the XXN-type ( DDN , DNN , NDN ) genes peaked at earlier time points ( 2 , 24 , 48 , or 72 h; S13 Table ) ., However , the upregulated genes were not enriched at a specific stage ., In addition , 45% ( 54/120 ) of the genes that were downregulated in OM2 exhibited the highest transcript accumulation at 72 h ( Table 3 ) ., We used PBM technology 34 , 35 to identify a putative binding site for the MAT1-2-1 protein ( S7 and S8 Figs ) ., Two PBMs ( Q9-PBM , FgPBM ) were hybridized to the DNA-binding HMG motif of MAT1-2-1 , which was fused to DsRed fluorescent protein , and then expressed in E . coli ., A comparison of the putative consensus binding sequences identified using both PBM methods indicated that ATTGTT could be the core binding sequence for the HMG domain of MAT1-2-1 ( Fig 7 ) , which is complementary to the core-binding element ( AACAAT ) of the mammalian sex-determining region Y ( SRY ) or SRY-related HMG box gene ( SOX ) 36–38 ., Electrophoretic mobility shift assay ( EMSA ) revealed that the quadruple sequences of the identified motifs ( ATTAAT and ATTGTT ) had binding activities for the MAT1-2-1 HMG box domain ( S9A Fig ) ., The promoter regions of three genes ( FGSG_04946 , FGSG_08467 and FGSG_06480 ) could bind to MAT1-2-1 , but their binding activities were relatively weak ( S9B Fig ) ., Experimental details and discussion from the PBM and EMSA analyses were described in S2 Text ., To assess the possible regulatory relationship among the sexual development-specific MAT downstream regulator genes ( four transcription factors and one RNAi regulator ) , we used qPCR to examine their expression patterns in fungal strains in which each gene was deleted during sexual development ( 3 and 6 days after perithecial induction ) ( Table 4 ) ., Using a fold-change threshold of 3 . 0 , because most genes were expressed at relatively low levels ( based on previous transcriptome data 23 ) , a map of the regulatory interactions among the genes was constructed , as previously performed 39 ., Because the core binding sequence of MAT1-2-1 was found in only FGSG_01366 , the transcription factor carrying the HMG-box motif , we assumed that FGSG_01366 is the first putative target of MAT1-2-1 in the network ( Fig 8 ) ., However , the deletion of FGSG_06966 , and FGSG_11826 had a significant effect on the expression of FGSG_01366 and other regulatory genes , suggesting that interregulatory networks operate among these genes ., Interestingly FGSG_00348 , which encodes an Argonaute-like protein was downregulated in strains in which all four transcription factor genes had been individually deleted , suggesting that it was a downstream target of these genes ., In addition , none of the target gene deletions had a significant effect on the transcription of MAT1-1-1 and MAT1-2-1 , which confirms that these regulatory genes are downstream of the MAT loci in F . graminearum ., The most significant achievement in this study is that it provided a comprehensive investigation of the putative target genes of the MAT loci during sexual development in self-fertile F . graminearum ., Genome-wide transcriptional profiling in various MAT genetic backgrounds , and subsequent in-depth and high-throughput analyses allowed us to explore the regulatory networks and function of MAT-target genes during the early sexual developmental process when the major regulators ( MAT1-1-1 , MAT1-2-1 ) of each MAT locus are expressed at their peak levels 14 ., Similar to other filamentous ascomycetes , F . graminearum undergoes various cellular processes during each stage of sexual development including mating , cell fusion , nuclear division , fusion , meiosis , ascus/ascospore development , and perithecium maturation ., These aspects of sexual development could include pheromone-mediated membrane function , signal transduction , cytoskeleton dynamics , secretory pathways , cell cycle , cell adhesion , apoptosis , and differentiation 40 ., The GO terms associated with genes that are differentially expressed in strains carrying a single MAT gene or no MAT gene were significantly enriched for terms related to sexual development processes ., In particular , the terms of metabolism including cell wall organization , developmental processes involved in reproduction , and the cellular response to chemical stimulus were enriched among the DDN- and UUN- type DEGs , which are the most frequent groups ., Similarly , those described above as well as related to signaling , cellular homeostasis , and cell cycle were enriched among the DEGs that were found in only ΔMAT1-1 or ΔMAT1-2 ( DNN- , U

Introduction, Results, Discussion, Materials and Methods

Fusarium graminearum , the causal agent of Fusarium head blight in cereal crops , produces sexual progeny ( ascospore ) as an important overwintering and dissemination strategy for completing the disease cycle ., This homothallic ascomycetous species does not require a partner for sexual mating; instead , it carries two opposite mating-type ( MAT ) loci in a single nucleus to control sexual development ., To gain a comprehensive understanding of the regulation of sexual development in F . graminearum , we used in-depth and high-throughput analyses to examine the target genes controlled transcriptionally by two-linked MAT loci ( MAT1-1 , MAT1-2 ) ., We hybridized a genome-wide microarray with total RNAs from F . graminearum mutants that lacked each MAT locus individually or together , and overexpressed MAT1-2-1 , as well as their wild-type progenitor , at an early stage of sexual development ., A comparison of the gene expression levels revealed a total of 1 , 245 differentially expressed genes ( DEGs ) among all of the mutants examined ., Among these , genes involved in metabolism , cell wall organization , cellular response to stimuli , cell adhesion , fertilization , development , chromatin silencing , and signal transduction , were significantly enriched ., Protein binding microarray analysis revealed the presence of putative core DNA binding sequences ( ATTAAT or ATTGTT ) for the HMG ( high mobility group ) -box motif in the MAT1-2-1 protein ., Targeted deletion of 106 DEGs revealed 25 genes that were specifically required for sexual development , most of which were regulated transcriptionally by both the MAT1-1 and MAT1-2 loci ., Taken together with the expression patterns of key target genes , we propose a regulatory pathway for MAT-mediated sexual development , in which both MAT loci may be activated by several environmental cues via chromatin remodeling and/or signaling pathways , and then control the expression of at least 1 , 245 target genes during sexual development via regulatory cascades and/or networks involving several downstream transcription factors and a putative RNA interference pathway .

The production of sexual propagules via a self-fertile mating strategy in Fusarium graminearum , an important cereal pathogen , is essential for overwintering and dissemination during the recurrent disease cycle caused by this fungus ., Genome-wide microarray analyses allow the identification of gene sets that are regulated by the mating-type ( MAT ) loci , which is a master regulator of sexual reproduction in F . graminearum ., By employing in-depth and high-throughput functional analyses , the current study provides novel insight into our understanding of the regulation of sexual developmental processes by the MAT loci ., MAT genes , which are located at two linked MAT loci , play important roles in even the late stages of sexual development by controlling regulatory pathways involving several sexual-specific transcription factors and putative RNA interference regulators ., This study could be significant both practically and fundamentally because of the ecological impact of sexual reproduction by F . graminearum during disease development in the field .

journal.pgen.1004338

Ras-Mediated Deregulation of the Circadian Clock in Cancer

All mammalian cells hold an internal circadian clock able to generate daily-endogenous rhythms with a period of approximately 24 hours ., Circadian clocks are evolutionary conserved and regulate the expression of about 10% of all genes 1–3 ., This time-generating mechanism enables the organism to react to external clues , to anticipate environmental changes and to adapt molecular and behavioural processes to specific day-times with the advantage of separating incompatible metabolic processes ., In mammals , the circadian system is hierarchically organized into two major levels of regulation including a main clock , located within the suprachiasmatic nucleus ( SCN ) and peripheral oscillators 4 , 5 ., The peripheral clocks can be found in almost all cells in the body ., These are able to respond to and synchronize to output signals of the SCN clock , thereby assuring time-precision of molecular processes throughout the organism 6 , 7 ., Interconnected genetic networks of transcriptional and translational steps drive the oscillator in each individual clock within a cell 8 , 9 ., The network system can be represented by a core of two main feedback loops: the RORs/Bmal1/REV-ERBs loop and the PERs/CRYs loop which generate oscillations 10 , 11 ., The core-clock network regulates a series of clock-controlled genes ( CCGs ) with relevant functions in several cellular and biological processes ., CCGs are involved in metabolism , detoxification , cell cycle , cell growth and survival , DNA damage responses and the immune system 12–15 ., Malfunctions of the circadian clock have been reported in the context of many disorders 16–19 ., Epidemiological studies have shown that increasing nocturnal light and overnight shift work coincided with a steady increase in the incidence of cancer 20–22 ., Moreover , the long-term survival rate of patients with metastatic colorectal cancer was about 5-fold higher in those patients with normal clock compared to patients with a severely disrupted clock 23 ., Additionally , accelerated malignant growth was observed in mice with an ablated SCN or subjected to experimental chronic jet-lag 24 ., Hence , the clock regulation of molecular processes has severe consequences on therapy optimization , and timing of drug intake in cancer 25–28 ., Cancer patients with altered circadian rhythms have a poorer prognosis 29 and chronotherapy , the administration of anti-cancer drugs at specific times of the day , can improve treatment efficacy as chemotherapeutics may act differently on their targets depending on the time of administration 25 , ., Furthermore , the rhythmic delivery of cancer therapeutics in colorectal cancer increased the efficacy of oxaliplatin on treatment and patient survival 32 , 33 ., A molecular basis for the effects of circadian rhythms on cancer patients might be provided by the association of several core-clock genes with cancer promoting mechanisms such as DNA damage 34 35 , metabolism 17 and cell cycle 36–38 ., Cell cycle check point regulators such as Wee1 ( G2- M transition ) , Myc ( G0- G1 transition ) , and cyclin D1 ( G1- S transition ) have been shown to be under the direct regulation of the circadian clock and could represent one way in which the circadian clock regulates cell division 3 , 18 , 28 ., The histone deacetylase sirtuin 1 ( SIRT1 ) , a key regulator of metabolism , has recently been identified as a core-clock component 39 , 40 ., The PER1 and Timeless proteins interact with proteins involved in DNA damage response and Per1 overexpression suppresses growth of human cancer cell lines 26 , 41 ., Expression of Per1 and Per2 is downregulated in colon , breast and endometrial carcinoma 41 , 42 ., Per2 is also downregulated in several human lymphoma cell lines and in non-small-cell lung cancer tissues 26 , 43–45 ., In addition , mutations in the clock gene Npas2 , a paralog of Clock , have been associated with increased risk of breast cancer and non-Hodgkins lymphoma 27 , 46 ., Furthermore , mutations in the gene Clock were found in colon cancer cell lines 47 ., These results suggest the existence of a strong cross-regulation between the components of the circadian clock and proto-oncogenes or tumour suppressors ., The circadian clock may act as tumour suppressor , whereas a disturbed clock might render the organism more cancer prone ., However , a comprehensive view of how cancer genes and clock can influence each other is missing ., This motivated us to investigate the mechanism by which the circadian clock might be altered in cancer models ., In the present manuscript , we provide a systems biology approach for the investigation of the circadian clock in several cancer cell lines including colon and skin ., Surprisingly , we found strong and weak circadian oscillators within the same type of cancer ., We recovered a set of genes which allow the discrimination between the two types of oscillators ., Using a theoretical bioinformatics approach , we extended the core-clock network to include a larger set of clock/controlled genes involved in several biological processes and present its interconnection to the discriminative set of genes ., Furthermore , we analysed the connections of such discriminative list of genes to cancer pathways among which is the RAS/MAPK pathway ., Using experimental data and mathematical modelling , we provide evidence for a putative connection between both systems ., Our work provides novel evidence pointing to RAS oncogene being one of the modulators of the mammalian circadian clock ., To investigate a possible link between the circadian core-clock oscillator and tumour-associated pathways , we tested the circadian properties of colorectal cancer cell lines well characterized for their genetic properties and oncogenic pathways ., The impact of the circadian systems is apparent for this cancer type , because chronotherapy has shown promising results in colon cancer patients 32 , 33 ., We studied the oscillation dynamics in the colon carcinoma cell lines HT29 , RKO , SW480 , LIM1215 , CaCo2 and HTC116 using a live-cell imaging approach based on ectopic expression of a luciferase reporter construct driven by the 0 . 9 kb Bmal1 promoter fragment ., As a control we analysed the human osteosarcoma cell line U2OS , a widely used in vitro model to study properties of the mammalian circadian clock ., We define a cell line with a clear circadian period and an amplitude variation of at least 20% as a strong oscillator ., Surprisingly , the clock properties were very diverse among the colon cancer cell lines ( Table 1 ) , which showed strong ( Figure 1B , C ) and weak to no-oscillation phenotypes ( Figure 1D–G ) ., The strongly oscillating cell lines HCT116 and SW480 exhibited shorter doubling times ( unpublished observations ) indicating an association of perturbations of the circadian clock and effects on the cell cycle and in agreement with previous studies 48 ., Moreover , we observed differences in the mRNA levels of the core-clock genes ., In Figure 1H , we plotted the relative fold change to Bmal1 for each clock gene and cell line ., The highest value for Cry1 is observed in a weak oscillator cell line ( CaCo2 ) , while the highest value for Cry2 is observed in the strong oscillator cell lines ., For Per1 and Per2 , the highest fold change is observed in the strong oscillators ., For Clock and Npas2 , weak oscillators show the highest relative change to Bmal1 ., These data clearly show a correlation between expression levels of core-clock genes and the cell oscillator phenotype ., To further investigate potential differences between strong and weak oscillators at the gene expression level , we performed transcriptome analysis for each cell line ., A leave-one-out cross validation strategy identified a set of differentially expressed genes that discriminate between strong and weak oscillators ( Figure 2A ) ., For each cell line we excluded both replica samples once and determined the 100 most significantly expressed probe sets , allowing the identification of two clock groups using a moderated t-test to select the top genes by confidence ., The resulting list of 45 best p-value discriminative genes ( Table 2 and Figure 2A ) allows positioning of U2OS accordingly within the strong oscillator cluster , although this cell line was not previously used for the list generation ., Additionally we tested 5 other colon cancer cell lines ( Colo205 , SW620 , SW403 , HKe3 , HKe-clone8 with/without mifepristone ) and two keratinocyte cell lines ( HaCaT and A5RT3 ) ., Data is shown in Figure S5 and Text S4 ., The predefined list of 45 discriminative genes could be used to correctly classify seven out of eight cell lines ( Text S4 ) ., Using the binomial test , we calculated the probability of observing the same or better classification result by random chance ., According to this experiment our classifier performs significantly better than randomly expected ( p\u200a=\u200a0 . 03516 ) ., Moreover , previous data 49 shows that the siRNA-dependent -knockdown of the majority of these 45 genes confers a circadian phenotype in U2OS cell lines , underlining their potential importance as regulators of the circadian system ., The methodology used to generate the list of 45 discriminative genes , including their p-values , is explained in detail in Text S1 and the resulting p-values for the short list of 45 genes are additionally given in Table 2 ., The composition of the set of top genes indicates that phenotypic circadian clock differences are reflected by gene expression differences both in genes of the core network ( Figure 1H ) , but also in additional genes not directly associated with circadian clock functions ., To explore a potential correlation of the discriminative set of genes to the mammalian circadian clock we extend the known core-clock network , which encompasses several genes and proteins interconnected by positive and negative feedback loops 10 , to a layer of next neighbours ., We scanned a total of 21 . 4 million abstracts by text mining ., After careful curation of the text mining results , we assembled a comprehensive genetic network for the mammalian circadian clock ( Figure 2B ) ., First , we gathered a network containing the currently known core-elements of the circadian pathway including the 14 genes: Per1 , 2 , 3 , Cry1 , 2 , Bmal1 , 2 , Rev-Erbα , β , Rorα , β , γ , Clock and Npas2 ., In the next step , we added 16 elements reported to be directly interacting to this core set 13 ., Subsequently , we used our text mining software GeneView 50 to extract all reported interactions between, ( i ) any two elements of the network and, ( ii ) new elements with direct connections to the 14-element core ., As a validation of the performance of GeneView , we analysed the part of the network containing common elements to our previously published network 13 ., The software provided evidence for 85% of the interactions described ( manuscript in preparation ) ., Additionally , we found 17 new elements in the outer shell and 108 novel interactions , after curation , supported by 132 PubMed references ., The enriched circadian-core network contains a total of 47 elements and 229 interactions ( Figure 2B , Text S2 ) ., This network provides an improved level of specificity regarding interactions of core-clock and clock-controlled genes , in comparison to previously published networks 12 , 13 ., It includes both protein-protein and DNA-protein interactions and selectively assembles elements which are as well reported to be able to influence the core-clock Figure 2B ., To analyse whether and how the network might convey circadian information to specific output pathways , we performed a detailed analysis of the KEGG pathways associated with the different elements as well as of the data set obtained by text mining analysis ( Text S2 ) ., We found that many of the genes in the periphery of the network play important roles in pathways frequently deregulated in cancer ., Examples are the Wnt pathway ( CSNK1ε , CSNK2α , p300 , GSK3β , βTRC ) , the TGF-β signalling pathway ( CBP , p300 , TNFα ) , the Jak-STAT signalling pathway ( CBP , p300 , IFNα ) and the MAPK signalling pathway ( CBP , p300 , GSK3β , PPARγ , TNFα ) ., We also found many genes involved in cell cycle and repair mechanisms ( NONO , PPAR1 , GSK3β ) and in immune defense ( CREB , GSK3β , AMPK , TNFα , PGC-1α ) ., Moreover , many of the new network elements were also found to link the circadian system to metabolism and xenobiotics detoxification mechanisms and as such are also potentially relevant in terms of therapy and drug response ( PPARα , TEF , ALAS1 , AhR , CAR , HLF , E4BP4 , DEC2 , DEC1 , DBP ) ., Altogether , these results clearly support the view that the circadian clock and oncogenic pathways are strongly connected ., To investigate how the identified set of discriminative genes links cancer genes and circadian regulators we again used the text mining software GeneView to create interaction networks between, i ) the core-clock genes ( Figure 2B ) and the 45 discriminative genes ( Table 2 , Text S1 ) ;, ii ) a set of known colon cancer-related genes ( 51 , Table 3 ) and the 45 discriminative genes and, iii ) between all three sets of genes ., Only interactions involving elements of two different gene sets were considered ( Figure 2B , C ) ., These interactions include both protein-protein as well as DNA-protein interactions ., All interactions extracted by the text mining pipeline as well as 391 , 434 interactions contained in the STRING database version 9 . 0 52 were collected ., In total 646 Interactions were found to be involved in the assembly of the network ( Figure S1 ) ., We found 184 connections between our discriminative set of genes and the genes of interest ( clock genes and cancer genes , Figure 2C , D ) ., To test whether the number and counts of interactions point to a specific connection between the discriminative genes and the clock/cancer genes , we now tested the significance of the number of connections by comparing to networks build from random gene sets of the same size ( 45 genes ) ., We created a total of 50 such random sets where each gene of the set was randomly selected from a bin containing genes with the same number of PubMed citations as a gene in the discriminative set ., We obtained an average of 165 interactions , standard deviation 28 yielding a p-value of 4 . 4e-5 when using a non-parametric Wilcoxon-Test ., This indicates that a specific connection between the discriminative genes and the clock/cancer genes exists , which points to the relevance of this gene set in the clock-cancer context ., Furthermore , from the network analysis , we found that 20 out of the 45 discriminative genes were associated with clock genes ( Figure 2C ) , 27 were found to be associated with cancer-related genes ( Figure 2D ) among which 18 intersect with the set of clock genes ( Table S1 ) ., Discriminative genes associated with cancer pathways ( Table 2 ) include IFNGR2 ( involved in the Jak-STAT pathway ) 53 , PITX2 ( TGF-β pathway ) , RFWD2 ( p53 signalling ) , PPARγ ( Wnt pathway ) ., Moreover , several genes are also involved in the MAPK/RAS pathway: LOXL2 , PPARD and CTSB are RAS target genes; Rab6 is a RAS family member; SPARC is also targeted by the RAS pathway and it was also shown to be a key modulator of extra-cellular matrix ( ECM ) remodelling , it affects cell proliferation and differentiation and it was recently reported to downregulate VEGF and thereby suppressing angiogenesis ., In addition , we searched among our genes of interest for circadian properties by evaluating their expression profiles in published microarray data ., We found 83% of the cancer related genes and 24% of the discriminative genes to show measurable oscillations in gene expression in several tissues and cell lines ( Text S3 ) ., Taken together , these discriminative genes are likely relevant for the analysis of possible clock malfunctions in a cancer model as a clear connection of these 45 genes to core-clock genes and to distinct cancer-related genes exists ., The link of the circadian network to distinct cancer related pathways such as RAS/MAPK , Wnt and Jak/STAT led us to investigate the connection of clock regulation and signalling pathways in an experimental model ., As a model pathway , we chose RAS/MAPK signalling , which is one of the most frequently altered signalling pathways in human cancer ., KRAS mutations have a high prevalence in colorectal cancers ., However , the colon cancer cells analysed exhibited a highly variable genetic background and thus introduced an extra level of complexity when used as models for functional analysis and for studying deregulation of the circadian clock ., Therefore , we used a well-established in vitro epithelial model , in which cellular transformation is triggered by the RAS oncogene , that simulates carcinogenesis development , for investigating the influence of RAS transformation on the circadian clock 54 ., Human HaCaT skin keratinocytes ( and their derivatives represent the different steps of malignant epithelial conversion from the immortal state ( HaCaT ) , to benign ( HaCaT I7 , class I tumours ) , advanced ( HaCaT II4 , class II tumours ) and metastatic states ( HaCaT A5RT3 ) ., The different HaCaT lines were lentivirally transduced with the Bmal1-driven luciferase reporter and the oscillation dynamics was monitored for 5 days ., Non-transformed human keratinocytes exhibited an average period of τ\u200a=\u200a23 . 4±0 . 4 hours ( mean ± SEM , n\u200a=\u200a5 ) with a strong oscillation that persisted for several days ( Figure 3A ) ., HaCaT I7 and HaCaT II4 periods were not significantly different compared to the immortalized cell line HaCaT ( τ\u200a=\u200a23 . 18±0 . 19 hours , n\u200a=\u200a5 and τ\u200a=\u200a22 . 96±0 . 3 hours , n\u200a=\u200a3 , respectively ) ., However , the metastatic cell line HaCaT A5RT3 showed a significantly longer period of τ\u200a=\u200a24 . 93±0 . 2 hours ( n\u200a=\u200a5 ) , than that observed in normal HaCaT cells ( p<0 . 05 , Students t-test ) ( Figure 3B ) ., Moreover , HaCaT A5RT3 showed a significantly delayed phase of approximately 1 . 5 hours ( 17 . 9±0 . 15 hours ) compared to HaCaT cells ( 16 . 37±0 . 21 hours; n\u200a=\u200a5 , p<0 . 05 , Students t-test ) ., HaCaT II4 showed the opposite effect , having an advanced phase of approximately 0 . 85 hour significantly earlier than in normal keratinocytes ( 15 . 52±0 . 35 hours; n\u200a=\u200a3 p<0 . 05 , Students t-test ) ( Figure 3C ) ., A larger amplitude was observed in both HaCaT I7 and II4 ( 1 . 49±0 . 04 , and 1 . 52±0 . 05 , respectively ) compared to normal keratinocytes ( 1 . 36±0 . 03 ) or HaCaT A5RT3 ( 1 . 33±0 . 02 ) ( Figure 3A ) ., To rule out possible secondary effects due to cell growth alterations caused by transfection with the BMAL1-promoter-driven luciferase ( BLP ) construct , proliferation of cells was measured by determining increased conductance of monolayers over time , using XCelligence technology ( Roche ) ( unpublished observations ) ., To investigate if the observed HaCaT A5RT3 phenotype is indeed due to a different phase of the oscillator , we carried out a temperature entrainment assay with metastatic HaCaT A5RT3 and immortal HaCaT cells ., This technique has the advantage that the cells do not respond to a single pulse of an external signal , but are entrained to follow an environmental cue ( temperature ) along a period of time ., Bioluminescence measurement for at least 3 days following the entrainment revealed that HRAS transformed human keratinocytes A5RT3 indeed have a different circadian phenotype in comparison to normal keratinocytes ( Figure 3D ) ., The first peak after release of the cells to a constant temperature was used to determine the phase of entrainment ., H-Ras-transformed HaCaT A5RT3 cells showed an advanced phase , of approximately 6 hours ( Figure 3D ) ., These data demonstrated that Ras transformation can induce a phase shift in the in vitro model system which is consistent with the observed period and phase perturbations in the circadian phenotype of A5RT3 cells ., To validate these results in an inducible RAS-system , we took the rat fibroblast cell line , 208F , and two derivative clones: IR2 and IR4 ., 208F cells are preneoplastic rat immortal fibroblasts ., IR2 and IR4 cell lines are clones obtained by stable transfection with the H-Ras oncogene ( G12V ) , which is under the lac regulatory control harbouring a Ras IPTG-inducible promoter 55 ., All cell lines exhibited oscillations with a period of approximately 24 hours ., The effect of the overexpression of H-Ras in rat fibroblasts was a clear phase shift in IPTG-treated cells ( Figure 4 ) ., The levels of RAS and phosphorylated ERK were analysed via Western Blot and are depicted in Figure S6 ., The results obtained with the HaCaT system and the rat fibroblast cell lines could be reproduced with colorectal cancer cells HKe3 and are presented in Figure 5 ., The HKe3 cells are derived from HCT116 colorectal cancer cell lines , but KRAS has been disrupted by genetic recombination ( HKe3; 56 ) ., HKe3 cells exhibit a period of τ\u200a=\u200a25 . 3±0 . 59 hours ., Furthermore , we used the Hke3-clone 8 , in which we introduced a conditional KRASV12 oncogene ., In the absence of KRAS induction Hke3 clone 8 exhibits a period of ( τ\u200a=\u200a25 . 1±0 . 3 hours ) , similar to HKe3 ., Most interestingly , we observe a clear period phenotype upon activation of the KRAS oncogene ( τ\u200a=\u200a37 . 9±0 . 96 hours ) ., This resembles our previous observations in HaCaT and IR2 cells ., With this data we could confirm our previous results and clearly showed that upon induction of RAS a larger period phenotype could be measured ., To investigate the clock specific changes at the gene expression level in human keratinocytes , we performed real-time PCR for the clock genes Cry1 , Bmal1 , Per2 , Rev-Erbα and Clock ., The measurement was started 24 hours after synchronization to avoid influence of immediate early gene response that may not reflect an accurate effect of the oscillator ., We measured the expression of the five clock genes in HaCaT and HaCaT A5RT3 cells over the course of 24 hours and all showed clear oscillations ( Figure 6A ) ., We found marked differences in the levels of Bmal1 , Per2 , Cry1 and Clock mRNAs , while Per1 and Rev-Erbα mRNAs were largely similar in both cell lines ., Consistent with the live-cell oscillation dynamics ( Figure 3 ) , Bmal1 transcript levels oscillate with a phase about 6 hours shorter in HaCaT cells than in HaCaT A5RT3 cells ., While Per2 mRNA levels are strongly reduced in H-Ras transformed HaCaT cells , Cry1 and Clock gene expression is markedly increased in H-Ras transformed keratinocytes compared to the HaCaT cells , where Cry1 gene expression remained at low levels ., Overall , gene expression in both cell lines is divergent , especially for Cry1 , where mRNA levels increased and for Per2 with decreased levels , in H-Ras transformed keratinocytes ., This indicates that oncogenic RAS is likely to directly impinge onto the regulation of the circadian clock , however by yet unknown means ., To unravel potential underlying mechanisms of RAS-mediated clock alterations we used our previously developed mathematical model for the mammalian circadian clock and carried out a control coefficient analysis over all model parameters and analysed the effects on period and magnitude 11 ., We filtered for parameters for which a perturbation could induce antipodal changes in the magnitude of Per and Cry and at the same time an increase in the period ., By perturbing a parameter involved in the transcription of the Cry gene ( kt2 , 11 ) such an effect could be simulated ( Figure 5B ) ., Combinations of several parameters could cause a similar effect ., However , in our mathematical model-system , the parameter described was the only which could , on its own , cause the observed phenotype ., Interestingly , in our model the parameter kt2 ( Cry transcription equation 11 ) regulates BMAL1/CLOCK-mediated transcription on Cry proposing one possible way of how RAS might interfere with the circadian rhythms: weakening BMAL1s role as a transcriptional activator ( perturbation of transcription activation for all clock genes 11 ) ., Figure 7 predicts a gene expression profile in agreement with the experimental data suggesting that RAS activation perturbs the clock possibly via modulating BMAL1/CLOCK transactivation activity ., To test our hypothesis regarding the regulatory effect of RAS on the circadian clock , we first investigated the potential influence of the MAPK/RAS signalling pathway on the circadian phenotype in HaCaT cells ., In our model the activation of RAS/MAPK signalling ( 60% reduction of the parameter which regulates BMAL1-mediated transcription , for each gene ) predicts an increase of the period ( τ\u200a=\u200a24 . 1 hours ) , while inhibition of the RAS/MAPK pathway ( 60% increase in the parameter which regulates BMAL1-mediated transcription , for each gene ) led to a shorter period phenotype ( τ\u200a=\u200a21 . 4 hours ) , as shown by the in silico expression profiles of Bmal1 ( Figure 8A ) ., To test the predicted effect of reducing RAS/MAPK-mediated signalling on the circadian phenotype experimentally , we treated synchronized HaCaT cells with the MEK inhibitor U0126 ( Figure 8B , C ) ., Indeed , U0126-treated cells showed a shorter period compared to vehicle-treated cells ( Figure 8D ) ., In HaCaT A5RT3 , the MAPK pathway seems to impinge on the length of the period ., This can be seen upon the comparison of HaCaT and A5RT3 and now much stronger when we inhibit the MAPK pathway in A5RT3 ( Figure S7 ) ., Together , our theoretical and experimental data indicate that the activity of the RAS/MAPK modulates the circadian period ( and thereby also the entrained phase ) , possibly by influencing the transcriptional activity of the CLOCK/BMAL1 ., In this study , we showed that the clock is perturbed differentially within the same cancer type and observed a rich variety of clock phenotypes ., To explain these effects , we correlated the gene expression profiles with the clock phenotype of the cells ., This analysis revealed a set of genes discriminating between strong and weak oscillator , thus functioning as a clock phenotype predictor ., In fact , this set of genes also allowed the correct classification of the osteosarcoma test cell line U2OS ., Moreover , previous data 49 showed that the siRNA dependent-knockdown of the majority of these 45 genes confers a circadian phenotype in U2OS cell lines , underlining their potential importance as regulators of the circadian system ., To what extend this set of genes exhibits robustness as a clock phenotype predictor beyond our experimental setup is currently unknown ., Of note , Caco2 cells have been described by others as a good oscillators , yet the experimental conditions were different 60 ., Thus , nature of the predictive gene set might undergo some alterations when alternative conditions are used ., The microarray data revealed important clock genes such as Nono ( a known splicing factor and recently described as connecting the clock to the cell cycle via PER 36 ) , Rac ( involved in proliferation ) and CkIIα to be highly expressed in all cell lines ., At the same time , the tumour suppressor gene Per3 appears weakly expressed in all cell lines as expected ., In addition , we identified a characteristic differential expression of several genes encoding epigenetic regulators , signalling molecules and transcriptional regulators not previously connected to circadian rhythm ., CBX7 is an essential component of the polycomb repressive complex 1 ( PRC1 ) involved in the control of histone methylation at tumour suppressor loci such as p16 61 ., CHD , encodes a chromodomain regulator of chromatin remodelling and was lost in approximately 50% of colorectal cancers 62 ., The YY1 interacting chromatin remodelling complex factor INO80 63 and the ubiquitous epigenetic regulator and insulator protein CTCF , are associated with the weak oscillators ., CTCF controls long-range chromatin interactions and functions in the establishment and maintenance of epigenetic signatures ., This renders CTCF a potentially important factor also for controlling circadian genes 64–66 ., A link between the circadian clock , energy metabolism and epigenetic ( re ) programming has been established earlier 67–69 ., Our new observations indicate that epigenetic events involving DNA methylation , histone modification and chromatin remodelling might also induce differential circadian oscillation in tumour cells ., Differential gene expression between strong and weak oscillators was also seen for the signalling molecules LOXL2 , SPARC , CTSB , IFNGR , WASF3 , GNG11 and the metabolism-associated PPARD gene 70 ., High expression of these genes was strongly associated with the strong oscillator phenotype , but low expression with the weak oscillators ., In contrast , high expression of FOXA1 and PDHX were associated with the weak oscillator phenotype ., The RAS pathway target genes LOXL2 , SPARC and CTSB exert functions in the remodelling of the extracellular matrix , thereby favouring invasion and metastasis of tumour cells 71–73 ., WASF3 74 and FOXA1 75 are downstream targets of the anti-apoptotic PI3K signalling pathway and have been shown to control actin polymerisation and invasion as well as differentiation of secretory gut epithelial cells , respectively ., GNG11 , a member of the heterotrimeric G-protein family , is involved in senescence induction via environmental stimuli 76 and has been reported recently to be deregulated in TGFIIR knock-out epithelial cells capable of increased metastasis 77 ., PPARD and PPARB expression is upregulated in colorectal cancer 78 and the gene was found activated by the K-RAS pathways in rat intestinal epithelial cells 79 ., The protein mediates activation of PI3K signalling via PTEN deregulation and can enhance anti-apoptotic signalling and cell survival 70 ., Taken together , these observations show that targets of RAS/MAPK-dependent signalling are associated with a certain oscillator phenotype ., With the exception of LOXL2 , the RAS target genes that play a role in cancer invasion and metastasis are downregulated in the weak oscillators , thus being associated with the bad prognosis phenotype ., In addition , PI3K signalling might play a functional role in clock deregulation as indicated by differential expression of WASF3 and FOXA1 ., Recently it was shown that GSK3β is able to phosphorylate and destabilize CLOCK 80 ., Overexpression of oncogenic RAS and subsequent activation of PI3K/Akt resulted in GSK3β inhibition and in an increased stabilization of CLOCK ., Comparing clock gene expression between immortal and RAS transformed HaCaT cell , we also detected increased levels of Clock mRNA ., To what extent the elevated Clock mRNA levels are due to RAS/MAPK or RAS/PI3K signalling in our cells needs to be investigated ., Furthermore , we correlated the set of discriminative genes found to an extended network of the mammalian circadian clock ., From our assembled network it is evident that the core-clock genes directly regulate a set of 33 clock-associated genes involved in the Wnt , the TGF-β , Jak-STAT and MAPK signalling pathway ., We also found many of the clock-associated genes to be involved in important biological processes such as cell cycle and repair mechanisms , immune defense , metabolism and the xenobiotics detoxification mechanisms ., All these processes and pathways are often found to be deregulated in cancer ., The assembled network established in this study provides a novel extension of the circadian system to output clock-associated genes which can potentially be relevant in terms of drug targets or even in the prognosis of cancer ., We also suggest that a feedback from the clock-associated genes to the c

Introduction, Results, Discussion, Materials and Methods

Circadian rhythms are essential to the temporal regulation of molecular processes in living systems and as such to life itself ., Deregulation of these rhythms leads to failures in biological processes and eventually to the manifestation of pathological phenotypes including cancer ., To address the questions as to what are the elicitors of a disrupted clock in cancer , we applied a systems biology approach to correlate experimental , bioinformatics and modelling data from several cell line models for colorectal and skin cancer ., We found strong and weak circadian oscillators within the same type of cancer and identified a set of genes , which allows the discrimination between the two oscillator-types ., Among those genes are IFNGR2 , PITX2 , RFWD2 , PPARγ , LOXL2 , Rab6 and SPARC , all involved in cancer-related pathways ., Using a bioinformatics approach , we extended the core-clock network and present its interconnection to the discriminative set of genes ., Interestingly , such gene signatures link the clock to oncogenic pathways like the RAS/MAPK pathway ., To investigate the potential impact of the RAS/MAPK pathway - a major driver of colorectal carcinogenesis - on the circadian clock , we used a computational model which predicted that perturbation of BMAL1-mediated transcription can generate the circadian phenotypes similar to those observed in metastatic cell lines ., Using an inducible RAS expression system , we show that overexpression of RAS disrupts the circadian clock and leads to an increase of the circadian period while RAS inhibition causes a shortening of period length , as predicted by our mathematical simulations ., Together , our data demonstrate that perturbations induced by a single oncogene are sufficient to deregulate the mammalian circadian clock .

Living systems possess an endogenous time-generating system – the circadian clock - accountable for a 24 hours oscillation in the expression of about 10% of all genes ., In mammals , disruption of oscillations is associated to several diseases including cancer ., In this manuscript , we address the following question: what are the elicitors of a disrupted clock in cancer ?, We applied a systems biology approach to correlate experimental , bioinformatics and modelling data and could thereby identify key genes which discriminate strong and weak oscillators among cancer cell lines ., Most of the discriminative genes play important roles in cell cycle regulation , DNA repair , immune system and metabolism and are involved in oncogenic pathways such as the RAS/MAPK ., To investigate the potential impact of the Ras oncogene in the circadian clock we generated experimental models harbouring conditionally active Ras oncogenes ., We put forward a direct correlation between the perturbation of Ras oncogene and an effect in the expression of clock genes , found by means of mathematical simulations and validated experimentally ., Our study shows that perturbations of a single oncogene are sufficient to deregulate the mammalian circadian clock and opens new ways in which the circadian clock can influence disease and possibly play a role in therapy .

oncology, systems biology, text mining, computer and information sciences, medicine and health sciences, computer modeling, theoretical biology, basic cancer research, genetics, biology and life sciences, information technology, computational biology, computerized simulations

journal.pgen.1005990

Arabidopsis thaliana DM2h (R8) within the Landsberg RPP1-like Resistance Locus Underlies Three Different Cases of EDS1-Conditioned Autoimmunity

In plants , receptors that sense pathogen attack are central players in the biotic stress signaling network ., Receptor activation triggers innate immunity pathways to protect cells and tissues from disease ., In a first line of defense , surface pattern recognition receptors ( PRRs ) bind microbial molecules to activate disease resistance programs leading to pattern-triggered immunity ( PTI ) ., A second critical immunity layer is mediated by intracellular nucleotide-binding/leucine-rich-repeat ( NLR ) receptors that recognize virulence factors ( called effectors ) which are delivered by pathogen strains to dampen PTI and promote disease 1 ., Structural counterparts of plant NLRs called NOD-LRR ( nucleotide-binding/oligomerization-domain/leucine-rich-repeat ) receptors also sense pathogen interference in mammalian systems 2 , 3 ., NLR and NOD-LRR proteins are ATP-driven molecular switches which become stimulated by direct binding of an effector molecule or effector modifications of an NLR-monitored host target 4 , 5 ., In plants , NLR activation induces a robust resistance response called effector-triggered immunity ( ETI ) involving the amplification of PTI-related transcriptional programs and , often , host cell death at infection sites ( a hypersensitive response , HR ) 6 ., NLRs are among the most rapidly evolving plant genes 7–9 , and expansion in NLR gene number and diversity , as paralogs within complex loci or allelic variants in different genotypes , is in part driven by pathogen effector pressure 10–13 ., Receptor monitoring ( or guarding ) of important defense hubs that are targeted by multiple pathogen effectors probably further increases NLR recognition space 14–17 ., Nevertheless , the rapid evolution of NLR genes creates potentially dangerous molecules if activated in the absence of a pathogen effector stimulus 4 , 18 ., Loss of NLR homeostasis caused by mutation , mis-expression or disturbance of NLR-monitored co-factors leads to autoimmunity ., Plant autoimmune backgrounds display constitutive defense gene expression and varying degrees of stunting , necrosis and reduced reproductive fitness 19 ., As in ETI , NLR autoimmune phenotypes are often conditional on temperature with high temperatures ( 25–28°C ) suppressing disease resistance , transcriptional activation of defense pathways and HR-related cell death 19–21 ., Temperature-conditioned autoimmunity can also arise in the progeny of inter- or intra-specific crosses between different genetic backgrounds to produce immune-related hybrid incompatibility ( HI ) ( known also as hybrid necrosis ) 19 , 22 ., HI is caused by deleterious epistatic interactions between two or more loci that have diverged through genetic drift or selection in the different parental lineages 23–25 ., Mapping of the causal interacting genes or allelic forms in several cases of temperature-conditioned HI shows that many are in NLR or immune-related loci 18 , 22 , 25–29 ., Therefore , HI might expose altered NLR regulation and/or associations with monitored co-factors as immunity systems evolve ., Effector-activated NLR receptors connect to a conserved basal resistance network to mobilize ETI defense pathways 6 ., Although the downstream events are not well understood , signals in ETI ultimately converge on the nuclear transcription machinery to boost PTI-related defense programs 6 ., A major NLR subclass in dicotyledenous species has an N-terminal Toll-Interleukin1-receptor ( TIR ) domain ( referred to as TNLs or TIR-NB-LRRs ) 9 , 30 and requires the nucleocytoplasmic , lipase-like protein ENHANCED DISEASE SUSCEPTIBILITY1 ( EDS1 ) for all measured ETI and autoimmunity outputs 21 , 31–34 ., Interactions between EDS1 and TNL proteins suggested that EDS1 provides an immediate link between TNLs and downstream resistance pathways 35–37 ., Importantly , EDS1 nuclear accumulation was found to be necessary for A . thaliana basal immunity against virulent pathogen strains and TNL-triggered ETI , consistent with a central EDS1 role in transcriptional reprogramming of cells for defense 21 , 32 , 38 ., Analysis of A . thaliana transgenic plants in which EDS1 was mis-localized to the cytoplasm or its nucleocytoplasmic trafficking disturbed , suggested also that the EDS1 cytoplasmic pool contributes to resistance 38 , 39 ., Unlike many mis-regulated NLRs , over-accumulation of functional , nucleocytoplasmic A . thaliana EDS1 does not cause autoimmunity 38 , 40 ., Here , we investigated the consequences of restricting A . thaliana EDS1 to the nuclear compartment ., Our analysis shows that a low-level EDS1 nuclear pool , operating with signaling partners , is sufficient for mediating A . thaliana basal and TNL immunity without deleterious consequences for the plant ., However , raising nuclear EDS1 amounts above a certain threshold leads to autoimmunity with many features of a deregulated TNL immune response ., In a screen for genetic suppressors , we discover that the nuclear EDS1 autoimmune phenotype depends on presence of the ‘DANGEROUS MIX2’ ( DM2 ) RPP1-like TNL gene cluster ., The DM2 locus is a hotspot for genes underlying immune-related HI ., In our case , a cluster of eight RPP1-like TNL genes linked to an eds1 deletion mutation had been co-introgressed from A . thaliana accession Landsberg erecta into accession Columbia ( Col ) ., We identify one gene , DM2h , within the DM2 RPP1-likeLer locus as necessary for nuclear EDS1 autoimmunity ., We propose that a weak DM2hLer autoactivity which is normally constrained is exposed by nuclear EDS1 , producing EDS1-dependent defense expression and autoimmunity ., A corollary of this damaging co-action between a TNL and nuclear EDS1 is that in wild-type plants , regulating the nuclear EDS1 pool likely helps to maintain TNL immune pathway homeostasis and growth ., We tested whether increased targeting of EDS1 to nuclei affects its disease resistance activity ., For this , A . thaliana stable transgenic lines expressing genomic EDS1 under control of its native promoter and fused to a C-terminal yellow fluorescent protein ( YFP ) tag and SV40 nuclear localization signal ( NLS ) were generated in an eds1-2 deletion mutant in accession Col-0 ( Col ) ( Fig 1A ) ., The eds1-2 mutation had been introgressed originally from accession Landsberg erecta ( Ler ) over eight backcrosses because Col contains a tandem duplication of two functional EDS1 genes 41 ., Three independent EDS1-YFPNLS lines ( #A3 , #A5 and #B2 ) were taken to homozygosity and tested alongside a previously characterized Col eds1-2 transgenic line expressing functional , genomic EDS1-YFP 38 ., EDS1-YFP protein accumulation in leaf extracts of the different transgenic lines was compared to that of native EDS1 in Col by immunoblotting with anti-EDS1 antibodies ., The EDS1-YFPNLS protein levels ranged from lower than wild-type EDS1 ( in EDS1-YFPNLS line #B2 ) to higher than wild-type EDS1 ( EDS1-YFPNLS line #A5 ) , with highest accumulation in EDS1-YFPNLS line #A3 ( Fig 1B ) ., Accumulation of EDS1-YFP ( without an additional NLS ) was intermediate between that of EDS1-YFPNLS lines #A5 and #A3 ( Fig 1B ) ., Confocal laser scanning microscopy of leaf epidermal cells showed that EDS1-YFP distributed in the cytoplasm and nucleus , as expected 38 , whereas EDS1-YFPNLS was detected only in nuclei in lines #B2 , #A5 and #A3 ( Fig 1C ) ., Biochemical purification of nuclei from leaf tissues showed that there was strong nuclear enrichment of EDS1 protein in the EDS1-YFPNLS line #A5 compared to EDS1-YFP ( Fig 1D ) ., Growth of EDS1-YFP and EDS1-YFPNLS #B2 , #A5 and #A3 plants in soil under short-day conditions ( 10 h light period at 22°C ) was monitored over several weeks ., EDS1-YFP and the EDS1-YFPNLS low expressor line #B2 were undistinguishable from wild-type Col or Col eds1-2 ( Figs 1E and S1A ) ., By contrast , EDS1-YFPNLS #A3 seedlings became stunted and chlorotic after the first true leaves emerged at ~ 2 weeks and were dead at 4 weeks ( Figs 1E and S1A ) ., EDS1-YFPNLS #A5 plants displayed stunting , curling of leaves and chlorosis from 4–5 weeks but remained viable and partially fertile ( Figs 1E and S1A ) ., The developmental defects of EDS1-YFPNLS lines #A3 and #A5 co-segregated with the T-DNA selection marker ., Also , the T-DNA insertion in EDS1-YFPNLS #A3 mapped to the first exon of At4g28490 , in which an insertion mutation ( in the haesa ( hae ) single mutant ) does not have a visible phenotype 42 ., These results suggest that increased EDS1 nuclear localization or an imbalance in EDS1 nucleocytoplasmic partitioning , rather than EDS1 over expression , leads to EDS1 dose-dependent growth defects ., We also generated Col eds1-2 transgenic lines expressing EDS1-YFP fused to a mutated , inactive NLS ( Figs 1A and S1 ) 43 ., No line was found that expressed EDS1-YFPnls protein as highly as EDS1-YFPNLS in line #A5 ., Two EDS1-YFPnls lines ( nls*#α5 and nls*#β5 ) were selected that had moderately high EDS1-YFP expression ( S1B Fig ) ., These showed a nucleocytoplasmic distribution of EDS1-YFP ( S1C Fig ) and grew normally at 22°C ( S1D Fig ) ., Because the developmental phenotypes in EDS1-YFPNLS lines #A3 and #A5 resemble A . thaliana autoimmunity backgrounds we measured expression of the EDS1-dependent defense marker genes PATHOGENESIS RELATED1 ( PR1 ) and AvrPphB SUSCEPTIBLE3 ( PBS3 ) in EDS1-YFPNLS transgenic and control lines ., PR1 and PBS3 expression remained low in Col , Col eds1-2 and the phenotypically normal EDS1-YFP or EDS1-YFPNLS #B2 lines over a 3–6 week growth period ( Fig 2A ) ., From ~ 4 weeks on , PR1 and PBS3 expression increased in EDS1-YFPNLS line #A5 ( Figs 2A and S1E ) , consistent with the appearance of macroscopic growth defects ., High PR1 and PBS3 expression was also detected in 3-week-old dying EDS1-YFPNLS #A3 plants ( Fig 2A ) ., By contrast , the EDS1-YFP-nls lines *#α5 and *#β5 did not have elevated PR1 expression ( S1F Fig ) ., A gradual increase in EDS1 total protein accumulation over 3–6 weeks development was detected in both the EDS1-YFP and EDS1-YFPNLS lines ( Fig 2B ) , suggesting that there is a general rise in EDS1 steady state levels as plants age , regardless of EDS1 nucleocytoplasmic or nuclear distribution ., Total and free SA levels were unchanged in 5-week-old EDS1-YFPNLS #B2 , Col and Col eds1-2 plants , but were high in line EDS1-YFPNLS #A5 ( Fig 2C ) ., Hence , during development , accumulation of nuclear EDS1 in EDS1-YFPNLS lines #A3 and #A5 appears to reach a threshold for causing defense gene activation and disturbed growth ., These results show that nuclear EDS1-YFPNLS in line #A5 , and more acutely in #A3 , has the capacity to transcriptionally activate defense pathways in the absence of a pathogen stimulus ., We tested whether EDS1 targeted to nuclei is sufficient to confer basal disease resistance by spray-infecting leaves with the virulent bacterial pathogen Pseudomonas syringae pv ., tomato strain DC3000 ( Pst DC3000 ) ., As expected , Pst DC3000 growth was higher at 3 d post-infection ( 3 dpi ) in Col eds1-2 than in wild-type Col leaves , indicative of a loss of basal resistance in Col eds1-2 ( Fig 2D ) ., The eds1-2 defect was fully complemented in EDS1-YFPNLS #B2 expressing low levels of EDS1-YFPNLS ( Figs 1B and 2D ) ., Pst DC3000 growth was marginally reduced on EDS1-YFPNLS #A5 compared to wild-type Col plants ( Fig 2D ) ., Similar resistance trends were observed in these transgenic lines in response to infection by a virulent oomycete pathogen , Hyaloperonospora arabidopsidis ( Hpa , isolate Noco2 ) ( Fig 2E ) ., We then tested whether nuclear-enriched EDS1 functions in ETI by inoculating plants with Pst DC3000 delivering the Type-III secreted effector AvrRps4 ( Pst AvrRps4 ) , or with an incompatible Hpa isolate , Emwa1 ., In accession Col , AvrRps4 is recognized by the nuclear TNL receptor pair RRS1/RPS4 32 , 44–46 and Hpa Emwa1 by the TNL receptor RPP4 47 , in EDS1-dependent ETI ., Accordingly , Pst AvrRps4 growth at 3 dpi was restricted in wild-type Col in an EDS1-dependent manner ( Fig 2F ) ., RRS1/RPS4 ETI against Pst AvrRps4 was also fully restored in EDS1-YFPNLS lines #A5 and #B2 ( Fig 2F ) , as well as in EDS1-YFPnls lines *#α5 and *#β5 ( S1G Fig ) ., EDS1-YFPNLS #A5 and #B2 restricted Hpa Emwa1 growth as efficiently as wild-type Col , with all lines exhibiting a host hypersensitive response ( HR ) at attempted Hpa infection sites , as measured by Trypan Blue ( TB ) -staining of infected leaves ( Fig 2G ) ., As expected , Col eds1-2 plants were fully susceptible to Hpa Emwa1 infection ( Fig 2G ) ., No HR lesioning was observed in mock-inoculated EDS1-YFPNLS lines #A5 or #B2 , indicating that the host HR is pathogen-triggered ( Fig 2G ) ., We concluded that even low levels of nuclear-targeted EDS1 , as in EDS1-YFPNLS #B2 , are sufficient for Arabidopsis basal and TNL-conditioned immunity ., Because many Arabidopsis effector-triggered TNL and autoimmunity phenotypes are attenuated at elevated temperatures , we tested whether high temperature alters EDS1-YFP nuclear accumulation ., At 28°C , accumulation of the nucleocytoplasmic TNL proteins tobacco N , Arabidopsis RPS4 and SNC1 ( SUPPRESSOR OF npr1-1 CONSTITUTIVE1 ) inside nuclei and EDS1-dependent transcriptional reprogramming are reduced 21 , 48 , 49 ., Macroscopic growth defects and enhanced PR1 expression in EDS1-YFPNLS lines #A3 and #A5 at 22°C were also suppressed when plants were propagated at 28°C ( S1A and S1E Fig ) ., Confocal laser scanning microscopy of leaves taken directly from plants grown at 22°C or 28°C showed that the distribution of nucleocytoplasmic EDS1-YFP or nuclear EDS1-YFPNLS fluorescence signals did not change substantially between the two temperature regimes ( Fig 3A ) ., Therefore , high temperature suppression of EDS1-YFPNLS autoimmunity in line #A5 is not due to a failure in EDS1 nuclear import ., However , steady state levels of EDS1-YFPNLS were lower in plants grown at 28°C compared to 22°C , as monitored on immunoblots with anti-EDS1 antibodies ( Fig 3B ) ., A decrease in native EDS1 protein accumulation was also detected in wild-type Col grown at 28°C ( Fig 3B ) ., Therefore , growth at 28°C leads to reduced EDS1 protein accumulation regardless of whether EDS1 is confined to the nucleus or free to shuttle between the nucleus and cytoplasm 38 ., This is in line with a reported lowering of EDS1 transcript levels under high temperature conditions 50 ., We concluded that suppression of autoimmunity in EDS1-YFPNLS #A5 , and probably also #A3 at 28°C ( S1A Fig ) , is caused by reduction of nuclear EDS1 to below a threshold needed to elicit autoimmunity ., A . thaliana EDS1 forms resistance signaling complexes with either one of two sequence-related partners , PHYTOALEXIN DEFICIENT4 ( PAD4 ) and SENESCENCE ASSOCIATED GENE101 ( SAG101 ) 31 , 40 , 51 , 52 ., Whereas PAD4 compensates genetically for a loss-of-function sag101 mutation , SAG101 only partially compensates for loss of PAD4 in basal resistance against virulent pathogens and in TNL mediated ETI 40 , 51 , 53 ., The enhanced disease susceptibility phenotype of a pad4 sag101 double mutant is as penetrant as an eds1 loss-of-function mutation and is not alleviated by over-expressing functional EDS1-HA 40 , 51 ., Thus , EDS1 requires PAD4 and , in the absence of PAD4 , SAG101 for disease resistance signaling in basal immunity and ETI ., We tested the genetic dependence of EDS1-YFPNLS #A5 autoimmunity on PAD4 and SAG101 by crossing EDS1-YFPNLS #A5 with Col pad4-1 and sag101-1 single null mutants or a Col pad4-1 sag101-1 double mutant and selecting lines that were homozygous for the EDS1-YFPNLS transgene and eds1-2 in the respective homozygous mutant backgrounds ., Developmental ( Fig 4A ) and PR1 expression ( Fig 4B ) autoimmune phenotypes of EDS1-YFPNLS #A5 were fully rescued by pad4-1 and pad4-1 sag101-1 but not by sag101-1 ., This indicates that autoimmunity caused by nuclear-enriched EDS1 has the same genetic requirements for PAD4 and SAG101 as EDS1-mediated basal immunity and ETI in wild-type plants ., EDS1-YFPNLS protein abundance was substantially lower in pad4-1 and pad4-1 sag101-1 mutant backgrounds , and similar to levels of native EDS1 in Col wild-type ( Fig 4C ) ., Reduced EDS1 accumulation can be largely attributed to reduced EDS1 expression as measured by qRT-PCR in the same plants ( Fig 4D ) ., EDS1 is also directly stabilized by PAD4 or SAG101 51 , 52 ., A . thaliana RRS1/RPS4 TNL resistance in EDS1-YFPNLS #A5 against Pst AvrRps4 displayed the same genetic dependence on PAD4 and SAG101 as wild-type EDS1 in Col ( Fig 4E ) ., We conclude that the defense-promoting actions of PAD4 or SAG101 predominantly target the EDS1 nuclear pool in plant immunity ., We next tested whether EDS1-YFPNLS #A5 autoimmunity requires signaling by the defense hormone salicylic acid ( SA ) because EDS1-PAD4 promote SA-dependent and SA-independent defense pathways 41 , 54–56 ., Also , SA feeds-forward to induce PAD4 expression 53 ., For this , loss-of-function mutations in the SA biosynthetic enzyme gene ISOCHORISMATE SYNTHESIS1 ( ICS1 , Col sid2-1 ) or the SA-response regulator gene NON-EXPRESSOR OF PR GENES1 ( NPR1 , Col npr1-1 ) were introduced into the EDS1-YFPNLS #A5 background ., High accumulation of SA in 5-week-old EDS1-YFPNLS #A5 was abolished in EDS1-YFPNLS #A5/sid2-1 plants ( S2A Fig ) , confirming that SA in this line is produced mainly by ICS1 57 ., SA levels were not lower in EDS1-YFPNLS #A5 /npr1-1 , consistent with NPR1 operating downstream of SA accumulation 58 ., Both sid2-1 and npr1-1 abolished enhanced expression of the SA-dependent PR1 marker gene in EDS1-YFPNLS #A5 ( S2B Fig ) , but only slightly compromised accumulation of EDS-YFPNLS protein ( S2C Fig ) ., Strikingly , neither sid2-1 nor npr1-1 suppressed EDS1-YFPNLS #A5 stunting ( S2D Fig ) ., We concluded that EDS1-YFPNLS #A5 immune-related growth defects are SA-independent or have a lower SA threshold ., Altogether , the genetic epistasis data suggest that EDS1-YFPNLS autoimmunity operates by similar mechanisms as pathogen-elicited basal resistance or ETI , with EDS1-PAD4 controlled pathways branching into SA-dependent and SA-independent signaling sectors ., Previously , we found that shifting plants from high ( 28°C , permissive ) to moderate ( 19°C , restrictive ) temperature can be used to trigger EDS1-dependent autoimmunity in a transgenic A . thaliana RPS4 over-expression line ( 35S:RPS4-HS ) 21 ., Analysis of global gene expression changes in 35S:RPS4-HS and 35S:RPS4-HS eds1-2 leaf tissues over a 24 h time course showed that temperature-conditioned RPS4 autoimmunity at 8 h and 24 h post temperature shift ( pts ) largely mirrors EDS1-dependent transcriptional reprogramming in RRS1/RPS4 ( TNL ) ETI against Pst AvrRps4 21 ., Moreover , a set of EDS1-dependent induced or repressed marker genes from Pst AvrRps4-triggered tissues at 6 h post infection ( hpi ) displayed the same EDS1-dependent trends in 35S:RPS4-HS leaves at 8 h pts 21 ., We performed Affymetrix ATH1 GeneChip analysis of 4-week-old untreated EDS1-YFPNLS line #A5 and wild-type Col plants grown at 22°C to measure the extent to which EDS1-YFPNLS #A5 autoimmunity resembles pathogen-elicited or temperature-induced A . thaliana immune responses ., More than 2000 genes were significantly up- or down-regulated ( p-value < 0 . 01 , > 2-fold change ) in EDS1-YFPNLS line #A5 compared to Col at 22°C ., Genes exhibiting at least 4-fold transcriptional differences in EDS1-YFPNLS #A5 compared to Col were then used for hierarchical clustering with transcriptome data sets from bacterial NLR-conditioned PTI or ETI , as well as 35S:RPS4-HS temperature-conditioned autoimmunity ( Fig 5 and S1 Table ) ., This analysis established that the EDS1-YFPNLS #A5 transcriptome was most similar to 35S:RPS4-HS gene expression changes at 8 h and 24 h pts and to those of ETI interactions ( Pst AvrRps4 , Pst AvrRpm1 6 h; Fig 5 ) ., The EDS1-YFPNLS #5 transcriptome was most different to those of Pst AvrRps4-elicited or temperature-shift induced eds1 mutant backgrounds ( Fig 5 ) ., Notably , EDS1-dependent induced and repressed genes in the EDS1-YFPNLS #A5 transcriptome overlapped substantially with EDS1-dependent induced and repressed gene sets in RRS1/RPS4-mediated ETI or 35S:RPS4-HS autoimmunity ( Fig 5 ) ., Two clusters of induced and repressed genes were unique to EDS1-YFPNLS #A5 ( a and b in Fig 5 , S2 Table ) and might correspond to adaptation to prolonged defense activation in the EDS1-YFPNLS #A5 line ., The above results suggest that EDS1-YFPNLS transgenic line #A5 behaves much like a TNL autoimmune background ., Therefore , expressing high levels of nuclear targeted EDS1 is sufficient to induce transcriptional defense reprogramming without pathogen activation of a TNL receptor ., We performed a genetic suppressor screen of the EDS1-YFPNLS #A3 seedling lethality to identify components contributing to nuclear EDS1 autoimmunity ., As shown above , high levels of EDS1-YFPNLS expression in EDS1-YFPNLS line #A3 caused rapid decline and eventual death of 3- to 4-week-old plants at moderate temperature ( 22°C ) ( Figs 1 and S1 ) ., The lethality phenotype was fully penetrant at 22°C and stable after three generations of propagating EDS1-YFPNLS #A3 at 28°C ., Seeds of EDS1-YFPNLS line #A3 were mutagenized with ethyl methane sulfonate ( EMS ) ., This led to the isolation of mutants we have named ‘near death experience’ ( nde ) , which exhibited restored seedling viability and vigor to varying extents at 22°C ., Seven putative dominant mutations ( nde1 to nde7 ) were identified by screening EMS mutagenized seedlings directly in the M1 generation ( Fig 6A ) ., A further 175 M2 pools ( nde8–175; each derived from ~ 100 M1 plants propagated at 28°C ) were screened at 22°C and ~ 50 additional nde mutants isolated from independent M2 pools ( Fig 6A ) ., Here , we describe analysis of a single nde complementation group containing alleles isolated in both the M1 ( nde1-1 , nde1-3 ) and M2 ( nde1-13 , nde1-150 and nde1-175 ) suppressor screens ., nde1-1 and nde1-3 were initially scored as dominant suppressor mutations ., When grown at 22°C , homozygous nde1-1 and nde1-3 M3 generation seedlings were indistinguishable from wild-type Col , whereas the parental EDS1-YFPNLS line #A3 was severely stunted ( Fig 6B ) ., Further lowering of the growth temperature to 16°C did not produce nde1-1 and nde1-3 stunting or chlorosis ., Homozygous nde1-1 and nde1-3 plants were backcrossed to the parental EDS1-YFPNLS #A3 line and segregation of the seedling lethality phenotype at 22°C recorded in the F2 generation ( BC1-F2 ) ., In both mutants , fully rescued nde , intermediate , and seedling lethal phenotypes segregated in a 1:2:1 ratio ( nde1-1: 79:150:59 , Chi2 = 3 . 28 ) ., This mode of inheritance suggests that nde1-1 and nde1-3 are loss-of-function alleles at single semi-dominant loci ., EDS1-YFPNLS localization remained entirely nuclear in nde1-1 and nde1-3 leaves , although YFP fluorescence intensity in the mutant lines was reduced compared to EDS1-YFPNLS line #A3 , assessed by confocal laser-scanning microscopy ( Fig 6C ) ., Therefore , we reasoned that phenotypic rescue was not due to interference with EDS1-YFP nuclear import but more likely reduced EDS1-YFP nuclear accumulation in nde1-1 and nde1-3 ., The SA-response marker gene PR1 was strongly induced in 3-week-old EDS1-YFPNLS line #A3 seedlings shifted to 18°C for 24h , but not in nde1-1 and nde1-3 ( Fig 6D ) ., EDS1 displayed a similar expression pattern to PR1 in these seedlings ( Fig 6D ) ., Therefore , mutations in nde1-1 and nde1-3 attenuate EDS1 mRNA accumulation under conditions inducing autoimmunity in the parental NLS#A3 line ., Accumulation of EDS1-YFPNLS protein was monitored in the same plants ., EDS1 levels in nde1-1 and nde1-3 were lower than in the parental NLS#A3 line and comparable to those in line NLS#A5 showing autoimmunity under the same conditions ( Fig 6E ) ., Thus , suppression of autoimmunity in nde1-1 and nde1-3 is not solely caused by a reduction of EDS1-YFPNLS levels ., The similarity of nde1-1 and nde1-3 phenotypes ( Fig 6B ) prompted us to perform an allelism test ., nde1-1 x nde1-3 F1 plants grew normally at 22°C ( S3A Fig ) ., Approximately 400 F2 plants originating from four individual nde1-1 x nde1-3 F1 plants also showed no signs of stunting or chlorosis at 22°C ( S3A Fig ) ., Therefore , the possibility of F1 phenotypic rescue through actions of independent semi-dominant alleles ( non-allelic non-complementation ) was excluded , unless the independent non-allelic variants are closely linked ., Segregation of a specific PCR marker for the nde1-1 mutation generated after A . thaliana whole genome sequencing ( see below ) confirmed that nde1-1 x nde1-3 F1 plants were derived from true crosses ( S3B Fig ) ., The nde1-13 , nde1-150 and nde1-175 mutations obtained in screens of EMS-mutagenized M2 plants fully rescued viability of EDS1-YFPNLS #A3 at 22°C and were inherited in a semi-dominant manner ., Also , nde1-13 , nde1-150 and nde1-175 were found to be allelic with nde1-1 after crossing and growing PCR-validated seedlings in the F2 generation ., We concluded that nde1-1 , nde1-3 , nde1-13 , nde1-150 and nde1-175 form a single complementation group of semi-dominant suppressors of nuclear EDS1 autoimmunity ., We performed mapping-by-sequencing of the nde1-1 and nde1-3 mutations ( see Materials and Methods ) 59–61 ., A . thaliana Col x Ler SNPs were used to delineate the introgressed Ler portion of DNA containing the eds1-2 mutation 41 to an approximately 6 Mb region in the parental EDS1-YFPNLS#A3 line ( S4 Fig ) ., Few polymorphisms with the Col reference sequence were detected in the remainder of the genome ., Using SHOREmap 62 , nde1-1 and nde1-3 were mapped to an approximately 5 Mb candidate region on the lower arm of chromosome 3 , coinciding with the parental Ler introgression ( Figs 7A and S5 ) ., However , no locus containing a mutation in both nde1-1 and nde1-3 bulk sequences , expected for allelic mutations , was identified ., We considered that NDE1 might be a Ler-specific gene or structural variant that is not present in the Col reference genome ., Genetic crosses of EDS1-YFPNLS #A3 and nde1-1 to Col and Col eds1-2 , respectively , confirmed that NDE1 encodes a Ler-specific autonecrosis-inducing factor which is lacking in Col ( S3 Table ) ., NDE1 was fine-mapped to a 90 kb interval in the Col reference genome by recombination mapping , and a physical contig of this region , which in accession Ler spans 134 kb , assembled using a previous construction of the same locus in Ler 27 ( see Materials and Methods ) ., Notably , the NDE1 mapping interval contained QTL3Ler , a polymorphic region covering two TNL RPP1-like paralogs in Col 27 ., The RPP1-like nomenclature derives from its close relatedness to a cluster of TNL RPP1 genes in A . thaliana accession Ws-2 whose different paralogs confer isolate-specific Hpa ( formally Peronospora parasitica ) resistance 27 , 63 , 64 ., In accession Ler , the QTL3 region has expanded to contain seven complete and one truncated RPP1-like genes ( denoted R1-R8 , Fig 7B ) 27 and the RPP1-likeLer cluster was found to be the causal locus in a recessive deleterious epistatic interaction with Strubbelig-Receptor Family 3 ( SRF3 ) allelic forms from A . thaliana accessions Kashmir ( Kas-2 ) and Kondara ( Kond-0 ) , producing immune-related HI 25 ., RPP1-likeLer R1-R8 correspond to DM2a-h paralogs of the DANGEROUS MIX2 locus which underlies multiple negative epistatic interactions among A . thaliana genetic accessions leading to HI 18 , 22 ., For simplicity , we now refer to the RPP1-likeLer R1-R8 genes as RPP1-likeLer DM2a-h ( Fig 7B ) ., We reasoned that the nde1 mutations might affect one or more of the RPP1-likeLer DM2a-h ( R1-R8 ) genes ., Illumina reads from mapping-by-sequencing were re-analyzed against a reference genome containing the Ler NDE1 mapping interval ., No canonical EMS changes were identified within the NDE1 mapping interval but manual inspection revealed a prominent drop in read coverage along the RPP1-likeLer cluster in nde1-3 , extending from the DM2c-d ( R3-R4 ) intergenic region to the DM2h ( R8 ) 5’ region ( Fig 7B ) ., This was consistent with a large deletion or structural rearrangement in this line , which was confirmed by diagnostic PCR ( S6A Fig ) ., Similarly , a 14 bp deletion leading to a premature STOP was detected in the fifth exon of DM2h in nde1-1 ( Figs 7C and S6B ) ., No additional SNPs were detected within the mapping interval in nde1-1 bulk sequencing data , indicating that NDE1 is DM2h ., The DM2h coding region from nde1-13 , nde1-150 and nde1-175 was therefore obtained by Sanger-sequencing ., From this , EMS mutations leading to a premature stop in nde1-13 ( W1129Stop ) or amino acid exchanges R1069C and C945Y , respectively in nde1-150 and nde1-175 , were detected ( Fig 7C ) ., Also , EDS1-YFPNLS #A3 necrosis was restored in T2 progeny of the nde1-1 mutant transformed with a RPP1-likeLer genomic DM2h construct ( S7 Fig ) ., These results show that DM2h ( R8 ) within the RPP1-likeLer TNL gene cluster interacts genetically with EDS1-YFPNLS resulting in autoimmunity ., Having identified RPP1-likeLer DM2h as causal in nuclear EDS1 autoimmunity , we tested whether the EDS1-YFPNLS #A3 or #A5 autoimmune response is accompanied by induced DM2h expression ., DM2h expression was significantly reduced in nde1 alleles compared to autoimmune lines EDS1-YFPNLS #A5 and #A3 , but there was only a two-fold increase in DM2h expression in the autoimmune lines ( S8A Fig ) , although these had induced PR1 expression ( Fig 6D ) ., This suggests that the DM2h gene itself is not strongly responsive to autoimmunity , in agreement with Alcazar et al ( 2014 ) ., A previous screen for senescence-associated mutants in A . thaliana accession Ler identified an EMS-induced mutation , onset of leaf death 3–1 ( old3-1 ) in the cysteine metabolic enzyme-coding locus O-acetylserine ( thiol ) lyase A1 , which also displays negative epistasis with the RPP1-likeLer gene cluster 65 , 66 ., Notably , old3-1 caused autonecrosis in Ler , but not Col , and was suppressed by amiRNA silencing of the RPP1-likeLer cluster 65 ., More specifically , silencing of DM2g ( R7 ) most closely correlated with the suppression of old3-1 dwarfism 65 ., Here , we tested whether the RPP1-likeLer DM2h gene contributes to autonecrosis induced by old3-1 ., From a Col x Col eds1-2 cross , we selected two independent near isogenic lines ( NILs ) containing the RPP1-likeLer locus and wild-type EDS1 from Col ( Col-RPP1-likeLer ) ., Similarly , we selected two independent NILs containing the RPP1-likende1-1 locus and wild-type EDS1 , but not the EDS1-YFPNLS #A3 transgene from a Col x nde1-1 cross ( Col-RPP1-likende1-1 ) ., Hence , the NILs differ mainly in the presence of a 14bp deletion in DM2h ( R8 ) in Col RPP1-likende1-1 but not Col-RPP1-likeLer ., We used these NILs first to test whether DM2h ( R8 ) contributes to other resistance responses not related to autoimmunity ., NILs were infected with virulent ( Pst DC3000 , Hpa Noco2 ) and avirulent ( Pst AvrRps4 , Hpa Cala2 ) pathogen isolates ( S9 Fig ) ., There were no measurable differences in resistance between the NILs , suggesting that DM2h does not act as a helper NLR or generally lower NLR resistance thresholds ., The NILs developed normally and were crossed with Ler old3-1 ., F2 plants homozygous for old3-1 and either RPP1-likeLer or RPP1-likende1-1 were selected and symptoms of autonecrosis monitored in F3 progeny ., old3-1 plants grown at 28°C were not autonecrotic ( Fig 8A ) 65 ., At 18°C , Col and Ler were healthy but old3-1 plants became necrotic ( Fig 8A ) ., Col/Ler hybrids containing old3-1 and RPP1-likeLer , but not hybrids containing old3-1 and RPP1-likende1-1 ( lacking functional DM2h ) , also became necrotic ( Fig 8A ) ., Similarly , PR1 and EDS1 expression was upregulated in Ler old3-1 and Col/Ler RPP1-likeLer old3-1 plants , but not Col/Ler RPP1-likende1-1 old3-1 , old3-1 grown

Introduction, Results, Discussion, Materials and Methods

Plants have a large panel of nucleotide-binding/leucine rich repeat ( NLR ) immune receptors which monitor host interference by diverse pathogen molecules ( effectors ) and trigger disease resistance pathways ., NLR receptor systems are necessarily under tight control to mitigate the trade-off between induced defenses and growth ., Hence , mis-regulated NLRs often cause autoimmunity associated with stunting and , in severe cases , necrosis ., Nucleocytoplasmic ENHANCED DISEASE SUSCEPTIBILITY1 ( EDS1 ) is indispensable for effector-triggered and autoimmune responses governed by a family of Toll-Interleukin1-Receptor-related NLR receptors ( TNLs ) ., EDS1 operates coincidently or immediately downstream of TNL activation to transcriptionally reprogram cells for defense ., We show here that low levels of nuclear-enforced EDS1 are sufficient for pathogen resistance in Arabidopsis thaliana , without causing negative effects ., Plants expressing higher nuclear EDS1 amounts have the genetic , phenotypic and transcriptional hallmarks of TNL autoimmunity ., In a screen for genetic suppressors of nuclear EDS1 autoimmunity , we map multiple , independent mutations to one gene , DM2h , lying within the polymorphic DANGEROUS MIX2 cluster of TNL RPP1-like genes from A . thaliana accession Landsberg erecta ( Ler ) ., The DM2 locus is a known hotspot for deleterious epistatic interactions leading to immune-related incompatibilities between A . thaliana natural accessions ., We find that DM2hLer underlies two further genetic incompatibilities involving the RPP1-likeLer locus and EDS1 ., We conclude that the DM2hLer TNL protein and nuclear EDS1 cooperate , directly or indirectly , to drive cells into an immune response at the expense of growth ., A further conclusion is that regulating the available EDS1 nuclear pool is fundamental for maintaining homeostatic control of TNL immune pathways .

Plants tune their cellular and developmental programs to different environmental stimuli ., Central players in the plant biotic stress response network are intracellular NLR receptors which intercept specific disease-inducing molecules ( effectors ) produced by pathogenic microbes ., Variation in NLR gene repertoires between plant genetic lines is driven by pathogen selection pressure ., One evolutionary question is how new , functional NLRs are assembled within a plant genome without mis-activating defense pathways , which can have strong negative effects on growth and fitness ., This study focuses on a large , polymorphic sub-class of NLR receptors called TNLs present in dicotyledenous plant lineages ., TNL receptors confer immunity to a broad range of pathogens ., They also frequently underlie autoimmunity caused by their mis-regulation or deleterious allelic interactions with other genes in crosses between different genetic lines ( hybrid incompatibility , HI ) ., TNL pathogen-triggered and autoimmune responses require the conserved nucleocytoplasmic protein EDS1 to transcriptionally reprogram cells for defense ., We discover in Arabidopsis thaliana that high levels of nuclear-enriched EDS1 induce transcriptional activation of defenses and growth inhibition without a pathogen effector stimulus ., In a mutational screen , we identify one rapidly evolving TNL gene , DM2hLer , as a driver of nuclear EDS1 autoimmunity ., DM2hLer also contributes to two separate cases of EDS1-dependent autoimmunity ., Genetic cooperation between DM2hLer and EDS1 suggests a functional relationship in the transcriptional feed-forward regulation of defense pathways .

biotechnology, plant anatomy, medicine and health sciences, immunology, brassica, plant biotechnology, plant science, model organisms, genetically modified plants, plants, extraction techniques, genetic engineering, research and analysis methods, arabidopsis thaliana, genetically modified organisms, protein extraction, gene expression, leaves, genetic loci, agriculture, plant and algal models, phenotypes, genetics, biology and life sciences, agricultural biotechnology, autoimmunity, organisms

journal.pbio.2005127

Neural responses to natural and model-matched stimuli reveal distinct computations in primary and nonprimary auditory cortex

One definition of understanding a neural system is to be able to build a model that can predict its responses ., Responses to natural stimuli are of particular interest , both because natural stimuli are complex and varied and thus provide a strong test of a model , and because sensory systems are presumably adapted to represent features present in natural stimuli 1–3 ., The evaluation of models by their ability to predict responses to natural stimuli is now widespread in sensory neuroscience 4–16 ., A challenge for this approach is that because natural stimuli are richly structured , the features of a set of natural stimuli in one model ( or model stage ) are often correlated with the features in other models ( or model stages ) 17 , 18 ., Model features can thus in principle predict neural responses to a natural stimulus set , even if the neural responses are in fact driven by other features not captured by the model ., Related issues have been widely discussed in the receptive field estimation literature 4 , 19 but have been less noted in cognitive neuroscience 17 , 18 ., A canonical example of this phenomenon occurs in the auditory domain , where there is still considerable uncertainty regarding computational descriptions of cortical processing ., Consider a common model of auditory processing , in which a sound waveform is processed by two stages of filters intended to mimic cochlear and cortical filtering , respectively 20 ( Fig 1A ) ., The filters in the second model stage are tuned to temporal and spectral modulations in the spectrogram-like representation produced by the cochlea ., Such filters and variants thereof are commonly used to account for human perceptual abilities 21–25 and to explain neural responses throughout the auditory pathway 2 , 7 , 11 , 12 , 26–36 ., But in natural stimuli , the responses of these second-stage filters are often correlated with other sound properties , such as semantic categories ( Fig 1B ) 37 , which can confound the interpretation of neural responses ., Speech , for instance , has a distinctive temporal modulation rate that corresponds loosely to the rate of syllabic patterning 38 , music has distinctive temporal modulations reflective of its beat structure 39 , and both speech and music have characteristic spectral modulations due to harmonic frequency structure 20 ., However , speech , music , and other natural sounds also have many unique properties that are not captured by spectrotemporal modulation alone 40 ., Thus , if a neuron responds more to speech than to other sounds , modulation filters may be able to predict the neuron’s response , even if the response is driven by another property of speech that is not captured by such filters ., This is what we term a “stimulus-driven response correlation , ” created when different stimulus properties ( e . g . , spectrotemporal modulations and semantic categories ) are correlated within a particular stimulus set , making their contribution to the neural response difficult to tease apart ., Here , we propose a complementary method for evaluating models that circumvents the challenge of stimulus-driven response correlations ., The idea is simple: we synthesize a stimulus that yields the same response in a model as a natural stimulus , and then test whether the “model-matched” stimulus elicits the same neural response as the natural stimulus ., The synthesized sounds are not influenced by the correlations between different feature sets that may exist in natural stimuli because they are constrained only by the features in the model ., As a result , they generally differ in other properties that could potentially be important to the neural response , and often sound markedly different from their natural counterparts ., Comparing responses to natural and model-matched sounds thus provides a strong test of the model’s explanatory power ., We demonstrate the method by using it to evaluate whether a common filter bank model of auditory cortex can explain human cortical responses to natural sounds measured with fMRI ., Many prior fMRI studies of auditory cortex have identified aspects of cortical tuning that are unique to nonprimary regions 16 , 17 , 41 , such as selectivity for voice 42 , speech 43 , 44 , and music 45–47 ., At the same time , other studies have demonstrated that the standard filter bank model has relatively good predictive accuracy throughout primary and nonprimary regions 7 , 12 , 16 , raising the possibility that primary and nonprimary regions encode sound using similar representations ., Alternatively , such predictions could in part reflect stimulus-driven correlations ., Here , we addressed this question by comparing cortical fMRI responses to natural and model-matched stimuli ., The model-matched stimuli were synthesized to yield the same response as a natural sound in one of several models of varying complexity , ranging from a model of just the cochlea’s response to the two-stage spectrotemporal filter bank model shown in Fig 1A 20 ., Our results show that tuning for temporal and spectral modulations explains much of the voxel response to natural sounds in human primary auditory cortex ( PAC ) but much less of the response in nonprimary areas ., This functional difference between primary and nonprimary regions was much less evident using conventional model predictions because of the effect of stimulus-driven response correlations ., Our findings provide novel evidence for functional differentiation between primary and nonprimary auditory cortex , and suggest that nonprimary regions build higher-order representations that cannot be explained by standard models ., Our methodology could provide stronger tests of neural models in any system for which models are used to predict neural responses ., The goal of this paper was to test whether conventional auditory models can explain voxel responses in auditory cortex to natural sounds ., The models we consider are described by a set of model features ( mk ( t ) ) , each of which has a time-varying response to sound determined by the feature’s filter ( Fig 2A ) ., In general , the response of these features will differ across natural sounds , both in their temporal pattern and their time-averaged properties ( S1A Fig ) ., The BOLD signal reflects a time-averaged measure of neural activity , and thus we expect that if a model provides a good description of the underlying neural responses , any two sounds with the same time-averaged model responses should yield the same fMRI response , even if the temporal pattern of the response is different ., To test this prediction , we iteratively modified a noise stimulus ( that was initially unstructured ) so as to match the time-averaged model responses ( S1B Fig ) , similar to methods for texture synthesis 40 , 48–50 ., Because the temporal patterns of the model responses are unconstrained , the model-matched sounds differ from the natural sounds to which they were matched ., Formally , we assume that the response of a voxel to a sound can be approximated as the weighted sum of time-averaged neuronal firing rates ., Here , we assume the voxel response to be a single number because the sounds we present are short relative to the timescale of the BOLD response ., Our goal is to test whether these model feature responses approximate neuronal responses within a voxel , in which case we should be able to approximate the voxel’s response ( vi ) as a weighted sum of time-averaged model responses ( ak ) ( Fig 2A ) :, ak=1T∫0Tg ( mk ( t ) ) dt, ( 1 ), vi=∑k=1Nzk , iak, ( 2 ), where g is an ( unknown ) point-wise function that maps the model responses to a neuronal firing rate ( e . g . , a rectifying nonlinearity ) , zk , i is the weight of model feature k in voxel i , and T is the duration of the response to a sound ., The most common approach for testing Eqs 1 and 2 is to estimate the weights ( zk , i ) that best predict a given voxel’s response to natural sounds ( for a particular choice of g ) and to assess the cross-validated prediction accuracy of the model using these weights ( via explained variance ) ., Here , we instead test the above equations by synthesizing a “model-matched” sound that should yield the same voxel response as a natural sound for all voxels that are accurately described by the model ( Fig 2A ) ., We then test the model’s validity by assessing whether the voxel responses to the two sounds are similar ., In principle , one could synthesize a separate model-matched sound for each voxel after learning the weights ( zk , i ) ., However , this approach is impractical given the many thousands of voxels in auditory cortex ., Instead , we matched the time-averaged response of all features in the model ( i . e . , all ak in Eq 2 are matched; see Fig 2A ) , which guarantees that all voxel responses that can be explained by the model should be matched , regardless of that voxel’s weights ., We accomplished this objective by matching the histogram of each feature’s response ( S1 Fig; see “Model-matching synthesis algorithm” in Materials and methods ) 48 ., Histogram matching implicitly equates the time-averaged response of the model features for any point-wise transformation ( g ) since , for any such transformation , the time-averaged response can be approximated via its histogram ., It thus obviates the need to choose a particular nonlinearity ., Whether or not a voxel responds similarly to natural and model-matched sounds depends on the response properties of the model features and underlying neurons ., If the model features are good approximations to the neurons in a voxel , then the voxel response to natural and model-matched sounds should be similar; if not , they could differ ., Here , we consider model features that are tuned to different patterns of temporal and/or spectral modulation 20 in a “cochleagram” ( Fig 1A ) produced by passing a sound through filters designed to mimic cochlear tuning ., Each model feature is associated with a time-frequency filter tuned to a particular temporal rate and/or scale , as well as to a particular audio frequency ., The response of each model feature is computed by convolving the spectrotemporal filter with the cochleagram ., Although the response time courses of the models considered here are sufficient to reconstruct the stimulus with high accuracy , the time-averaged properties of the filters , as captured by a histogram , are not ., As a consequence , the model-matched sounds differed from the natural sounds they were matched to ., Indeed , many of the model-matched stimuli sound unnatural ( see http://mcdermottlab . mit . edu/svnh/model-matching/Stimuli_from_Model-Matching_Experiment . html for examples ) ., This observation demonstrates that the time-averaged properties of the model’s features , which approximately capture the modulation spectrum ( Fig 2A ) , fail to capture many perceptually salient properties of natural stimuli ( e . g . , the presence of phonemic structure in speech or melodic contours in music ) ., This additional structure is conveyed by temporal patterns in the feature responses , which are not made explicit by the model but which might be extracted by additional layers of processing not present in modulation filter models ., If the neurons in a voxel respond to such higher-order properties ( e . g . , the presence of a phoneme or melodic contour ) , we might expect their time-averaged response to differ between natural and model-matched sounds ., Thus , by measuring the similarity of voxel responses to natural and model-matched sounds , we can test whether the features of the filter bank model are sufficient to explain their response , or whether other features are needed ., We measured fMRI responses to a diverse set of 36 natural sounds and their corresponding model-matched sounds ( Fig 2B ) ., Each sound was originally 10 seconds in duration , but the sounds were broken up into successively presented 2-second excerpts to accommodate the fMRI scanning procedure ( S2 Fig; see “Stimulus presentation and scanning procedure” in Materials and methods ) ., The model-matched sounds were constrained by all of the features from the two-stage filter bank model shown in Fig 1A ( see below for results from sounds constrained by simpler models ) ., We first plot the response of two example voxels from a single subject ( Fig 3A ) , which illustrate some of the dominant trends in the data ., One voxel was located in the low-frequency area of the “high-low-high” tonotopic gradient thought to span PAC , and which is organized in a roughly V-shaped pattern 51–55 ., Another voxel was located outside of tonotopically defined PAC ., We note that how best to define PAC is a matter of active debate 54 , 56–59 , and thus we have quantified our results using both tonotopic and anatomical definitions of PAC ( described below ) ., As shown in Fig 3A , the response of the primary voxel to natural and model-matched sounds was similar ., By contrast , the nonprimary voxel responded notably less to the model-matched sounds ., We quantified the dissimilarity of responses to natural and model-matched sounds by computing the squared error between corresponding pairs of natural and model-matched sounds , normalized by the squared error that would be expected if there was no correspondence between the two sound sets ( see “Normalized squared error” in Materials and methods ) ., We quantified response differences using the squared error rather than the correlation because model matching makes no prediction for how responses to natural and model-matched sounds should differ if the model is inaccurate , and , in practice , responses to model-matched sounds were often weaker in nonprimary regions , a phenomena that would not have been captured by correlation ., At the end of the results , we quantify how natural and model-matched sounds differ by comparing correlation and squared error metrics ., For these example voxels , the normalized squared error ( NSE ) was higher for the nonprimary voxel ( NSE = 0 . 729 ) than the primary voxel ( NSE = 0 . 101 ) , reflecting the fact that the nonprimary voxel showed a more dissimilar response to natural and model-matched sounds ., Moreover , most of the error between responses to natural and model-matched sounds in the primary voxel could be attributed to noise in the fMRI measurements , because a similar NSE value was observed between two independent measurements of the voxel’s response to natural and model-matched sounds ( NSE = 0 . 094 ) ( Fig 3B ) ., By contrast , in the nonprimary voxel , the test-retest NSE ( NSE = 0 . 082 ) was much lower than the NSE between responses to natural and model-matched sounds , indicating that the difference in response to natural and model-matched sounds cannot be explained by a lower signal-to-noise ratio ( SNR ) ., We quantified these effects across voxels by plotting the NSE between responses to natural and model-matched sounds for each voxel ( Fig 3C ) ., Maps were computed from voxel responses in eight individual subjects who were scanned substantially more than the other subjects ( see “Participants” in Materials and methods for details ) and from responses that were averaged across all twelve subjects after aligning their brains ., Data were collected using two different experiment paradigms that differed in the sounds that were repeated within a scanning session ., The results were similar between the two paradigms ( S3 Fig ) , and so we describe them together ( see Materials and methods for details; subjects S1 , S2 , S3 , S7 , and S8 were scanned in Paradigm I; subjects S4 , S5 , and S6 were scanned in Paradigm II . Group results are based on data from Paradigm I ) ., In Paradigm I , only responses to natural sounds were repeated , while in Paradigm II , both natural and model-matched sounds were repeated ., Only voxels with a reliable response are plotted ( test-retest NSE < 0 . 4; see “Evaluating the noise-corrected NSE with simulated data” in Materials and methods for a justification of this criterion; reliability was calculated using natural sounds for Paradigm I and both natural and model-matched sounds for Paradigm II ) ., Subjects have been ordered by the overall reliability of their data ( median test-retest NSE across the superior temporal plane and gyrus , evaluated using natural sounds so that we could apply the same metric to subjects from Paradigms I and II ) ., These maps have been corrected for noise in the fMRI measurements ( see “Noise-correcting the NSE” in Materials and methods ) , but the results were similar without correction ( S4 Fig ) ., Both group and individual subject maps revealed a substantial change across the cortex in the similarity of responses to natural and model-matched sounds ., Voxels in PAC showed a similar response to natural and model-matched sounds with noise-corrected NSEs approaching 0 , indicating nearly identical responses ., Moving away from PAC , NSE values rose substantially , reaching values near 1 in some voxels far from PAC ( Fig 3C ) ., This pattern of results suggests that the filter bank model can explain much of the voxel response in primary regions but much less of the response in nonprimary regions , plausibly because nonprimary regions respond to higher-order features not made explicit by the model ., This result is suggestive of a hierarchy of feature selectivity in auditory cortex and demonstrates where in the cortex the standard filter bank model fails to explain voxel responses ., We quantified the gradient we observed between primary and nonprimary voxels by binning the NSE of voxels from individual subjects based on their distance to PAC ., Similar results were observed for tonotopic ( Fig 3D ) and anatomical definitions of PAC ( S5 Fig; PAC was defined either as the center of the high-low-high gradient or as the center of anatomical region TE1 . 1 58 , in posteromedial Heschl’s gyrus ( HG ) ) ., To directly compare primary and nonprimary regions , we then averaged NSE values within the three bins nearest and farthest from PAC ( Fig 3D , inset ) ., This analysis revealed that responses to natural and model-matched sounds became more dissimilar in nonprimary regions in both the left and right hemisphere of every subject tested , leading to a highly significant difference between primary and nonprimary regions ( p < 0 . 01 via sign test for both hemispheres and for both tonotopic and anatomical definitions of PAC ) ., The gradient between primary and nonprimary regions was observed in both scanning paradigms , regardless of smoothing ( S3 Fig ) , and could not be explained by selectivity for intelligible speech ( a similar pattern was observed when intelligible speech sounds were excluded from the analysis; see S6 Fig ) ., These results also could not be explained by variations in voxel reliability across brain regions , both because our NSE measures were noise-corrected and because voxel responses were similarly reliable throughout primary and nonprimary regions ( S4C Fig ) ., As a consequence of the similar reliability across auditory cortex , the increase in the NSE between natural and model-matched sounds between primary and nonprimary regions was significantly greater than the change in voxel reliability ., This was true using both corrected and uncorrected values for the natural versus model-matched NSE , both tonotopic and anatomical definitions of PAC , and with reliability measured using just natural sounds ( for Paradigm I ) and both natural and model-matched sounds ( for Paradigm II ) ( p < 0 . 01 via sign test in all cases; see S3 Fig for a breakdown by paradigm ) ., Thus , our results demonstrate that the modulation filter bank model is worse at accounting for voxel responses in nonprimary regions ., We next used a similar approach to test whether responses in PAC could be explained by simpler models ., For example , if neurons in a voxel are tuned primarily to audio frequency , then all sounds with similar spectra should produce similar responses , regardless of their modulation properties ., To test such alternative models , we synthesized three new sounds for each natural sound ., Each synthetic sound was matched on a different subset of features from the full model ( Fig 4A ) ., One sound was synthesized to have the same marginal distribution of cochlear envelopes as a natural sound and , thus , a similar audio spectrum , but its modulation properties were otherwise unconstrained ., Another sound was constrained to have the same temporal modulation statistics within each cochlear frequency channel , computed using a bank of modulation filters modulated in time but not frequency ., A third sound was synthesized to have matched spectral modulation statistics , computed from a bank of filters modulated in frequency but not time ., All of the modulation-matched sounds also had matched cochlear marginal statistics , thus making it possible to test whether adding modulation structure enhanced the similarity of cortical responses to natural and model-matched sounds ., The results of this analysis suggest that all of the model features are necessary to account for voxel responses to natural sounds in PAC ( Fig 4B and 4C; S7 Fig ) ., Responses to model-matched sounds constrained just by cochlear statistics differed substantially from responses to natural sounds even in PAC , leading to significantly larger NSE values than those observed for the full model ( p < 0 . 001 in PAC via bootstrapping across subjects; see “Statistics” in Materials and methods ) ., Thus , even though PAC exhibits selectivity for frequency due to tonotopy , this selectivity only accounts for a small fraction of its response to natural sounds ., Responses to natural and model-matched sounds in PAC became more similar when the sounds were constrained by either temporal or spectral modulation properties alone ( NSE temporal < NSE cochlear: p < 0 . 001 via bootstrapping; NSE spectral < NSE cochlear: p < 0 . 001 ) ., However , we only observed NSE values near 0 when sounds were matched in both their temporal and spectral modulation properties ( NSE full model < NSE temporal: p < 0 . 001; NSE full model < NSE spectral: p < 0 . 001 ) ., These results provide further support for the idea that selectivity for both temporal and spectral modulation is a prominent feature of cortical tuning in PAC 7 , 32 , 33 ., In nonprimary auditory cortex , we also observed more similar responses when matching sounds on spectrotemporal modulation compared with simpler models ( NSE spectrotemporal < NSE cochlear: p < 0 . 001; NSE spectrotemporal < NSE temporal: p < 0 . 05; NSE spectrotemporal < NSE spectral: p < 0 . 01 ) ., However , the noise-corrected NSE values were high for all of the models tested , indicating that the modulation model fails to account for a substantial fraction of nonprimary responses ., Part of the motivation for using model-matched stimuli comes from the more common approach of predicting responses to natural stimuli from the features of a model ( e . g . , via linear regression ) ., As discussed above , good predictive accuracy is not sufficient to guarantee that the features of a model drive a neural response , due to the potential for correlations between different feature sets across natural stimuli ., Model matching provides one way to circumvent this issue , because the synthesized sounds are only constrained by the statistics of the particular model being tested ., Here , we test whether our approach yields novel insights compared with simply predicting cortical responses to natural sounds from model features ., We attempted to predict responses to the 36 natural sounds from time-averaged statistics of the same model features used to generate the model-matched sounds ( Fig 5A; see S8 Fig for individual subject prediction error maps for the full spectrotemporal model ) ., Specifically , we used ridge regression to predict voxel responses from the amplitude of each model feature’s response to each natural sound 7 , 16 , measured as the standard deviation across time ( for the cochlear model , we used the mean rather than the standard deviation because the features were the result of an envelope extraction operation , and the mean thus conveyed the amplitude of the filter’s response ) ., Because histogram matching approximately matches all time-averaged statistics of a distribution , predictions based on a single time-averaged statistic , such as the standard deviation , provide a conservative estimate of the predictive power of time-averaged statistics ., Good predictions in voxels whose responses to model-matched sounds deviated from those to natural sounds would thus suggest that prediction-based analyses overestimate the model’s explanatory power ., We quantified prediction accuracy by measuring the NSE between measured and predicted responses for left-out sounds that were not used to learn the regression weights ( see “Model predictions” in Materials and methods ) ., Overall , we found that voxel responses to natural sounds were substantially more similar to the predicted model responses than to the measured responses to the model-matched stimuli ( Fig 5B and 5C ) , leading to smaller NSEs for model predictions compared with model-matched stimulus responses ., This difference was particularly pronounced in nonprimary regions , where we observed relatively good predictions from the full two-stage model despite highly divergent responses to model-matched sounds , leading to a significant interaction between the type of model evaluation ( model prediction versus model matching ) and region ( primary versus nonprimary ) ( p < 0 . 01 via sign test for both tonotopic and anatomical definitions of PAC; a sign test was used to evaluate whether the change in NSE values between primary and non-primary regions was consistently larger for model matching compared with model prediction ) ., Because the natural and model-matched sounds were matched in the features used for prediction , the divergent responses to the two sound sets imply that the features used for prediction do not in fact drive the response ., Thus , good predictions for natural sounds in the presence of divergent model-matched responses must reflect the indirect influence of correlations between the features of the model and the features that actually drive the neuronal response ., Model matching thus reveals a novel aspect of functional organization not clearly evident from model predictions by demonstrating the failure of the filter bank model to account for nonprimary responses ., Our prediction analyses were based on responses to a set of 36 natural sounds that was smaller than the sound sets that have been used elsewhere to evaluate model predictions 7 , 16 , 45 , 60 ., Because our analyses were cross-validated , small sound sets should reduce prediction accuracy and thus cannot explain our finding that model predictions were better than would be expected given responses to model-matched sounds ., Nonetheless , we assessed the robustness of our findings by also predicting responses to a larger set of 165 natural sounds 45 ., We observed similar results with this larger sound set , with relatively good prediction accuracy for the full spectrotemporal model throughout primary and nonprimary auditory cortex ( S9 Fig ) ., Another way to assess the utility of the model-matching approach is to train a model to predict natural sounds , and then test its predictive accuracy on model-matched sounds ( and vice versa ) ., In practice , this approach yielded similar results to directly comparing responses to natural and model-matched sounds: good cross-predictions in PAC but poor cross-predictions in nonprimary auditory cortex ( S10 Fig ) ., This observation is expected given that, ( a ) the model predictions for natural sounds were good throughout auditory cortex and, ( b ) responses to natural and model-matched sounds diverged in nonprimary regions , but it provides a consistency check of the two types of analyses ., All of our analyses described thus far were performed on individual voxels , summarized with maps plotting the NSE between each voxel’s response to natural and model-matched sounds ., However , these error maps do not reveal in what respect the responses to natural and model-matched sounds differ , and , because of the large number of voxels , it is not feasible to simply plot all of their responses ., We previously found that voxel responses to natural sounds can be approximated as a weighted sum of a small number of canonical response patterns ( components ) 45 ( Fig 6A ) ., Specifically , six components explained over 80% of the noise-corrected response variance to a diverse set of 165 natural sounds across thousands of voxels ., We thus used these six components to summarize the responses to natural and model-matched sounds described here ., This analysis was possible because many of the subjects from this experiment also participated in our prior study ., As a consequence , we were able to learn a set of voxel weights that reconstructed the component response patterns from our prior study and then apply these same weights to the voxel responses from this experiment ( see “Voxel decomposition” in Materials and methods ) ., We found that all six components exhibited reliable responses to the natural sounds from this experiment ( Fig 6B ) ., Two of the components ( 5 and 6 ) responded selectively to speech and music , respectively , replicating the selectivity we found previously ( last two columns of 6B ) ., Critically , responses to the model-matched sounds were much weaker in these speech- and music-selective components , even for sounds matched on the full model ( Fig 6C , last two columns; see S11 Fig for sounds matched on subsets of model features ) , leading to high NSE values ( speech NSE = 0 . 45; music NSE = 0 . 55 for the full model , noise-corrected; Fig 6D ) ., By contrast , the other four components , all of which overlapped PAC to varying extents , responded similarly to natural and model-matched sounds constrained by the full model , leading to smaller errors ( NSE for Component 1: 0 . 06 , Component 2: 0 . 12 , Component 3: 0 . 26 , Component 4: 0 . 19 ) than those for the speech- and music-selective components ( p < 0 . 001 for all direct comparisons between the speech- and music-selective components and Components 1 , 2 , and 4; for Component 3 , which had the lowest test-retest reliability , the direct comparison with the music-selective component was significant , p < 0 . 01 , and the direct comparison with the speech-selective component was nearly significant , p = 0 . 076; statistics computed via bootstrapping across subjects ) ., These results indicate that selectivity for music and speech cannot be purely explained by standard acoustic features that nonetheless account for much of the voxel response in primary regions ., Our model-matching approach posits that responses should be exactly matched if the model is accurate ., If the model is not accurate , the approach makes no prediction about how the responses should differ ., Nonetheless , the divergent responses to natural and model-matched sounds in Components 5 and 6 appeared to be largely driven by weaker responses to the model-matched sounds ., We verified this observation by comparing the standard deviation of responses to natural and model-matched sounds: the response variation for model-matched sounds decreased sharply in Components 5 and 6 , driven by lower overall responses to the model-matched sounds ( Fig 6E ) ., In contrast , the noise-corrected correlation remained high ( Fig 6F ) ., A similar pattern was also evident in whole-brain maps ( S12 Fig ) : the variation in voxel responses to model-matched sounds constrained by the full model dropped in nonprimary regions ( driven by lower responses to the model-matched stimuli ) , while the correlation remained high ., For Components 5 and 6 , the high correlations were driven by the fact that model-matched sounds from the component’s preferred category produced a higher response than model-matched sounds from other categories ( as is evident in Fig 6C ) ., For example , in Component 6

Introduction, Results, Discussion, Materials and methods

A central goal of sensory neuroscience is to construct models that can explain neural responses to natural stimuli ., As a consequence , sensory models are often tested by comparing neural responses to natural stimuli with model responses to those stimuli ., One challenge is that distinct model features are often correlated across natural stimuli , and thus model features can predict neural responses even if they do not in fact drive them ., Here , we propose a simple alternative for testing a sensory model: we synthesize a stimulus that yields the same model response as each of a set of natural stimuli , and test whether the natural and “model-matched” stimuli elicit the same neural responses ., We used this approach to test whether a common model of auditory cortex—in which spectrogram-like peripheral input is processed by linear spectrotemporal filters—can explain fMRI responses in humans to natural sounds ., Prior studies have that shown that this model has good predictive power throughout auditory cortex , but this finding could reflect feature correlations in natural stimuli ., We observed that fMRI responses to natural and model-matched stimuli were nearly equivalent in primary auditory cortex ( PAC ) but that nonprimary regions , including those selective for music or speech , showed highly divergent responses to the two sound sets ., This dissociation between primary and nonprimary regions was less clear from model predictions due to the influence of feature correlations across natural stimuli ., Our results provide a signature of hierarchical organization in human auditory cortex , and suggest that nonprimary regions compute higher-order stimulus properties that are not well captured by traditional models ., Our methodology enables stronger tests of sensory models and could be broadly applied in other domains .

Modeling neural responses to natural stimuli is a core goal of sensory neuroscience ., A standard way to test sensory models is to predict responses to natural stimuli ., One challenge with this approach is that different features are often correlated across natural stimuli , making their contributions hard to tease apart ., We propose an alternative in which we compare neural responses to a natural stimulus and a “model-matched” synthetic stimulus designed to yield the same responses as the natural stimulus ., We tested whether a standard model of auditory cortex can explain human cortical responses measured with fMRI ., Model-matched and natural stimuli produced nearly equivalent responses in primary auditory cortex , but highly divergent responses in nonprimary regions , including those selective for music or speech ., This dissociation was not evident using model predictions because of the influence of feature correlations in natural stimuli ., Our results provide a novel signature of hierarchical organization in human auditory cortex , and suggest that nonprimary regions compute higher-order stimulus properties that are not captured by traditional models ., The model-matching methodology could be broadly applied in other domains .

auditory cortex, acoustics, medicine and health sciences, statistical noise, diagnostic radiology, functional magnetic resonance imaging, engineering and technology, statistics, signal processing, matched filters, brain, neuroscience, signal filtering, magnetic resonance imaging, mathematics, forecasting, brain mapping, neuroimaging, research and analysis methods, sensory physiology, imaging techniques, animal cells, mathematical and statistical techniques, bioacoustics, research assessment, physics, auditory system, cellular neuroscience, radiology and imaging, diagnostic medicine, anatomy, cell biology, research validity, physiology, neurons, biology and life sciences, sensory systems, cellular types, physical sciences, statistical methods

journal.pbio.1001985

PI3K Signaling and Stat92E Converge to Modulate Glial Responsiveness to Axonal Injury

Glial cells are extraordinarily sensitive to disruptions in central nervous system ( CNS ) homeostasis and exhibit an impressive ability to respond to a diversity of neural injuries including hypoxia , chemical insults , and mechanical injury ( e . g . , axotomy or traumatic brain injury ) 1–3 ., During injury responses , glia exhibit robust changes in gene expression , migrate or extend membranes to sites of trauma , and phagocytose degenerating neuronal debris ., Glial reactive responses after injury can be beneficial and promote recovery ., For example , glial clearance of degenerating neuronal debris is thought to suppress nervous system inflammation and facilitate remyelination 4–8 ., However , glia can also exacerbate damage in the CNS by driving inflammation through the release of pro-inflammatory cytokines and actively destroying healthy cells 9–14 ., Whether reactive gliosis is ultimately more beneficial or harmful to the nervous system remains an open question and is likely context dependent ., However , mounting evidence points to reactive glial cells as an excellent target for therapeutically modifying CNS disorders 15 ., Reactive glial cells have been studied in a variety of neurodegenerative diseases and injury models but surprisingly little is known about how glial sensitivity to neuronal health is initially established in the healthy brain , or how glial perception of neuron-derived “injury signals” is translated into dynamic reactive responses after trauma ., A wealth of studies indicate that reactive gliosis is not an all-or-none response , but rather includes a wide range of graded responses that correlate with the severity of nervous system injury 15 ., The ability of glia to “measure” the severity of trauma in the nervous system and respond accordingly implies a tight functional coupling between signaling pathways sensing neural injury and those executing glial reactive responses 16 ., In some cases reactive glia can modulate the intensity of reactive gliosis in an autocrine way ., For example , focal demyelination leads to activation and release of endothilin-1 ( ET-1 ) from astrocytes , which in turn can signal back to astrocytes ( through ET-1 receptors ) and further enhance astrocyte cellular hypertrophy , proliferation , and GFAP expression 17–22 ., However , the initial injury signal which activates ET-1 expression remains to be identified ., An intriguing possibility is that so-called “eat me” cues on degenerating neuronal debris directly activate receptors on glial cells , which in turn modulate initiation of reactive gliosis , but evidence for such molecular regulation is lacking ., We have previously shown that Draper , a glial-expressed immunoreceptor-like molecule , is a key surface receptor required for glial engulfment of degenerating axons after axotomy in the adult Drosophila brain 23–25 ., Draper is the Drosophila ortholog of CED-1 , an engulfment receptor that is essential for engulfment of cell corpses in Caenorhabditis elegans 26 ., Within hours after axotomy of Drosophila antennal olfactory receptor neuron ( ORN ) axons , Draper protein , and mRNA levels are dramatically increased in glia surrounding degenerating axons ., Glial membranes are then recruited to severed axons , glia phagocytose axonal debris , and ultimately glia return to a resting state 24 , 27 ., Loss of Draper function blocks all glial morphological and molecular responses to axonal injury and axonal debris lingers in the brain for the life of the animal ., Similar phenotypes have been observed when the components of the Src-family kinase signaling cascade that acts downstream of Draper ( i . e . , Shark and Src42a ) are eliminated specifically from glial cells 25 ., These data argue that Draper acts very early in the activation of Drosophila glia after axonal injury , perhaps even in the recognition of cues presented by engulfment targets like degenerating axons ., Recently the mammalian orthologs of CED-1/Draper , MEGF10 , and Jedi , have been implicated in satellite glial engulfment of neuronal cell corpses in developing mouse dorsal root ganglia 28 and MEGF10 has been shown to engulf pruned synapses in the postnatal dorsal lateral geniculate nucleus during activity-dependent synaptic refinement 29 ., Thus , Draper/MEGF10/Jedi engulfment signaling appears to be a conserved feature of glial cells in evolutionarily distant species ., To further explore the molecular mechanisms by which glial cells establish competence to respond to axonal injury and dynamically regulate reactive responses , we performed an in vivo RNAi screen for novel signaling molecules required for glial engulfment of degenerating ORN axons in Drosophila ., We identified PI3K signaling and the Stat92E transcription factor as important regulators of glial Draper expression ., Interestingly , both of these pathways were necessary for the expression of Draper in the resting , uninjured brain ., However , while PI3K signaling was dispensable for injury-induced up-regulation of Draper , we found that Stat92E was necessary in glial cells for both injury-induced up-regulation of Draper and clearance of degenerating axonal debris ., Surprisingly , we find that Stat92E acts downstream of Draper to activate transcription ., We propose a simple model for glial activation after axotomy whereby Draper signaling is stimulated in a graded fashion according to the level of axonal debris ( i . e . , the severity of axonal injury ) , which in turn promotes Stat92E-dependent changes in glial gene expression in a way that is proportional to the strength of Draper pathway signaling ., Such a mechanism places levels of glial activation directly downstream of the total amount of axonal debris present in the adult brain ., In order to identify new glial engulfment genes we performed an in vivo RNAi screen using a previously established assay 24 ., Briefly , ∼300 candidate engulfment genes were knocked down specifically in glia by driving UAS-regulated RNAi constructs with the glial specific repo-Gal4 driver ., To assay the ability of glia to clear degenerating axonal debris we labeled a subset of maxillary palp ORNs with green fluorescent protein ( GFP ) , severed the axons by surgically removing the maxillary palps , and assayed for clearance of GFP-labeled ORN axonal debris at various timepoints ., Interestingly , we found glial-specific knockdown of pi3k92e , raptor , or pdk-1—key components of the phosphoinositide 3-kinase ( PI3K ) signaling pathway—led to a decrease in glial clearance of axonal debris 5 days after axotomy ( Figure 1A and 1B; Data S1 ) ., Similar results were found when we drove glial expression of a dominant-negative version of PI3K92e ( UAS-pi3k92eDN; Figure S1; Data S2 ) ., We note that while axon clearance was delayed in these backgrounds , ultimately all axonal debris was cleared from the brain 7–10 days after axotomy ( unpublished data ) , indicating that glia exhibited a delay in clearance rather than a blockade ., To further explore the cellular basis for this delayed axon clearance phenotype , we assayed glial expression of the engulfment receptor Draper ., Surprisingly , glial knockdown of pi3k92e , raptor , or pdk1 , or over expression of pi3k92eDN reduced Draper expression significantly in the uninjured brain ( Figures 1C and S1; Data S3 ) ., Reductions in Draper protein levels were confirmed and quantified on Western blots ( Figures 1D , 1E , and S1; Data S3 ) ., Reciprocally , we found that over-expression of a constitutively activate version of PI3K92E ( UAS-pi3k92eCAAX ) in glia led to a dramatic increase in Draper levels in uninjured brains ( Figure 1C–1E; Data S3 ) ., Thus , Draper expression was tightly correlated with glial PI3K signaling levels in the healthy uninjured brain ., To our knowledge this is the first pathway shown to modulate the establishment of Draper expression levels in glia ., While maxillary palps house ∼60 ORN cell bodies and ablation of maxillary palps results in a modest increase in Draper protein expression , more severe injury of ORN axons by removal of antennae ( which house ∼600 ORN cell bodies ) results in a dramatic up-regulation of Draper protein and mRNA levels 24 , 27 ., To determine whether this response was normal when the PI3K pathway was compromised , we ablated antennae in control , pi3k92eRNAi , raptorRNAi , and pdk1RNAi backgrounds ., Despite the fact that knockdown of pi3k92e , raptor , or pdk1 significantly reduced Draper levels in the brains of uninjured control animals , we found that antennal ORN axotomy induced significant up-regulation of Draper levels in glia surrounding the antennal lobe ( Figure 1C ) ., While it is possible that this injury-induced Draper up-regulation could be the result of incomplete RNAi mediated knockdown of components of the PI3K signaling pathway , on the basis of the consistency in results among the different components of the pathway we favor the notion that basal levels of Draper ( i . e . , those present in the healthy brain ) are regulated by PI3K signaling and injury-induced up-regulation of Draper is regulated by alternate signaling pathways ., Our observations that basal and injury induced Draper expression were controlled by distinct molecular pathways prompted us to attempt to identify draper gene enhancer elements responsible for establishing basal levels of draper expression in adult brain glia , and/or increasing draper expression specifically after ORN axotomy ., We focused our search on an ∼40 kb region centered around the draper locus ( Figure 2A ) ., We cloned nine different potential draper enhancer elements ( termed dee2-dee10 ) from 5′ , intronic , or 3′ regions of the draper gene into the Gal4-based pBGW vector 30 and inserted these elements into identical genomic locations ( Figure 2A ) ., Each dee-Gal4 line was then used to drive two copies of UAS-mCD8::GFP in vivo and expression patterns were examined in the adult brain before and after injury ., No expression in ensheathing or cortex glia was observed with any of the enhancer element lines in the healthy , uninjured brain ( unpublished data ) ., We therefore failed to identify any single enhancer element that was capable of driving glial expression of reporters in the adult brain in a pattern similar to endogenous Draper protein ., This observation suggests that PI3K-dependent regulation of Draper levels might be governed by an enhancer element some distance from the draper gene , requires the convergent activity of multiple enhancers along the draper gene , or could be controlled through post-transcriptional mechanisms ., To determine whether any of these potential DEEs were responsive to axotomy , we ablated antennae or maxillary palps and assayed reporter activation in glia one day after injury ., We did not observe glial expression after axonal injury with dee2-6- or dee8-10-Gal4 lines ( unpublished data ) ., However , one day after antennal ORN axotomy , we observed a striking increase in glial expression of mCD8::GFP in the dee7-Gal4 reporter background , and GFP levels were further increased four days after axon injury ( Figure 2B ) ., We note that in uninjured animals we observed low level expression of this element in astrocyte-like glia , which were randomly distributed in the neuropil ( Figure 2B ) but it did not drive GFP expression prior to axon injury in ensheathing or cortex glia , those adult brain glia which normally express Draper 23 ., Interestingly , not only was injury-induced reporter expression strong in ensheathing glia surrounding the antennal lobe—those glia that normally engulf degenerating ORN axons 23—but the reporter expression also increased robustly in cortex glia throughout the brain ( Figure 2B and 2D ) ., This widespread activation of the reporter supports the notion that glia , even at locations distant from the injury site , can respond molecularly to axonal damage ., Notably , we found the activation of the dee7-Gal4 element appeared to scale with the severity of the axonal injury: when we ablated one maxillary palp ( ∼60 ORNs ) , two maxillary palps ( ∼120 ORNs ) , one antenna ( ∼600 ORNs ) , or both antennal ( ∼1 , 200 ORNs ) , we observed a correlated increase in dee7-Gal4-driven mCD8::GFP ( Figure 2D ) ., We further note that ablation of maxillary palps , whose axons are found in the maxillary nerve and ventro-medial regions of the antennal lobe , resulted in a much reduced and more localized increase in dee7-Gal4-driven mCD8::GFP in cortex glia located in the ventral region of the antennal lobe ( Figure 2D ) ., Sequence analysis of the 2619 bp dee7 element led to the discovery of three consensus Stat92E binding sites ( TTC3n/4nGAA ) 31 , the sole member of the signal transducer and activator of transcription ( STAT ) family of molecules in Drosophila 31 , 32 ., Of these three Stat92E sites , two were also present in dee6-Gal4 , which was not responsive to axonal injury ( Figure 2A ) ., We mutated the Stat92E binding site specific to the dee7-Gal4 element , integrated this dee7MUT-Gal4 construct into the same genomic location used for the previously generated reporter lines , and examined its responsiveness to axonal injury ., While baseline levels of mCD8::GFP expression were similar to dee7-Gal4 , dee7MUT-Gal4 exhibited an ∼30%–40% decrease in the injury-induced expression of mCD8::GFP at both 1 and 4 days after axotomy ( Figure 2B and 2C; Data S4 ) ., In contrast , simultaneous mutation of the two other Stat92E binding sites within the dee7 element had no effect on transcriptional activation of the dee7-Gal4 reporter ( Figure S2A–S2C; Data S5 ) ., From these data we conclude that dee7 contains a glial transcriptional regulatory element that is responsive to axonal injury , and our data suggest that at least one Stat92E binding site is required for maximal activation of this element after axotomy ., We next sought to determine whether Stat92E was required for modulating glial responses to axonal injury and clearance of degenerating axonal debris ., Stat92E function was knocked down specifically in glia by expressing a UAS-stat92eRNAi construct with the pan-glial repo-Gal4 driver ., In controls the majority of GFP+ maxillary palp ORN axon material was cleared from the brain 5 days after injury ( Figure 3A–3C; Data S6 ) ., However , in stat92eRNAi animals , GFP+ axonal debris persisted 5 days after injury ( Figure 3A–3C; Data S6 ) ., In contrast to depletion of the PI3K signaling cascade , glial stat92eRNAi suppressed clearance of antennal ORN axons even 15 days after injury ( Figure S3A ) , indicating an essential requirement for STAT92E in glial clearance of axonal debris ., We were able to confirm the UAS-Stat92ERNAi line efficiently targets stat92e as glial co-expression of a GFP-tagged stat92e molecule with the stat92eRNAi construct eliminated all Stat92E-GFP expression compared to controls ( Figure S3B ) ., These observations argue that Stat92E is an important regulator of glial engulfment activity in the adult brain ., On the basis of our identification of a Stat92E-dependent injury-responsive element in the draper gene , we predicted Stat92E would modulate glial phagocytic activity by regulating draper expression after axotomy ., Draper is normally expressed in ensheathing and cortex glia throughout the uninjured brain and is up-regulated around the antennal lobe after antennal ORN axotomy ( Figure 3D ) ., Glial-specific knockdown of Stat92E led to nearly undetectable levels of Draper expression in the adult brain even prior to injury ( Figure 3D and 3E; Data S7 ) ., We confirmed this widespread loss of Draper by performing Western blots on dissected adult brains from control , stat92eRNAi , and draperRNAi animals ( Figures 3F , 3G , and S3; Data S8 ) ., In contrast to loss of PI3K signaling , knockdown of Stat92E in glia was also sufficient to potently suppress glial activation of draper after axotomy: stat92eRNAi animals exhibited no detectable increase in Draper levels after antennal ablation compared to controls ., In addition , while Draper was localized specifically to severed ORN axons after maxillary palp ablation in controls , we found no detectable Draper localization to severed axons in stat92eRNAi animals ( Figure 3D and 3E; Data S7 ) ., These data suggest that Stat92E is required to establish normal basal levels of Draper expression in glia , and dynamically regulates increased Draper levels as glia respond to axotomy ., STAT signaling is involved in multiple cellular processes including cell survival , differentiation , motility , and immunity 33–42 ., To exclude the possibility that the defects we observed in stat92eRNAi animals were the result of abnormalities in glial cell development we used the conditional Gal80ts system to specifically activate stat92eRNAi at adult stages ., When temperature sensitive stat92eRNAi animals were raised and tested at 18°C , we found that glia efficiently cleared axonal debris and expressed normal levels of Draper ( Figure S4; Data S9 ) ., However , when they were shifted to and tested at the restrictive temperature during adult stages ( thereby activating the RNAi construct only after development was complete ) , we found that stat92eRNAi animals exhibited reduced expression of Draper and failed to clear degenerating axons ( Figure S4; Data S9 ) ., Glial cell morphology ( visualized with membrane-tethered GFP ) and numbers ( counted with α-Repo antibody nuclear staining ) appear grossly normal in these animals , arguing that these phenotypes are not the result of glial cell loss in stat92eRNAi backgrounds ( Figure S5A and S5B ) ., Moreover , we found that adult-specific activation of the RNAi was reversible , as shifting these animals back to 18°C ( thereby turning the RNAi off ) re-established normal levels of Draper and initiated clearance of axonal debris ( Figure S4; Data S9 ) ., Thus Stat92E functions in adult brain glia , where it modulates Draper expression and glial phagocytosis of degenerating axons ., Draper and the PTB domain-containing protein dCed-6 are both required for glial engulfment of degenerating axons and are expressed in glial cells in the adult brain 23 ., To determine whether Stat92E broadly regulates engulfment gene expression we assayed dCed-6 levels in the adult brain in animals expressing stat92eRNAi in glia and found that dCed-6 was still present at high levels throughout the brain ( Figure S5C and S5D ) ., Interestingly , while dCed-6 levels appeared grossly normal in a stat92eRNAi background , dCed-6 was not recruited to severed maxillary palp axons 1 day after axotomy ( Figure S5C ) ., Thus , while Stat92E is necessary for dCed-6 recruitment to severed axons ( i . e . , glial responses to injury ) , basal levels of dCed-6 do not appear to be regulated by a Stat92E dependent mechanism ., Finally , we sought to determine whether Draper levels were regulated transcriptionally by STAT92E and/or PI3K signaling ., We performed quantitative real-time PCR to measure draper transcript levels in dissected brains from control animals and animals expressing glial draperRNAi , pi3k92eCAAX ( gain-of-function ) , pi3k92eRNAi , or stat92eRNAi ., Consistent with STAT92E regulating draper at the transcriptional level , we found that glial RNAi for stat92e severely reduced draper transcripts to a level comparable to depletion with draperRNAi ( Figure 3H; Data S10 ) ., In contrast , neither loss- or gain-of-function manipulation of PI3K92E resulted in a statistically significant difference in draper transcript levels ., While this argues for a post-transcriptional mechanism of Draper regulation by PI3K signaling , we cannot rule out the possibility that PI3K at least partially regulates draper at the transcriptional level considering loss of PI3K resulted in a trend of decreased ( ∼40% ) draper mRNA ( Figure 3H; Data S10 ) ., To explore the dynamics of Stat92E signaling in adult brain glia we examined the expression patterns of transcriptional reporters for Stat92E activity 43 ., These reporters have been previously shown to accurately reflect Stat92E transcriptional activity during development as well as in the adult 43–46 ., We first used the 10XStat92E-GFP , which harbors ten Stat92E binding sites driving expression of enhanced GFP ., In co-stains with α-Draper and α-Repo ( a glial nuclear marker ) we observed quite specific glial expression of the Stat92E reporter in uninjured controls ( Figure 4A ) ., Moreover , after ablation of antennae we found strong GFP labeling of antennal lobe glia , and the GFP signal completely overlapped with Draper ( Figure 4A ) ., After ablation of maxillary palps we found GFP signals co-localized with Draper in glomeruli housing severed axons ( Figure 4A ) ., These data argue that Stat92E is active at a transcriptional level in adult brain glia ., The GFP driven by the 10XStat92E-GFP reporter is quite stable and can perdure in cells for ∼20 hours after activation , which precludes our use of this construct to examine dynamic changes in Stat92E transcriptional activity ., We therefore used a second reporter , 10XStat92E-dGFP , which drives the expression of a rapidly degraded , destabilized GFP ( dGFP ) , thereby allowing for increased temporal resolution of Stat92E activity ., Prior to injury , we were unable to detect any activation of this Stat92E transcriptional reporter in adult brains ( Figure 4B ) ., However , beginning ∼16 hours after antennal ablation we detected 10XStat92E-dGFP expression in cells surrounding the antennal lobe ( Figure 4B , arrows ) ., GFP intensity peaked at ∼24 hours after antennal ablation and disappeared by 48 hours after axotomy ( Figure S6 ) ., To confirm that activation of the 10XStat92E-dGFP reporter after axotomy is Stat92E-dependent and glial specific , we knocked down Stat92e specifically in glia , severed axons , and assayed 10XStat92E-dGFP activity ., We found that glial-specific knockdown of Stat92E completely suppressed the axotomy-induced activation of the 10XStat92E-dGFP transcriptional reporter ( Figure 4B ) ., Consistent with our observations of widespread activation of the dee7-Gal4 driver in glia throughout the brain after injury , we also found that axonal injury led to broad activation of the 10XStat92E-dGFP reporter in glial cells ( Figure 4C ) and expression of these two reporters colocalize in glial cells after axotomy ( Figure 4D ) , indicating they are active in the same cells ., Together these data indicate that Stat92E can transiently increase the transcriptional activation of target genes in glial cells throughout the brain after axonal injury , with the strongest increases in target gene activation occurring adjacent to injury sites ., STAT activity is generally regulated by the JAK signaling platform , and this pathway is conserved in all higher metazoans ., The Drosophila JAK/STAT signaling pathway consists of a single JAK molecule , Hopscotch ( hop ) 47 , and the cytokine like receptor Domeless ( Dome ) 48 ., To determine if Stat92E-dependent activation of draper is mediated through canonical JAK/STAT signaling we drove RNAi constructs targeted against hop , and a dominant negative Domeless molecule , DomelessΔCYT 49 , in glial cells , and assayed Draper expression and clearance of severed axons ., Surprisingly , in each of these backgrounds we found Draper levels were similar to control animals and axons were efficiently cleared 5 days after injury ( Figure S7A and S7B; Data S11 ) ., Reciprocally , we found a gain-of-function allele of hop , hopTUM , which has been shown in numerous assays to activate Stat92E transcriptional activity 43 , 50 , 51 , failed to activate the10XStat92E-dGFP reporter in adult brain glia ( Figure S7C ) ., Notably , expression of an activated version of Stat92E , Stat92EΔNΔC 52 , led to strong activation of the 10XStat92E-dGFP reporter ( Figure S7C ) ., Stat92EΔNΔC has previously been shown to require phosphorylation at Y711 for activation 52 ., We therefore expressed a version of Stat92EΔNΔC with a Y711F mutation in glia , and found it was insufficient for activation of the 10XStat92E-dGFP reporter ( Figure S7C ) ., Together these data argue that while phosphorylated Stat92E mediates activation of Stat92E reporters in glia , canonical JAK/STAT signaling is neither necessary nor sufficient to activate Stat92E-dependent glial responses to axon injury ., To date Domeless is the only receptor known to positively regulate Stat92E signaling under normal physiological conditions 48 , 53 , 54 ., Considering we were unable to demonstrate a role for Domeless signaling in activation of Draper after injury , we sought to determine whether Draper itself might have a role in modulating glial gene expression ., We assayed 10XStat92E-dGFP activation after axonal injury in draperΔ5 null mutants and , intriguingly , loss of Draper resulted in a complete lack of 10XStat92E-dGFP activation after axotomy ( Figure 5A ) ., Consistent with a direct requirement for Draper signaling in STAT92E-dependent activation of draper after axotomy , we also found a lack of activation of the dee7-Gal4 reporter after axonal injury in draperΔ5 animals ( Figure 5B ) ., These data demonstrate a direct role for the Draper receptor in modulating Stat92E-dependent changes in glial gene expression following local axonal injury ., We next explored whether other identified components of the Draper signaling pathway modulate Stat92E transcriptional activity after axon injury ., Draper is thought to be phosphorylated by Src42a upon activation , initiating binding of the non-receptor tyrosine kinase Shark , which together , with Rac1 and dCed-6 , promote engulfment 23 , 25 ., Interestingly , we found that glial-specific knockdown of Shark , Src42a , or Rac1 blocked injury-induced activation of the10XSTAT92E-dGFP transcriptional reporter ( Figures 5A and S8A ) ., This finding was confirmed using multiple RNAi lines and/or dominant negative alleles for each gene in the pathway ( Figure S8A ) ., However , while RNAi mediated knockdown of dCed-6 efficiently eliminated dCed-6 immunoreactivity ( Figure S8B ) , axotomy-induced activation of 10XSTAT92E-dGFP was still detectable in dCed-6RNAi animals ( Figure 5A ) ., Thus Draper , Src42a , Shark , and Rac1 , but not dCed-6 , are essential for Stat92E-dependent activation of transcriptional targets in glia responding to axonal injury ., These are the first data that demonstrate a functional divergence between dCed-6 and Draper/Src42a/Shark function during engulfment signaling ., Since Src42a is known to signal downstream of Draper and has been shown to mediate Stat92E signaling in multiple contexts 25 , 55 , 56 , we sought to determine whether Src42a activity was sufficient to activate Stat92E transcriptional reporters ., Indeed , expression of a constitutively active Src42a molecule ( Src42aCA ) resulted in robust 10XStat92E-dGFP reporter activation throughout brain ( Figure 5C , refer to Figure 4B for control ) ., Based on our previous genetic studies it appears that Src42a acts through modulating Rac1 activity ., We therefore drove glial expression of a constitutively active Rac1 molecule ( Rac1v12 ) and found this also robustly activated the 10XStat92E-dGFP reporter throughout the adult brain ( Figure 5C , refer to Figure 4B for control ) ., These data are consistent with the notion that Src42a acts in glia downstream of Draper to activate Stat92E signaling through Rac1 after axonal injury ., In summary , our data suggest that axonal injury leads to Stat92E-dependent transcriptional up-regulation of draper through a Draper/Src42a/Shark/Rac1-dependent signaling cascade ., While activation of Stat92E , Src42A , or Rac1 led to robust activation of the 10XStat92E reporter , none were sufficient to increase basal Draper levels ( unpublished data ) , suggesting additional transcriptional inputs are required for injury-induced transcriptional activation of draper ., Together our data argue that draper transcription is regulated by Stat92E after axotomy ., Stat92E may regulate many genes after axotomy , or only a few critical targets essential for engulfment ., To further explore the relationship between Stat92E and the draper gene , we over-expressed Draper in a stat92eRNAi background to determine whether resupplying Draper was sufficient to overcome the engulfment deficit observed in Stat92E knockdown animals ., Remarkably , over-expression of Draper in stat92eRNAi animals led to a near complete rescue of the engulfment defect ( Figure 6A and 6B; Data S12 ) ., As an alternate method to increase Draper levels we also drove glial expression of the activated PI3K molecule , PI3K92eCAAX , in the presence of stat92eRNAi or control draperRNAi animals ., We found that activation of PI3K signaling was sufficient to increase Draper levels ( Figures 6E , 6F , and S9; Data S13 ) and rescue engulfment defects in stat92ERNAi backgrounds ( Figure 6C and 6D; Data S14 ) ., Consistent with the ability of PI3K signaling to drive Draper expression independently of Stat92E , we also found that activated PI3K signaling was not sufficient to activate expression of the 10XStat92edGFP reporter ( Figures 6E , 6F , and S9 ) ., However , PI3K signaling appeared to sensitize brain glia to injury as glial PI3K92eCAAX enhanced 10XStat92E-dGFP activation after antennal ablation compared to controls ( Figure S9 ) , perhaps through driving increased Draper levels ., From these data we conclude that draper is a critical target of Stat92E during glial responses to axonal injury ., In addition , our data argue that activated PI3K signaling results in Stat92E-independent increases in Draper levels ., Our work identifies two new signaling pathways important for regulating glial engulfment function in vivo ., First , we show that the PI3K signaling pathway modulates Draper levels in the healthy , uninjured brain as reduced PI3K signaling leads to dramatically decreased glial Draper and constitutive activation of PI3K signaling leads to Draper up-regulation ., However , depletion of PI3K signaling components delays but does not completely block the ability of glial cells to up-regulate Draper levels or clear axonal debris in response to axotomy , perhaps due to positive signaling through the small amount of Draper that remains under these conditions ., Second , we identify Stat92E as a potent regulator of both basal and injury induced Draper levels in adult brain glia ., Loss of Stat92E in mature glia results in significantly decreased draper transcript levels and a near complete loss of Draper protein in the uninjured brain ., What is the relationship between PI3K signaling and Stat92E in regulating basal levels of Draper in the healthy brain ?, On the basis of our analysis of draper mRNA levels we speculate that STAT92E regulates draper at least in part at the transcriptional level ., It appears unlikely that Stat92E functions downstream of PI3K signaling since loss of Stat92E in a constitutively activated PI3K background did not suppress PI3K-dependent increases in Draper levels , and gain-of-function PI3K was sufficient to rescue reduced Draper levels and engulfment defects in stat92eRNAi animals ., STAT molecules have been shown to be capable of acting as adaptor molecules for receptors that ultimately lead to activation of PI3K signaling 57–59 , therefore Stat92E could function in adult brain glia upstream of PI3K signaling in a non-transcriptional manner to regulate basal levels of Draper ., Finally , Stat92E might transcriptionally regulate key molecules required for activation or execution of PI3K signaling ., In such a situation Stat92E and PI3K signaling could modulate Draper levels through parallel mechanisms , but both would be required for expression of appropriate levels of Draper in the adult brain ., A role for PI3K signaling in phagocytic function appears to be conserved from Drosophila to mammals ., Activation of PI3K signaling occurs downstream of the Fcγ receptor 60–63 ., This finding is intriguing in light of the fact that Draper appears to act as an ancient immunoreceptor , activating a Src family kinase signaling cascade through ITAM/ITIM-dependent m

Introduction, Results, Discussion, Materials and Methods

Glial cells are exquisitely sensitive to neuronal injury but mechanisms by which glia establish competence to respond to injury , continuously gauge neuronal health , and rapidly activate reactive responses remain poorly defined ., Here , we show glial PI3K signaling in the uninjured brain regulates baseline levels of Draper , a receptor essential for Drosophila glia to sense and respond to axonal injury ., After injury , Draper levels are up-regulated through a Stat92E-modulated , injury-responsive enhancer element within the draper gene ., Surprisingly , canonical JAK/STAT signaling does not regulate draper expression ., Rather , we find injury-induced draper activation is downstream of the Draper/Src42a/Shark/Rac1 engulfment signaling pathway ., Thus , PI3K signaling and Stat92E are critical in vivo regulators of glial responsiveness to axonal injury ., We provide evidence for a positive auto-regulatory mechanism whereby signaling through the injury-responsive Draper receptor leads to Stat92E-dependent , transcriptional activation of the draper gene ., We propose that Drosophila glia use this auto-regulatory loop as a mechanism to adjust their reactive state following injury .

Acute injuries of the central nervous system ( CNS ) trigger a robust reaction from glial cells—a non-neuronal population of cells that regulate and support neural development and physiology ., Although this process occurs after all types of CNS trauma in mammals , how it is activated and its precise role in recovery remain poorly understood ., Using the fruit fly Drosophila melanogaster as a model , we previously identified a cell surface receptor called Draper , which is required for the activation of glia after local axon injury ( “axotomy” ) and for the removal of degenerating axonal debris by phagocytosis ., Here , we show that regulation of Draper protein levels and glial activation through the Draper signaling pathway are mediated by the well-conserved PI3K and signal transducer and activator of transcription ( STAT ) signaling cascades ., We find that STAT transcriptional activity is activated in glia in response to axotomy , and identify an injury-responsive regulatory element within the draper gene that appears to be directly modulated by STAT ., Interestingly , the intensity of STAT activity in glial cells after axotomy correlates tightly with the number of local severed axons , indicating that Drosophila glia are able to fine-tune their response to neuronal injury according to its severity ., In summary , we propose that the initial phagocytic competence of glia is regulated by setting Draper baseline levels ( via PI3K ) , whereas injury-activated glial phagocytic activity is modulated through a positive feedback loop that requires STAT-dependent activation of draper ., We speculate that the level of activation of this cascade is determined by glial cell recognition of Draper ligands present on degenerating axon material , thereby matching the levels of glial reactivity to the amount of injured axonal material .

neuroglial development, molecular neuroscience, developmental neuroscience, cellular neuroscience, neural homeostasis, biology and life sciences, neuroscience

Activation of glial cells following axon injury is mediated by a positive feedback loop downstream of the glial phagocytic receptor Draper, allowing the strength of the response to match the severity of injury.

journal.pcbi.1005932

Bayesian inference of phylogenetic networks from bi-allelic genetic markers

The availability of genome-wide data from many species and , in some cases , many individuals per species , has transformed the study of evolutionary histories , and given rise to phylogenomics—the inference of gene and species evolutionary histories from genome-wide data ., Consider a data set S = {S1 , … , Sm} consisting of the molecular sequences of m loci under the assumptions of free recombination between loci and no recombination within a locus ., The likelihood of a species phylogeny Ψ ( topology and parameters ) is given by, L ( Ψ | S ) = ∏ i = 1 m L ( Ψ | S i ) = ∏ i = 1 m ∫ G p ( S i | g ) p ( g | Ψ ) d g ( 1 ), where the integration is taken over all possible gene trees ., The term p ( Si|g ) is the likelihood of gene tree g given the sequence data of locus i 1 ., The term p ( g|Ψ ) is the density function ( pdf ) of gene trees given the species phylogeny and its parameters ., For example , Rannala and Yang 2 derived this pdf under the multispecies coalescent ( MSC ) ., This formulation underlies the Bayesian inference methods of 2–4 ., Debate has recently ensued regarding the size of genomic regions that would be recombination-free ( or almost recombination-free ) and could truly have a single underlying evolutionary tree 5 , 6 ., One way to overcome this issue is to use unlinked single nucleotide polymorphisms ( SNPs ) or amplified fragment length polymorphisms ( AFLPs ) ., Such data provide a powerful signal for inferring species phylogenies and the issue of recombination within a locus becomes irrelevant ., Furthermore , as long as those markers are sampled far enough from each other the assumption of free recombination among loci holds ., Indeed , this is the basis of the SNAPP method that was recently introduced in 7 ., Since a bi-allelic SNP or AFLP marker has no signal by itself to resolve much of the branching patterns of a gene genealogy , a major contribution of Bryant et al . was an algorithm for analytically computing the integration in Eq ( 1 ) for bi-allelic markers ., While trees constitute an appropriate model of the evolutionary histories of many groups of species , it is well known that other groups of species have evolutionary histories that are reticulate 8 ., Horizontal gene transfer is ubiquitous in prokaryotes 9 , 10 , and several bodies of work are pointing to much larger extent and role of hybridization in eukaryotic evolution than once thought 8 , 11–15 ., Not only does hybridization play an important role in the genomic diversification of several eukaryotic groups , but increasing evidence is pointing to the adaptive role it has played , for example , in wild sunflowers 16 , humans 17 , macaques 18 , mice 19 , butterflies 20 , and mosquitoes 21 , 22 ., Reticulate evolutionary histories are best modeled by phylogenetic networks ., Two statistical methods were recently introduced for inference under the formulation given by Eq ( 1 ) , when Ψ is a phylogenetic network 23 , 24 , and other methods were also introduced for statistical inference of phylogenetic networks using gene tree estimates as the input data 25–29 ., The methods of 23 , 24 assume that the data for each locus consists of a sequence alignment that has no recombination ., In this paper , we devise an algorithm that builds on the algorithm of 7 for analytically computing the integral in Eq ( 1 ) when Ψ is a phylogenetic network ., In other words , our algorithm allows for computing the likelihood of a phylogenetic network from unlinked bi-allelic markers while analytically integrating out the gene trees for the individual markers ., We couple this likelihood function with priors on the phylogenetic network and its parameters to obtain a Bayesian formulation , and then employ the reversible-jump MCMC ( RJMCMC ) kernel from 23 to sample the posterior of the phylogenetic networks and their associated parameters given the bi-allelic data ., We implemented our algorithm and the RJMCMC sampler in PhyloNet 30 , which is a publicly available open-source software package for inferring and analyzing reticulate evolutionary histories ., We studied the performance of our method on simulated and biological data ., For simulations , we extended the framework of 7 so that the evolution of bi-allelic markers could be simulated within the branches of a phylogenetic network ., For the biological data , we analyzed two data sets of multiple New Zealand species of the plant genus Ourisia ( Plantaginaceae ) ., The results on the simulated data show very good accuracy and robustness as reflected by the method’s ability to recover the true phylogenetic networks and their associated parameters even when the underlying assumptions of the method are violated ., For the biological data , the method recovers two established hybrids and their putative parents correctly ., The proposed method and Bayesian sampler provide a new tool for biologists to infer reticulate evolutionary histories , while also account for the complexity arising from incomplete lineage sorting , from bi-allelic markers , thus complementing existing tools that use gene tree estimates or sequence alignments of the individual loci as the input data ., The use of such bi-allelic markers , particularly when they are sampled far enough across the genome , completely sidesteps potential problems that could arise due to the presence of recombination within loci ., A phylogenetic X -network , or X -network for short , Ψ is a rooted , directed , acyclic graph ( DAG ) whose leaves are bijectively labeled by set X of taxa ., We denote by V ( Ψ ) and E ( Ψ ) the sets of nodes and edges , respectively , of the phylogenetic network Ψ ., Every node of the network has in-degree 1 , which we call a tree node , or in-degree 2 , which we call a reticulation node ., The only exception is special node s whose in-degree is 0 and out-degree is 1; the edge ( s , r ) defines the branch above the root ., The edges whose head is a reticulation node are the reticulation edges of the network; all other edges constitute the tree edges of the network ., Every edge is directed forward in time ., We assume all phylogenies considered here ( trees and networks ) are binary—no node has out-degree higher than 2 ., Here , we use the bottom of a branch to refer to the end of the branch that is farther from the root of the network , and use the top of a branch to refer to the end of the branch that is closer to the root ., Given that the coalescent views the evolution of alleles backward in time , we say that a lineage enters a branch to mean a lineage that exists at the bottom of that branch ., Similarly , we say a lineage exits a branch to mean a lineage that exists at the top of that branch ., Each node in the network has a species divergence time and each edge b has an associated population mutation rate θb = 4Nbμ ., This parameter is typically referred to in the literature as the ( rescaled ) population size ., Given the length τ of a branch in units of expected number of mutations per site , the length of that branch in coalescent units is 2τ/θ , assuming diploid individuals ., The branch above the root , ( s , r ) , is infinite in length so that all lineages that enter it coalesce on it eventually ., For every pair of reticulation edges e1 and e2 that share the same reticulation node , we associate an inheritance probability , γ , such that γ e 1 , γ e 2 ∈ 0 , 1 with γ e 1 + γ e 2 = 1 ., We denote by Γ the vector of inheritance probabilities corresponding to all the reticulation nodes in the phylogenetic network ., We use Ψ to refer to the topology , species divergence times , population mutation rates , and inheritance probabilities of the phylogenetic network ., That is , here we include Γ as part of Ψ ., An X -phylogenetic tree , or X -tree , is an X -network with no reticulation nodes ., A gene tree is an X -tree ., Each node in the gene tree has an associated coalescence time ., In the algorithm below , we make use of a coloring function c: ( E ( g ) , t ) → {0 , 1} , similar to that used in 7 , where c ( e , t ) indicates the color , or allele , at time t along the branch e of gene tree g ., We will follow 7 in calling the two colors red and green ., Looking forward in time ( from the root toward the leaves ) , let u and v be the mutation rate from red allele to green allele and the mutation rate from green allele to red allele , respectively ., The stationary distribution of the red and green alleles at the root is given by v/ ( u + v ) and u/ ( u + v ) , respectively ., Observed alleles are indicated by values of the coloring function c at gene tree leaves ., Given a gene history embedded within the branches of the network , the numbers and types of lineages at both ends of each branch of the network are needed to compute the likelihood ., Let x be a branch in the phylogenetic network ., We denote by n x T and n x B the total numbers of lineages at the top and bottom of x , respectively , and by r x T and r x B the numbers of red lineages at the top and bottom of x , respectively ., See Fig 1 for an illustration ., Let x be an arbitrary branch in the phylogenetic network and let R x be the event that for every external branch z that is a descendant of x , the actual number of red alleles in z equals to r z B . We define two partial likelihoods: F x B is the product of the likelihood of a subtree rooted at the bottom of x and the probability P r n x B = n , and F x T is the product of the likelihood at the top of branch x and the probability P r n x T = n ., In the case of a species tree ( i . e . , no reticulation nodes in the species phylogeny ) , the partial likelihood vectors F x B and F x T are given by 7, F x B ( n , r ) = P r R x | n x B = n , r x B = r P r n x B = n ( 2 ), and, F x T ( n , r ) = P r R x | n x T = n , r x T = r P r n x T = n ., ( 3 ) Here F x B and F x T are indexed by nonnegative integers n and r , where r ≤ n ., Let M be the maximum possible value of n x B and n x T over all branches ., Then , each of F x B and F x T has at most l = ( 1 + ( M + 1 ) ) ( M + 1 ) /2 entries ., In the case of a species tree , the path from a leaf to the root is unique ., However , this might not be the case for phylogenetic networks: If there is a reticulation node on a path from a leaf to the root , then multiple paths exist between that leaf and the root ., This is the issue that necessitates modifying the algorithm of 7 significantly , and that leads to much larger computational requirements in the case of phylogenetic networks ., The key idea behind the modification is as follows ., As the algorithm proceeds to compute the likelihood in a bottom-up fashion from the leaves to the root , whenever a reticulation node is encountered , the current set of lineages is bipartitioned in every possible way so that one side of the bipartition tracks one parent of the reticulation node and the other side tracks the other parent ., As the network has a unique root , the two sides of each bipartition eventually come back together at an ancestral node ., At that point , these two sides are merged properly ., To achieve this proper merger , we introduce “labeled partial likelihoods , ” or LPL ., Like the case of 7 , LPLs are not “real” partial likelihoods ., The reason for this is that when partial likelihood vectors are split ( described below ) , those become symbolic terms that do not evaluate to partial likelihoods until they are merged later ., This is analogous to the difference between ancestral configurations on species trees 31 and their labeled counterparts on phylogenetic networks 32 , where the latter are in many cases just symbolic terms that do not evaluate to true ( partial ) likelihood values ., Given a phylogenetic network Ψ with k reticulation nodes numbered 0 , 1 , ⋯ , k − 1 , an LPL P is an element of 0 , 1 l × Z k , where the first element of the pair is a partial likelihood as in 7 ., The second element is the label to keep track of partial likelihoods that originated from a split of the same partial likelihood at a reticulation node so that these two could be merged ., More formally , we say two LPLs P1 = ( F1 , s1 ) and P2 = ( F2 , s2 ) , where |s1| = |s2| , are compatible if and only if for every 0 ≤ i < |s1| , either s1 ( i ) = s2 ( i ) or s1 ( i ) ⋅ s2 ( i ) = 0 ., We denote by P x T and P x B the sets of LPLs that are associated with the top and bottom of branch x , respectively ., These two quantities are computed in a bottom-up fashion , proceeding from the leaves of the network towards its root ., Once the LPLs at the root are computed , the overall likelihood of a given site is computed ., As the algorithm proceeds from the leaves towards the root , it needs to compute LPLs at the leaves , the top of a branch , the bottom of reticulation edges , and the bottom of tree edges ., We now describe each of those computations; the overall algorithm is simply a bottom-up traversal of the network while applying the appropriate computation as a node is encountered ., Our algorithm computes the likelihood of a phylogenetic network given a set of biallelic markers ., This algorithm computes matrix exponential along every branch , and processes the network’s nodes in a post-order traversal ., Computation at a leaf takes O ( 1 ) time ., At a reticulation node , the time consumption increases after each reticulation node is processed , due to the accumulation of ( split ) LPLs ., In the last processed reticulation node , the number of LPLs in its descendant is at most O ( n4 ( k−1 ) ) ., There are at most O ( n4 ) new LPLs generated due to decompose-and-split operation for each original LPL ., Therefore the time complexity of processing a reticulation node is at most O ( n4k ) ., We adopted the same approximation of matrix exponential as in 7 , so the time complexity of computing matrix exponentiation is O ( n2 ) , and computation along every branch is at most O ( n4k+2 ) ., At a tree node , computation is mostly spent on evaluating Eq ( 13 ) ., Let n be the number of individuals present under an internal tree node ., Then , this evaluation takes O ( n4 ) time for a pair of compatible LPLs ., The total time consumption of processing tree nodes also depends on the number of LPLs ., Assuming k reticulation nodes in the phylogenetic network , there are at most O ( n4k ) pairs of compatible LPLs ., Therefore the time complexity of processing a tree node is O ( n4k+4 ) ., In total , the time complexity of the algorithm is O ( mn4k+4 ) , where m is the number of species , n is the total number of lineages sampled from the species , and k is the number of reticulation nodes ., Notice that when k = 0 , which means the species phylogeny is a tree , the time complexity is O ( mn4 ) , which is the running time of the SNAPP algorithm without fast Fourier transforms ., To speed up computation , and since markers are independent , computations for the individual markers are parallelized by multi-threading ., Furthermore , the data is preprocessed so that the unique marker patterns are identified and their probabilities are computed only once and reused for for all markers with the same patterns ( states for the taxa ) ., The prior on the phylogenetic network is the same as that employed in 23 , which we review briefly here ., The prior is given by, p ( Ψ | ν , δ , η , ζ , α , β ) = p n u m r e t ( Ψ | ν ) × p d i a m ( Ψ | η ) × p d i v ( Ψ | δ ) × p p o p ( Ψ | ζ ) × p i n h ( Ψ | α , β ) ., ( 15 ), Here , p ( Ψ|ν ) is a Poisson prior on the number of reticulation nodes , normalized by the number of networks with the same number of reticulation nodes as Ψ ., pdiam ( Ψ|η ) is an exponential prior on the diameters of reticulation nodes ., The diameter of a reticulation node is the sum of the branch lengths on the cycle that contains the reticulation node in the underlying undirected graph of the network ., pdiv ( Ψ|δ ) is an exponential prior on the divergence times ., Rannala and Yang used independent Gamma distributions for time intervals ( branch lengths ) instead of divergence times ., However , in the absence of any information on the number of edges of the species network as well as the time intervals , it is computationally intensive to infer the hyperparameters of independent Gamma distributions ., Currently , we use a uniform distribution ( as in BEST 33 ) ., ppop ( Ψ|ζ ) is a Gamma prior on the population mutation rate ., For ppop , we use the Gamma distribution Γ ( 2 , ζ ) with mean value 2ζ and shape parameter 2 ., pinh ( Ψ|α , β ) is a Beta prior , with parameters α and β , on the inheritance probabilities ., Unless there is some specific knowledge on the inheritance probabilities , a uniform prior on 0 , 1 is adopted by setting α = β = 1 ., It is important to note here that if the topology of Ψ does not follow the phylogenetic network definition ( e . g . , has a cycle ) , then p ( Ψ|ν , δ , η , ψ ) = 0 ., This is crucial since , in the MCMC kernels we employ for sampling the posterior distribution , we allow the moves to produce directed graphs that slightly deviate from the definition; in this case , having the prior be 0 guarantees that the proposal is rejected ., Using the strategy , rather than defining only “legal” moves simplifies the calculation of the Hastings ratios ., However , the sampler always guarantees that the divergence times are consistent; that is , no node has a divergence time smaller than or equal to the divergence time of any of its descendants ., We employed the reversible-jump MCMC , or RJMCMC 34 algorithm implemented in PhyloNet 30 to sample from the posterior distribution given by, p ( Ψ | S ) ∝ L ( Ψ | S ) p ( Ψ ) , ( 16 ), where Ψ here denotes the topology of the network and all its parameters , and p ( Ψ ) is the prior on the network and its parameters as described above ., We make use of only the 12 proposals designed for sampling phylogenetic networks and their parameters described in 23 , but not the proposals aimed at sampling gene trees , as gene trees are integrated out ., We implemented in PhyloNet 30 a program to simulate bi-allelic markers on a given phylogenetic network ., Bryant et al . 7 simulated bi-allelic markers by first generating gene trees inside a species tree ( under the multispecies coalescent model ) , and then simulating the markers down the gene trees ., In our case , we replaced the first step by generating gene trees inside a phylogenetic network under the multispecies network coalescent 26; the second step of simulating bi-allelic markers down gene trees remains the same as that employed in 7 ., When requiring the data set to contain only polymorphic sites , if the generated site is not polymorphic , we discard both gene tree and markers , and repeat until a polymorphic site is generated ., Two small subsets of a larger AFLP data set of multiple New Zealand species of the plant genus Ourisia ( Plantaginaceae ) 39 were analyzed , including previously unpublished AFLP profiles from two different hybrid individuals O . × cockayneana and O . × prorepens ( herbarium codes follow 40 continuously updated ) ., There are both morphological 41 and molecular ( Meudt unpubl . ) data supporting the hybrid nature of these two individuals ., Although other Ourisia hybrid combinations have been reported in New Zealand 41 , O . × cockayneana and O . × prorepens are perhaps the most common , both involve O . caespitosa as a putative parent , and both have been formally named ., Each data subset comprised five diploid individuals in total , which means ten haploid individuals were effectively analyzed due to the correction for dominant markers ., A Poisson distribution with λ = 1 . 5 as the prior on the number of reticulations , an exponential prior with λ = 2 . 0 as the prior on the species divergence times , and a Gamma distribution with α = 2 . 0 and β = 0 . 05 as the prior on the population mutation rates were adopted ., An MCMC chain was run on each data subset for 1 . 5 × 106 iterations with 2 × 105 burn-in iterations , and a sample was collected every 500 iterations ., We used following commands:, Phylogenetic networks allow for representing evolutionary relationships that involve both vertical and horizontal transmission of genetic material ., Extensions of the multispecies coalescent process to include hybridization events have facilitated the development of statistical methods for inferring and analyzing phylogenetic networks from gene tree estimates and sequence data ., A major challenge with using gene tree estimates as the input to species phylogeny inference methods is the error in these estimates ., While using the sequence data directly overcomes this issue , the problem of recombinations within loci can confound inferences ., Using bi-allelic markers from individual , independent loci could provide a way to avoid both the gene tree uncertainty and recombination problems ( the two are not necessarily independent ) ., Furthermore , it is important to note that many biological studies use data sets that consists of bi-allelic markers and no available sequence alignment data for individual loci ., Bryant et al . recently devised an algorithm for inferring species trees from bi-allelic genetic markers while analytically integrating out the gene trees for the individual loci 7 ., In this paper , we extended their algorithm significantly so as the likelihood of a phylogenetic network given bi-allelic markers could be computed while integrating out the gene trees ., This method complements existing ones that use gene tree estimates or sequence alignments as input for statistical inference of phylogenetic networks ., We implemented a Bayesian method for sampling the posterior of phylogenetic networks and their associated parameters from bi-allelic data , and studied its performance on both simulated and empirical data ., The results indicate a very good performance of the method ., This work adds a powerful method to the biologist’s toolbox that allows for estimating reticulate evolutionary histories ., A major bottleneck of the method is its computational requirements ., While the SNAPP method is very time consuming on species trees , our method is much more time consuming given that reticulations in the phylogenetic network give rise to an explosion of the number of partial likelihoods that need to be computed and stored ., More generally , the number of taxa in a data set has more of an effect on the running time of the method than the number of loci does ., In particular , two aspects of the phylogenetic network under consideration affect the computational requirements of the method: The number of leaves under the reticulation nodes and the diameter of each of the reticulation nodes ., As discussed above , the set of lineages entering a reticulation node must be bipartitioned in every possible way ., This number of lineages is dependent on the number of leaves under that reticulation node ., For example , if a single individual is sampled from a single species that exist under the reticulation node , then the number of bipartitions is very small ( only two bipartitions exist ) ., However , if n individuals are sampled from a single species that exist under the reticulation node or one individual is sampled per n species that exist under the reticulation node , then a number of bipartitions on the order of 2n arises ., This computation becomes much more demanding if there are more reticulation nodes on the path to a lowest articulation node ., As for the diameter—which is the number of branches on the paths between the two parents of the reticulation node and a lowest articulation node above them , the larger its value , the more demanding the computation becomes ., An important direction for future research is improving the computational requirements of the method to scale up to data sets with many taxa .

Introduction, Materials and methods, Results, Discussion

Phylogenetic networks are rooted , directed , acyclic graphs that model reticulate evolutionary histories ., Recently , statistical methods were devised for inferring such networks from either gene tree estimates or the sequence alignments of multiple unlinked loci ., Bi-allelic markers , most notably single nucleotide polymorphisms ( SNPs ) and amplified fragment length polymorphisms ( AFLPs ) , provide a powerful source of genome-wide data ., In a recent paper , a method called SNAPP was introduced for statistical inference of species trees from unlinked bi-allelic markers ., The generative process assumed by the method combined both a model of evolution for the bi-allelic markers , as well as the multispecies coalescent ., A novel component of the method was a polynomial-time algorithm for exact computation of the likelihood of a fixed species tree via integration over all possible gene trees for a given marker ., Here we report on a method for Bayesian inference of phylogenetic networks from bi-allelic markers ., Our method significantly extends the algorithm for exact computation of phylogenetic network likelihood via integration over all possible gene trees ., Unlike the case of species trees , the algorithm is no longer polynomial-time on all instances of phylogenetic networks ., Furthermore , the method utilizes a reversible-jump MCMC technique to sample the posterior of phylogenetic networks given bi-allelic marker data ., Our method has a very good performance in terms of accuracy and robustness as we demonstrate on simulated data , as well as a data set of multiple New Zealand species of the plant genus Ourisia ( Plantaginaceae ) ., We implemented the method in the publicly available , open-source PhyloNet software package .

The availability of genomic data has revolutionized the study of evolutionary histories and phylogeny inference ., Inferring evolutionary histories from genomic data requires , in most cases , accounting for the fact that different genomic regions could have evolutionary histories that differ from each other as well as from that of the species from which the genomes were sampled ., In this paper , we introduce a method for inferring evolutionary histories while accounting for two processes that could give rise to such differences across the genomes , namely incomplete lineage sorting and hybridization ., We introduce a novel algorithm for computing the likelihood of phylogenetic networks from bi-allelic genetic markers and use it in a Bayesian inference method ., Analyses of synthetic and empirical data sets show a very good performance of the method in terms of the estimates it obtains .

taxonomy, genetic networks, genome evolution, applied mathematics, simulation and modeling, algorithms, phylogenetics, data management, mathematics, phylogenetic analysis, network analysis, research and analysis methods, computer and information sciences, genomics, molecular evolution, evolutionary systematics, genetic loci, gene identification and analysis, genetics, biology and life sciences, physical sciences, computational biology, evolutionary biology

journal.pntd.0002131

Analysis of Schistosomiasis haematobium Infection Prevalence and Intensity in Chikhwawa, Malawi: An Application of a Two Part Model

According to 1 , Schistosomiasis infections affect an estimated 779 million people , with consequences in health nutritional and educational development of infected individuals 2 ., The disease causes an annual loss of 4 . 5 million disability-adjusted-lifeyears ( DALYs ) 3 ., In SSA alone , 207 million individuals are estimated to be infected with Schistosomiasis: S . haematobium and S . mansoni 1 ., S . haematobium is reported to be endemic in 53 countries in the Middle east and most of the African continent including islands of Madagascar and Mauritius 4 , whereas S . mansoni is mostly endemic in sub-Saharan Africa 4 ., Schistosomiasis can be effectively treated with single dose oral therapies of praziquantel that are safe , inexpensive and required at periodic intervals 5 ., Treatment is typically implemented through mass chemotherapy whereby the entire at-risk population is treated , as part of either school or community- based campaigns , referred to as mass drug administration ( MDA ) ., The transmission intensity of Schistosomiasis is a function of parasitic worm load within a group of individuals , which can indirectly be quantified by the number of eggs that are excreted ., Host heterogeneities in exposure and susceptibility to infection may lead to an aggregated distribution of worm burden across individuals 6 ., For this reason , a few individuals would harbour large numbers of worms , whilst the majority of individuals are uninfected or only carry a low worm burden 6 ., In addition , widely used diagnostic approaches for Schistosomiasis like the Kato-Katz technique for S . mansoni diagnosis fail to detect some infected individuals , particularly when only a single stool sample is examined and infection intensities are light 7 ., Due to these two issues , often a large proportion of individuals are considered as “zero egg excretor” 6 ., The standard Poisson distribution , which assumes equal mean and variance , commonly employed to model such count data , is inappropriate to fit observed egg counts since the variance of the counts is much larger than their mean , a case known as over-dispersion 8 ., The use of negative binomial ( NB ) distribution has been proposed to model the extra-Poisson variation 9 , and applications of NB in analysing helminth egg counts are many 8 , 10 ., Although NB models may be ideal for over-dispersion , they may not be suitable when data is zero-inflated ., Other distributions like the hurdle models or zero inflated ( ZI ) or zero augmented models that may be more appropriate for modeling data with such excess zeros are reported 8 ., These models can have more than one mode , including a mode at zero ., ZI models attempt to account for excess zeros , i . e . , zero inflation arises when one mechanism generates only zeros and the other process generates both zero and nonzero counts hence they can be expressed as a two-component mixture model where one component has a degenerate distribution at zero and the other is a count model 11 ., ZI models estimate two equations , one for the count model and one for the excess zeros ., ZI models assume that a proportion of individuals have no chance to be infected , as they are not exposed ., In other words , there is a process which determines whether an individual is likely to be infected at all and a second process determining the number of excreted eggs among those who are at risk of infection ., Zero inflated Poisson ( ZIP ) models assume that the number of excreted eggs follows a Poisson distribution ., Zero-inflated negative binomial ( ZINB ) models assume that the number of worms among those who are at risk of infection has a negative binomial distribution 6 ., ZI count data are common in a number of applications ., Examples of data with too many zeros from various disciplines include agriculture , econometrics , patent applications , species abundance , medicine , and use of recreational facilities 8 ., The zero-inflated Poisson ( ZIP ) regression models with an application to defects in manufacturing is described in 12 , while zero-inflated binomial ( ZIB ) regression model with random effects into ZIP and ZIB models are defined in 13 ., The idea for a hurdle model , a modified count model in which the two processes generating the zeros and the positives are not constrained to be the same , was developed in 14 ., The two processes are modeled using a mixture of two models ( i . e , two part or a hurdle model ) ., The first part is a binary outcome model , and the second part is a truncated count model ., Such a partition permits the interpretation that positive observations arise from crossing the zero hurdle or the zero threshold ., The first part models the probability that the threshold is crossed , in our case thatan infection occurred ., In principle , the threshold need not be at zero; it could be any value , and it need not be treated as known ., The zero value has special appeal because in many situations it partitions the population into subpopulations in a meaningful way , one on infection status and the other for those infected it captures intensity ., In contrast to the zero-inflated model , the zero and non-zero counts are separated in the hurdle model 15 which makes them very useful in inferential studies ., Hurdle models are sometimes referred to as zero-altered models 16 ., Zero-altered Poisson and negative binomial models are thus referred to , respectively , as ZAP and ZANB ., They have also been termed overlapping models 17 ., The application of hurdle or two part models in epidemiology has not been common so far ., Use of ZI models have been reported ., One such an application was in Cote dIvore , in which a ZINB model within a model-based geostatistics ( MBG ) framework for S . mansoni infection was applied 6 ., This study showed that geostatistical ZI models produce more accurate maps of helminth infection intensity than the spatial negative binomial counterparts ., However , to our knowledge , no hurdle or two part model has been applied in Schistosomiasis or geohelminth epidemiology ., This paper demonstrates the applications of hurdle models to helminth epidemiology ( S . haematobium ) and encourage its wider application in helminth disease control programmes ., Its advantage is that it allows joint modeling of infection status and intensity ., Although , a multinomial model maybe used 18–20 , its limitation is that it involves stratifying egg counts , leading to a loss of information , whereas the negative binomial hurdle model approach makes full use of intensity data on a continuous scale , therefore , ideal to model latent infection intensity ., In addition , hurdle models are robust when over-dispersion is present ., In 8 , it was concluded that the ZIP models were inadequate for the data as there was still evidence of over-dispersion ., Moreover , the negative binomial hurdle model , which allows for over-dispersion and accommodates the presence of excess zeros through a two-part model has a natural epidemiological interpretation within the case study considered here ., The data which motivated this work were collected in 2004 in Chikhwawa district , in the Lower Shire Valley-southern Malawi ., This is a rural area whose population is mainly engaged in subsistence farming ., This area lies between 100 and 300 m above sea level ., The rainy season extends from December to March ., Temperatures can rise up to in months preceding rainy season ., Malaria is known to be holoendemic 21 ., Data were collected in eighteen villages , purposively selected from the control and intervention arms of a cluster randomized study design ., There was only one round of treatment following community based and house to house approaches for mass drug administration ( MDA ) ., Over 90 percent of the eligible population were treated ., All infected participants in non-intervention arm received appropriate treatment ., After the follow-up assessment , both arms had mass treatment ., In the study , polyparasitism was considered basing on the number of species an individual was hosting ., The focus was on Hookworm , S . mansoni , S . haematobium and Ascaris ., Polyparasitism is the epidemiology of multiple species parasite infections ., Ten percent of the households were randomly selected from the villages for baseline survey using random number tables 22 ., Subjects for geo-helminth survey were selected using a two stage-design ., Briefly , at first stage , villages were selected , then at second stage , sample of households was listed and chosen ., In the selected households all members aged one year and above were invited to participate ., Consenting individuals had their demographic details completed and were given full body clinical examinations ( except genitals for females ) for chronic manifestations of human helminths ., In addition they had anthropometric measurements taken and were asked to provide a single fresh stool and urine sample ., All individuals ( aged>1 year ) were requested to provide a finger prick blood sample 22 ., Further details are provided in 22 ., Fresh stool samples were transported in a cooler box to the laboratory and processed within four hours of collection ., A single Kato-Katz thick smear was prepared from each sample and immediately examined under a light microscope for parasite eggs ( within 15–20 minutes ) ., Standardized and quality controlled procedures were followed ., Briefly 41 . 7 mg of sieved stool was placed on a microscope slide through a punched plastic template ., Ova for each parasite observed were counted and expressed as eggs per gram ( epg ) of stool ., Five percent of the slides were randomly selected for re-examination for quality control purposes 22 ., Urine samples were processed on the day of collection ., A measured volume ( maximum 10 ml ) was centrifuged at 300 rpm for five minutes ., The sediment was then examined under a light microscope ., The eggs seen were counted and the intensity of infection per 10 ml of urine accordingly determined ., All those infected were treated with praziquantel at 40 mg/kg 22 ., The study that collected data from Chikhwawa , Malawi received ethical clearance from Malawis College of Medicine Research Ethics Committee ( COMREC ) 22 ., Individual informed consent was orally obtained from each participant or ( if they were aged<16 ) from one of their parents or a legal guardian ., COMREC approved oral informed consent because the study was determined to be of minimal risk ., The consent process was a four stage process ., First stage , oral informed consent was obtained at the traditional authority ( TA ) level ., Second stage , at village head level and third stage at the household level from the head of the household and fourth at individual level from each individual in the household ( if applicable ) else from parent/guardian if an individual was aged<16 ., Registers were kept for documentation whereby , for each individual in the selected household , a column was kept to indicate whether an individual had orally consented to participate in the study or not ., Various statistical models have been developed to model helminths disease burden as reviewed in the introduction ., For purposes of this paper , we assumed a negative binomial logit hurdle ( NBLH ) model for joint analysis of infection prevalence and intensity of Schistosomiasis hematobium in Malawi ., Following on 11 , a NBLH model can be written as: ( 1 ) ( 2 ) where are observed counts taking values for each individual ., The probability of infection is , such that indicates there are no zero counts and the model reduces to a truncated Negative Binomial distribution ( TNegBinom ) ; while means there are no infections ., The observed counts are modelled by assuming two processes: ( 3 ) The first is assumed to model the infection prevalence ( first hurdle ) and the other the intensity of infection ( second hurdle ) ., The first hurdle assumes a binary outcome defining whether an individual is infected or not ., This is modeled as a logit regression for a given set of risk factors ., After determining infection status we are interested in analyzing the number of eggs - as a measure of intensity of infection , which is defined by the second hurdle ., We model the second hurdle as a negative binomial regression model for a given set of risk factors ., The NB model is suited for count data with over-dispersion ., In many cases , the same risk factors are used in the logit and count regression models , i . e . ., The two regression models , incorporating the risk factors , are given by: ( 4 ) ( 5 ) The model parameters and are estimated using maximum likelihood estimation in which the likelihoods ( or log-likelihoods ) are maximized separately ., The covariates included in the model are given in Table 1 ., Age and polyparasitsm were fitted as continuous variables , while sex , education levels , village type , fishing , gardening and occpuation were entered in the model as categorical variables , with the first category of each variable selected as the reference group ., For both parts of the model we used the same set of covariates ., We also fitted a number of count models , with the Poisson as the null model , for comparison and evaluated the number of zeros each model correctly predicts ., We also compared model fit using AIC and zero capturing ., A difference of 10 indicates the model with the smallest AIC is superior to others ., Furthermore , deviance residuals were assessed for spatial correlation using variogram and were subsequently mapped using kriging to depict spatial variation in risk ., Statistical model fitting was carried out using Political Science Computational Laboratory ( PSCL ) package 23 in R statistical software ( The R Foundation for Statistical Computing , Version 2 . 14 . 0 ) ., Variogram analysis and kriging were implemented in geoR 24 ., Table 1 gives summary statistics for study participants ., The study had 1642 participants of which 55 . 4 % were female ., The mean age ( years ) of 32 . 4 ( standard deviation: 22 . 8 ) ., Of these , 324 had hookworm representing 19 . 7 % of sample population , 71 of these had S . mansoni representing 4 . 3 % and 233 had S . haematobium representing a prevalence of 14 . 2 % ., Figure 1 shows that a large proportion of individuals i . e . 85 . 8 % were “zero egg excretors” hence the data were inflated with zeros ., The likelihood ratio test for overdispersion between Poisson and negative binomial at =\u200a0 . 05 showed a critical value test statistic\u200a=\u200a2 . 7 with a test statistic\u200a=\u200a10606 . 5 , p-value<0 . 001 ., Indeed , there was overwhelming evidence of overdispersion ., This was confirmed by the presence of excess zeros ( Figure 1 ) ., Using the AIC and zero capturing , the predicted counts using the NBLH indicate a closer fit with the observed values ., In Table 2 , AIC results show that the NBLH offers a better fit compared to using Poisson Logit Hurdle ( PLH ) or a negative binomial ( AIC\u200a=\u200a3 , 482 for NBLH; AIC\u200a=\u200a6 , 854 for PLH and AIC\u200a=\u200a3 , 576 for NB respectively ) ., The AIC further showed a difference of 10 , 700 for the NBLH compared to the Poisson and a difference of 19 comparing NBLH with ZINB , thus NBLH is superior among all competing models ., With regards to zero capturing , the Poisson model was again not appropriate as it could only capture 515 of the zeros whereas the NB-Zero adjusted based models were much better in capturing the zero counts ., The NBLH model captured 971 zeros which were equal to the observed ( Table 3 ) ., Since NB logit hurdle model offered the best fit to zero inflated helminth data in terms of the AIC ( minimum value for all the models fitted ) as well as true zero count capturing , it therefore became a natural choice for fitting a final model to model helminth infection intensity and determination of factors that foster infections ., Table 4 provides estimates for the fixed effects ., The probability of infection was found to be associated with age ( Odds Ratio OR\u200a=\u200a0 . 97 , 95 % Confidence Interval CI: 0 . 96–0 . 99 ) , the risk of infection was decreasing with age ., This assumed a linear relationship with age; 6 years being the baseline age ., The risk of infection was low in males than in females ( OR\u200a=\u200a0 . 61 , 95 % CI: 0 . 41–0 . 89 ) ., The association between risk of infection with education at both primary level ( OR\u200a=\u200a1 . 18 , 95 % CI: 0 . 81–1 . 71 ) and secondary level ( OR\u200a=\u200a1 . 37 , 95 % CI: 0 . 41–4 . 60 ) relative to those with no education was not significant ( p-value\u200a=\u200a0 . 62 ) ., Infection probability was found to be associated with village type; whether one was in the intervention area or control area ( OR\u200a=\u200a0 . 38 , 95 % CI: 0 . 26–0 . 54 , p-value<0 . 001 ) ., Those in the intervention area were at a reduced chance of infection relative to those in control area ., We observed a negative association between infection probability and fishing ( OR\u200a=\u200a0 . 73 , 95 % CI: 0 . 44–1 . 20 ) though not significant; contrary to the expectation ., Working in the garden was observed not to be significant albeit it was positive ( OR\u200a=\u200a1 . 34 , 95 % CI: 0 . 90–1 . 99 ) ., Again , occupation ( farmer/other ) showed a negative association with infection probability though with marginal significance ( OR\u200a=\u200a0 . 61 , 95 % CI: 0 . 35–1 . 06 ) with a p-value\u200a=\u200a0 . 17 ., We also noted that chances of infection were increasing with number of parasite species an individual was hosting ( Table 4 ) ( OR\u200a=\u200a7 . 30 , 95 % CI:5 . 56–9 . 59 ) ., From Table 4 , it was observed that infection intensity reduced with an increase in age ( Relative Risk RR\u200a=\u200a0 . 96 , 95 % CI: 0 . 95–0 . 98 ) ., Similar to infection prevalence , a linear relationship was assumed between infection intensity and age ., There was no difference of infection intensity between males and females ( RR\u200a=\u200a1 . 03 , 95 % CI: 0 . 72–1 . 47 ) ., Primary school children showed a high infection intensity relative to those that are in pre-school level ( RR\u200a=\u200a1 . 54 , 95 % CI: 1 . 08–2 . 19 ) whereas those in secondary level showed a reduced infection intensity ( RR\u200a=\u200a0 . 34 , 95 % CI: 0 . 11–1 . 06 ) though not significant ., There was a reduced risk for those in intervention area relative to those in the control area , though , not significant ( RR\u200a=\u200a0 . 81 , 95 % CI: 0 . 58–1 . 13 ) ., A positive association was also observed between those who did fishing in Shire river relative to those who did not fish ( Table 4 ) ., We observed an increased infection intensity in those working in the gardens relative to those who did not ( RR\u200a=\u200a1 . 21 , 95 % CI: 0 . 82–1 . 81 ) , albeit not significant and also increased infection intensity for farmers compared to non-farmers ( RR\u200a=\u200a1 . 83 , 95 % CI: 1 . 16–2 . 91 ) ., Estimating the continuous surface using variogram analysis and kriging , spatial patterns in the residuals were observed and subsequently mapped ., There was some degree of spatial dependence in residuals distribution across the study area , as evidenced by the spherical model ( Figure 2 ) ., The magnitude of spatial correlation decreased with separation distance until at distance of 10 km ., The predicted spatial surface , in Figure 3 , showed a relatively increased risk of infection in the northern part of the study area compared to other areas ., Low risk areas were in the southern parts , more especially in the south-eastern part of the study region ( Figure 3 ) ., The current study found a prevalence of 14 . 2 % for S . haematobium in Chikhwawa district ., This prevalence was well below national estimates , which a previous study in Malawi indicated to be between 40 and 50 % 25 ., The finding serves to highlight the fact that Schistosomiasis infections are highly localised and that nationwide surveys tend to overlook the focus of heterogeneity of infection ., Indeed , in a study conducted in the northern lakeshore area 26 , school children from four schools screened for Schistosomiasis reported a wide range of prevalence: 5 %–57 % of S . haematobium infection ., A national survey , representative of all school children in the country , and undertaken just before the rainy season , showed far lower levels of 7 % for S . haematobium 25 ., We used robust , contemporary statistical methods in a two part application to analyse risk factors for S . haematobium infection intensity and prevalence ., This resulted in estimates of parasitic infection prevalence and intensity that could be used in control programme planning by channeling resources to areas with a known high disease burden ., In this study we have looked at the intensity and prevalence of S . haematobium in relation to factors such as age , sex , education level , village type , fishing in Shire river , working in gardens , occupation and polyparasitism ., Polyparasitism is the epidemiology of multiple species parasite infections ., In the study , polyparasitism was based on the number of species an individual was hosting ., The focus was on Hookworm , S . mansoni , S . haematobium and ascaris ., The study confirms the critical importance of ascertaining the infection intensity ., We found that S . haematobium infection intensity reduced with age , this confirms what previous studies found ., In common intestinal helminths such as Ascaris lumbricoides ( large roundworms ) and Trichuris trichiura ( whipworm ) and also Schistosomiasis , children are more heavily affected and infected than adults 27 ., Several other studies have reported that school-aged children show high infection intensity and prevalence 25 , 28 , 29 ., Fishing in Shire river and working in gardens along the river were potential risk factors for exposure to schistosomes and subsequent infection because transmission requires contact with the aquatic habitat of intermediate host snails 30 ., This is in line with results from a study that was conducted in western Africa 20 , that contact with water bodies that are a habitat for intermediate host snails is one of the main risk factors ., Results showed low probability of infection for males compared to females ., This could be explained by a number of factors including that Malawi being an agriculture based economy , and that mainly agricultural activities are carried out by females , hence they are more exposed to risk factors such as working in gardens and farming ., Schistosomiasis is water dependent disease and the incidence is usually more amongst people who constantly get into contact with the schistosome infected waters through activities such as farming , fishing , swimming and washing 30 ., Results from the study showed that individuals who had received chemotherapy cure for helminth showed reduced risk of infection as well as infection intensity as compared to those in the control area ., Studies have shown that MDA significantly reduces Schistosomiasis infection 31 , 32 ., Evidence has shown that , following chemotherapeutic cure of S . mansoni or S . haematobium infection , older individuals display a resistance to re-infection in comparison to younger children 33 ., Therefore there is need to channel integrated control and interventions for helminths to areas with diseases burden in order to reduce and/or eradicate the infections - more especially towards school age children ., Several studies have shown that having one infection , is a risk factor for having other infections 34 ., It is conceivable that the first parasite that establishes an infection may modulate the immune response in such a way that it makes it easier for the next 22 ., Worthy noting were differences that existed in associations between infection probability and infection intensity ., For gender , males had a reduced risk of infection as compared to females ( negative association ) but high infection intensity ( positive association ) ., This could possibly be explained by the fact that women were mostly involved in agricultural activities there by being more exposed ., Also for those infected , many studies find that men visit public health care facilities much less frequently than do women 35 hence the high intensity ., Poly-parasitism was positively associated with infection probability but had a negative association with infection intensity ., This could be explained by the fact that having other parasites increases the chance of the body being susceptible to new parasite infections 34 ., Again , secondary level of education had a positive association with infection probability but showed a negative association with infection intensity ., This finding could be explained by the fact that an increase in education level corresponds to increase in age which comes with increased risk-behaviour of older school children who frequently contact schistosome-infested water for both domestic and livestock purposes relative to younger children 36 hence increased infection prevalence ., At the same time , an increase in education may correspond to increased awareness and access to treatment 37 by this group hence reduced infection intensity ., Those with the highest level of education , through high school , have showed the lowest mean infection intensity 37 ., Being a farmer had a negative association with probability of infection and a positive association with infection intensity ., The finding was in line with what was reported in 37; farmers showed the highest levels of Schistosomiasis infection among occupational groups ., Both education and occupation are proxies for the nature and intensity of water contact 37 ., Individuals become infected by prolonged contact ( like irrigating farm , bathing , washing or swimming ) with fresh water containing free-swimming Cercariae 30 ., We believe that the apparent dominance of agricultural , socio-economic and demographic factors in determining S . haematobium infection risk in the villages carries important implications for disease surveillance and control strategies ., Prevalence of S . haematobium was highly associated with age of an individual as well as working in the garden and also number of parasites an individual hosted ., Furthermore , S . haematobium infection intensity was associated with gender , education level , garden , occupation and village type ( intervention ) ., Cercariae control control through environmental modifications and strategies involving socio-economic status improvement and MDA may be more promising approaches to disease control in this setting ., Finally , zero adjusted methods represents a key advance in the epidemiological analysis of helminth disease data inflated with zeros ., There are an increasing number of examples in the published literature where two part methods are being used for zero inflated data for helminths diseases control planning and implementation programmes 38 , 39 ., Ease of implementation and straightforward interpretation of the components and its direct link with the observed data , makes the negative binomial logit hurdle model definitely a valuable alternative for researchers analysing zero-inflated count data for helminths .

Introduction, Materials and Methods, Results, Discussion

Urinary Schistosomiasis infection , a common cause of morbidity especially among children in less developed countries , is measured by the number of eggs per urine ., Typically a large proportion of individuals are non-egg excretors , leading to a large number of zeros ., Control strategies require better understanding of its epidemiology , hence appropriate methods to model infection prevalence and intensity are crucial , particularly if such methods add value to targeted implementation of interventions ., We consider data that were collected in a cluster randomized study in 2004 in Chikhwawa district , Malawi , where eighteen ( 18 ) villages were selected and randomised to intervention and control arms ., We developed a two-part model , with one part for analysis of infection prevalence and the other to model infection intensity ., In both parts of the model we adjusted for age , sex , education level , treatment arm , occupation , and poly-parasitism ., We also assessed for spatial correlation in the model residual using variogram analysis and mapped the spatial variation in risk ., The model was fitted using maximum likelihood estimation ., The study had a total of 1642 participants with mean age of 32 . 4 ( Standard deviation: 22 . 8 ) , of which 55 . 4 % were female ., Schistosomiasis prevalence was 14 . 2 % , with a large proportion of individuals ( 85 . 8 % ) being non-egg excretors , hence zero-inflated data ., Our findings showed that S . haematobium was highly localized even after adjusting for risk factors ., Prevalence of infection was low in males as compared to females across all the age ranges ., S . haematobium infection increased with presence of co-infection with other parasite infection ., Infection intensity was highly associated with age; with highest intensity in school-aged children ( 6 to 15 years ) ., Fishing and working in gardens along the Shire River were potential risk factors for S . haematobium infection intensity ., Intervention reduced both infection intensity and prevalence in the intervention arm as compared to control arm ., Farmers had high infection intensity as compared to non farmers , despite the fact that being a farmer did not show any significant association with probability of infection ., These results evidently indicate that infection prevalence and intensity are associated with risk factors differently , suggesting a non-singular epidemiological setting ., The dominance of agricultural , socio-economic and demographic factors in determining S . haematobium infection and intensity suggest that disease transmission and control strategies should continue centring on improving socio-economic status , environmental modifications to control S . haematobium intermediate host snails and mass drug administration , which may be more promising approaches to disease control in high intensity and prevalence settings .

Schistosomiasis is one of the great causes of morbidity among school aged children in the tropical region and Sub Saharan Africa in particular ., Its mainly transmitted through contact with water infested with intermediate host snail Cercariae ., Currently , over 200 million people are estimated to be infected in SSA alone ., Here , we used robust and contemporary statistical methods in a two part application to analyse risk factors for S . haematobium infection intensity and prevalence ., We found that S . haematobium was more common in younger children as compared to older children , thus making the infection and prevalence age dependent ., We also found that mass chemotherapy reduced both infection prevalence and intensity ., We found that dominance of agricultural , socio-economic and demographic factors in determining S . haematobium infection risk in the villages carries important implications for disease surveillance and control strategies ., Therefore disease transmission and control strategies centered on improving strategies involving socio-economic status , environmental modifications to control S . haematobium intermediate host snails and mass drug administration may be more promising approaches to disease control in high intensity and prevalence settings .

medicine, public health and epidemiology, mathematics, epidemiology, statistics

journal.pgen.1006812

Ranking and characterization of established BMI and lipid associated loci as candidates for gene-environment interactions

Gene-environment ( G×E ) interactions may contribute to complex diseases , but their detection has proven challenging; hence , a variety of approaches have been developed to enhance power ., Most G×E analyses focus on loci that are strong biological candidates 1 or those with highly significant marginal effects 2 ., The latter approach is attractive because these loci are available in many large cohorts , and can be conveniently followed-up with interaction analyses if environmental data are accessible ., Moreover , selecting SNPs with strong and reproducible marginal effect signals is a pragmatic data-reduction step that may improve power 3 , although this approach risks omitting other promising candidates 4 ., In a linear regression setting , the presence of interaction effects drives phenotypic variance heterogeneity by genotype 3 , 5 ., Exploiting variance heterogeneity as a signature of interactions is appealing because , unlike standard approaches for assessing G×E interactions , no explicit information about environmental exposures is needed 6 and multiple exposures can be simultaneously considered ., Here we explored whether loci identified in large-scale genome-wide association studies ( GWAS ) of blood lipids and body mass index ( BMI ) are strong candidates for G×E interactions by comparing genome-wide variance heterogeneity P-value distributions generated using Levene’s test against P-value distributions for marginal effects and explicit G×E interaction effects ( for smoking and physical activity ) ., We assessed between-genotype variance heterogeneity for up to 1 , 927 , 671 directly genotyped or imputed SNPs ( HapMap II CEU reference panel 7 ) that passed quality control ( QC ) ., Meta-analyses of Levene’s test summary statistics 8 were performed for BMI ( n≤44 , 211 participants ) , and blood concentrations of high-density lipoprotein cholesterol ( HDL-C ) ( n≤34 , 315 ) , low-density lipoprotein cholesterol ( LDL-C ) ( n≤34 , 180 ) , total cholesterol ( TC ) ( n≤34 , 318 ) and triglycerides ( TG ) ( n≤34 , 110 ) ., We then obtained marginal effects results for the same index traits and SNPs from publicly available GWAS summary data from the GIANT ( Genetic Investigation of ANthropometric Traits ) Consortium 9 and GLGC ( Global Lipids Genetics Consortium ) 10 , 11 ., We compared the genome-wide marginal effects with between-genotype variance heterogeneity results for each of the five cardiometabolic traits by calculating the association between marginal effects ( Pm ) and variance heterogeneity ( Pv ) P-values using the rank-based Spearman correlation ( ρ ) ., This was done using a set of 42 , 710 pruned SNPs produced using the--indep-pairwise command in PLINK ( see Materials and Methods ) to account for linkage disequilibrium ( LD ) among variants ., As shown in Table 1 ( see also Fig 1A and S1 Table ) , the Spearman’s ρ for the association between Pm and Pv for all pruned SNPs was of very small magnitude and only statistically significant for BMI ., The exclusion of SNPs based on progressively more conservative Pm thresholds ( Pm<0 . 05; Pm<10−4; previously established loci with Pm<5×10−8 in external datasets ) , saw corresponding improvements in the magnitude of these correlations , which were statistically significant for all traits except TC when focusing on previously established loci ., The BMI correlation at the Pm<0 . 05 threshold , as well as the test of equality with ρ for all SNPs , was statistically significant , suggesting concordance between marginal and variance signals at a nominal level of significance ., The odds ratio ( OR ) for a SNP to have both Pm<0 . 05 and Pv<0 . 05 as compared to Pv≥0 . 05 was 1 . 33 ( 95% CI: 1 . 12 , 1 . 57 ) for BMI while the 95% CIs of ORs for other traits included 1 ., On the other hand , the P-value for a non-zero ρ for TG was statistically significant when focusing on the established loci and at Pm<10−4 , suggesting concordance between marginal and variance signals at more conservative Pm thresholds ., We further compared Pm with interaction P-values from exposure-specific ( smoking and physical activity ) genome-wide interaction tests for BMI ( Pint ) ; this was only done for BMI owing to the requirement for an adequately powered external dataset ( such a dataset was accessible through the GIANT consortium ) ( Table 2 ) ., Marginal effects GWAS were performed by strata of smokers vs . non-smokers and physically active vs . inactive participants ( n = 210 , 316 European-ancestry adults 12 ) respectively , and a heterogeneity test 12 was used to generate exposure specific Pint distributions ., Spearman ρ for the pruned set of SNPs in the SNP × physical activity and the SNP × smoking analyses were low and not statistically significant ( Table 2 ) ., We also compared Pint values and Pv values for BMI ., Spearman’s ρ for the pruned set of SNPs were low and not statistically significant ., We next tested if the number of previously established marginal effect SNPs ( Pm<5×10−8 ) that were also nominally significant ( Pv<0 . 05 ) for variance heterogeneity was greater than expected by chance ( Tables 3 and 4 , Fig 1 ) ., For 4 out of the 5 index traits , we observed enrichment at the lower end of the Pv distribution ( Pv<0 . 05 ) for the established GWAS-derived lead SNPs ., Thus , the nominally significant regions of the Pv distributions were generally enriched for GWAS-derived loci ., We also performed enrichment analyses to test if previously established marginal effects SNPs ( Pm<5×10−8 ) are enriched for nominally significant ( Pint<0 . 05 ) interactions in the SNP × physical activity or SNP × Smoking analyses , but no enrichment was observed ( Table 3; Fig 1B ) ., By contrast , for the physical activity and smoking interaction tests ( using all pruned SNPs ) , the lower end of the Pint distribution ( Pint<0 . 05 ) was enriched with SNPs that were nominally significant in the Levene’s test analysis ( Pv<0 . 05 ) ( Table 4 ) ., This enrichment translated into an OR of 1 . 08 ( 95% CI: 1 . 01 , 1 . 14 ) for a SNP to have Pint<0 . 05 given Pv<0 . 05 vs . Pv≥0 . 05 for SNP × physical activity interaction ., The corresponding OR for the SNP × smoking interaction test was not significant ( OR = 1 . 02; 95% CI: 0 . 96 , 1 . 08 ) ., Finally , in the pruned SNP-set we used the Mann–Whitney U test to probe for systematic differences in Pv and Pm ranks ., P-values were ordered from least significant to most significant , and the lowest 100th centile ( i . e . the most significantly associated SNPs ) was compared to the remaining 99th percentile for each of the five traits ., For BMI , SNPs in the lowest 100th centile of the Pm distribution had markedly higher Pv ranks ( i . e . more significant Pv ) than the remaining SNPs ( PMann–Whitney = 1 . 46×10−5; Table 5 ) ., Even when excluding previously established lead SNPs ( Pm<5×10−8 ) for BMI ( or SNPs +/-500kb proximal ) , SNPs from the lowest 100th centile of the Pm rank-ordered distribution had higher Pv ranks than the remaining SNPs ( PMann–Whitney = 4 . 30×10−4; Table 5 ) ., Conversely , no difference in Pv ranks was observed for SNPs from the lowest 100th centile of the Pm rank-ordered distribution for the four blood lipid traits; this may reflect trait-specific G×E effects or differences in statistical power by trait ., No differences in Pv ranks between SNPs from the lowest 99th centile of the Pm rank-ordered distribution compared to SNPs from the 98th to 1st centiles of the distribution were observed for any trait ( PMann–Whitney>0 . 05; Table 5 ) ., Similarly , no difference in Pm ranks was observed for SNPs from the lowest 100th centile of the Pv rank-ordered distribution for any traits ( PMann–Whitney>0 . 05; Table 6 ) ., To assess whether a trait with a non-normal distribution ( e . g . BMI ) or strong marginal associations could cause spurious association between the marginal and variance signals , we recapitulated the analysis pipeline ( correlation analysis , enrichment analysis , comparisons of rank Pm and Pv values ) in simulations described in the Materials and Methods ., Careful assessment of results emanating from these simulations did not reveal evidence of type I error inflation caused by the non-normal distribution of an outcome trait nor strong marginal effects ., For instance , we extracted correlation P-values of Pm , Pv and Pint generated from 5 , 000 simulations ., QQ-plots of the 5 , 000 correlation P-values , 2 , 500 binomial P-values , and 2 , 500 Mann-Whitney U test P-values revealed no inflation ( S1A–S1C Fig , S2A and S2B Fig and S3A and S3B Fig , respectively ) ., Repeating these analyses on subsets of SNPs with low Pm values did not materially change the results ., Collectively , our analyses highlight a few variants with genome-wide significant marginal effects that may be strong candidates for G×E interactions owing to their strong concurrent variance heterogeneity P-values ., For BMI , such SNPs are also overrepresented in the nominally significant part of the Pv distribution ., FTO is an excellent example , as it conveys strong marginal effects 13 , exhibits high between-genotype heterogeneity here ( Tables 2 and 3 and Fig 1B ) and elsewhere 5 , and reportedly interacts with physical activity , diet and other lifestyle exposures 2 , 14 , 15 and is associated with macronutrient intake 16 , 17 ., Although variance heterogeneity tests are potentially powerful screening tools for G×E interactions , like most interaction tests , they may be bias prone ., For example , apparent differences in phenotypic variances across genotypes may be caused by scaling , particularly when the phenotypic means also differ substantially 18 , such that the per-genotype means and variances for index traits are correlated ., However , where necessary we transformed variables , and the correlations between Pm and Pv were generally weak , excluding this as a likely source of bias ., Using simulated data , we investigated whether the non-normal distribution of a trait can cause a spurious association between marginal and variance signals , which we show is highly improbable ., Through further simulations , we assessed whether SNPs with large marginal effects inflate Pv , but observed no inflation , indicating that large genetic marginal effects do not artificially inflate variance heterogeneity to a meaningful extent , and SNPs with low Pm and low Pv-values are thus likely to be strong candidates for G×E interactions , at least in the case of BMI ., It might also be that combining populations from ancestral ( e . g . , hunter-gatherers ) and contemporary environments increases variance heterogeneity owing to diversity in population substructure rather than G×E interactions per se 19 ., However , this seems unlikely here , as the cohorts examined are from Westernized European-ancestry populations ., There are several additional explanations for between-genotype variance heterogeneity , such as variance misclassification that can occur when the index variant is located within a haplotype containing rare functional variants that convey strong marginal effects 5 ., Hence , although variance heterogeneity tests represent a useful data-reduction step , before conclusions are drawn about the presence or absence of G×E interactions , index variants should be validated by testing their interactions with explicit environmental exposures , as we did here with smoking and physical activity ., However , genome-wide G×E interactions datasets are not comprised of functionally validated G×E interactions , as no such resource is currently available for human complex traits ., This limitation inhibits the extent to which causal effects can be attributed to the top-ranking loci and their interactions with smoking or physical activity ., We conclude that the common approach of prioritizing loci with established genome-wide significant association signals without further discrimination for G×E interaction analyses might be useful , but the efficiency of such analyses could be substantially improved by focusing on variants with low P-values for both variance heterogeneity and marginal effects ., We provide these rankings here to facilitate this approach ., We performed a genome-wide search for SNPs whose associations with the following traits are characterized by high between-genotype variance heterogeneity: BMI , TC , TG , HDL-C and LDL-C ., The variance heterogeneity analyses were performed using Levene’s test 20 in up to 44 , 211 participants of European descent from seven population-based cohorts ., Descriptions of these cohorts are presented in S2 Table ., To minimize bias that might result from unequal sample sizes between SNPs when calculating the correlations between the P-values from the marginal ( Pm ) and variance heterogeneity ( Pv ) meta-analyses , we restricted the sample size for analyses to 26 , 000 participants for BMI and to 24 , 000 participants for lipid traits ( S4 Fig ) ., A detailed summary of sample sizes , genotyping platforms , genotype calling algorithms , sample and SNP quality control filters , and analysis software for all participating cohorts are provided in S2 and S3 Tables ., For each individual , SNPs were imputed using the CEU reference panel of HapMap II 7 ( S2 Table ) ., We excluded SNPs with low imputation quality ( below 0 . 3 for MACH , 0 . 4 for IMPUTE , and 0 . 8 for PLINK imputed data ) , Hardy-Weinberg equilibrium P <10−6 , directly genotyped SNP call rate < 95% , and minor allele frequency ( MAF ) < 1% ., We identified SNPs that have been robustly associated ( P<5x10-8 ) with the five cardiometabolic traits in European ancestry populations: 77 SNPs associated with BMI discovered by GIANT 9; and 58 SNPs associated with LDL-C , 71 SNPs associated with HDL-C , 74 SNPs associated with TC , and 40 SNPs associated with TG 10 , 11 discovered by GLGC ., We used Levene’s test 20 to identify SNPs that show heterogeneity of phenotypic variances ( σi2 ) across the three genotype groups at each SNP locus ( i = 0 , 1 , or 2 ) ., We first log10 transformed all five traits followed by a z-score transformation by subtracting the sample mean and dividing by the sample standard deviation ( SD ) , and further Winsorized the z-score values at 4 SD ., The transformed phenotype Y was then used to calculate Z , defined by the absolute deviation of each participant’s phenotype from the sample mean of his or her respective genotype group at a given SNP locus ., For each trait , participating cohorts provided the necessary summary statistics for each genotype at each marker 8 ., Specifically , the per genotype group counts ( n0s , n1s , n2s ) , per genotype means ( Z¯0s , Z¯1s , Z¯2s ) , and per genotype group variances of Z ( σ0s2 , σ1s2 , σ2s2 ) were centrally collected and meta-analyzed ., The minimum number of observations per genotype group required is 30 participants per cohort ., Meta-analyses were performed using the following formula , derived previously 8:, L= ( N−3 ) ( 3−1 ) ⋅ ( ∑i=02γi⋅ ( ∑sZ¯is⋅ωis ) 2− ( ∑i=02∑sZ¯is⋅ωis⋅γi ) 2 ) ∑i=02 ( ∑s ( σZis2⋅ωis−σZis2N⋅γi+Z¯is2⋅ωis ) ⋅γi− ( ( ∑sZ¯is⋅ωis ) 2⋅γi ) ), Where N is the combined sample size , Z¯is and σZis2 are the sample mean and variance of Z in the ith genotype group of the sth study , respectively ., When combining summary-level data to calculate the Levene’s test statistics L , the following natural weights ωis and γi were calculated: ωis=nis∑snis and γi=niN , where ni the sum of genotype counts in the ith genotype group across all participating cohorts ., These weights are determined by the frequency of the marker amongst the cohorts , such that the sum of both weights is equal to 1 , i . e . ∑sωis=1 and ∑iγi=1 ., The meta-analysis Levene’s test P-value is obtained by comparing L to an F-distribution with df1 = 2 and df2 = N-3 ., Marginal effects P-values for BMI and the relevant lipid traits were obtained from publically available GWAS summary data from the GIANT 9 and GLGC 10 , 11 consortia , respectively ( all cohorts included here in the Levene’s meta-analysis were also included in the GIANT and GLGC datasets ) ., To illustrate our findings , we rank-ordered the P-values ( from lowest to highest ) from both marginal effects and variance effects analyses for all 1 , 927 , 671 SNPs so that the lowest P-value for a given trait was assigned a rank equal to the lowest 100th centile ., These rank-scaled distributions for Pm for all five traits are presented in Fig 1 ., We calculated Spearman’s correlations for each of the five cardiometabolic traits between Pm and Pv ., This was done using a pruned set of SNPs ., Pruning was performed in the TwinGene cohort using the--indep-pairwise 50 5 0 . 1 command in PLINK 21 by calculating LD ( r2 ) for each pair of SNPs within a window of 50 SNPs , removing one of a pair of SNPs if r2>0 . 1; we proceeded by shifting the window 5 SNPs forwards and repeating the procedure ., Spearman’s correlations were computed for categories of SNPs:, i ) all pruned SNPs ,, ii ) the subset of SNPs that was nominally significant ( Pm<0 . 05 ) in the marginal effects analysis ,, iii ) the subset of SNPs with Pm<10−4 in the marginal effects analysis , and, iv ) SNPs that were previously established in conventional marginal effects GWAS meta-analyses ( Pm<5×10−8 ) ., We also compared Spearman’s correlations between these categories of SNPs using the test for equality of two correlations 22 ., Next , we performed enrichment analyses to test if there was a higher number of established SNPs in the nominally significant variance P-value ( Pv<0 . 05 ) distribution than expected by chance under the binominal distribution ., We also tested if there is a difference in Pv ranks for SNPs from the lowest 100th centile of the Pm rank-ordered distribution for all five traits and the rest of SNPs in the pruned set of SNPs using the Mann–Whitney U test , including and excluding established SNPs ( or SNPs that were +/-500kb from the reported lead SNP ) ., This analysis was repeated for SNPs from the 99th centile vs SNPs from 1st to 98th centiles of the Pm rank-ordered distribution ., The same Mann–Whitney U tests were used to study differences in Pm ranks for SNPs from the lowest 100th and 99th centiles of the Pv rank-ordered distribution and the rest of SNPs in the pruned set of SNPs ., All analyses were performed using Stata 12 ( StataCorp LP , TX , USA ) , unless specified otherwise ., We used now published data from 210 , 316 European-ancestry adults ( from the GIANT consortium ) pertaining to marginal effects meta-analyses for BMI that had been performed separately by strata of smoking ( 45 , 968 smokers vs . 164 , 355 non-smokers ) 23 ., The genetic marginal effect estimates , calculated separately within each of the two strata , were compared using a heterogeneity test 12 to infer the presence or absence of SNP × smoking interaction effects ., The same analyses were performed using physical activity as a binary stratifying variable in up to 180 , 287 European-ancestry adults ( 42 , 065 physically active vs . 138 , 222 physically inactive ) 24 ., We calculated Spearman correlations between the P-values derived from the marginal effects meta-analysis and the Pint from the interaction effects meta-analysis ( i . e . , the between-strata heterogeneity test for SNP × smoking and SNP × physical activity interactions from the GIANT consortium ) ; these tests were undertaken for all SNPs and those SNPs that were nominally significant ( Pm<0 . 05 ) in the marginal effects analysis ., We then performed enrichment analyses to test if the numbers of nominally significant ( Pint<0 . 05 ) GWAS-derived SNPs from both SNP × physical activity and SNP × smoking analyses were greater than expected by chance under the binomial distribution ., We further calculated the OR of having Pint<0 . 05 given Pv<0 . 05 versus Pv≥0 . 05 both SNP × physical activity and SNP × smoking interaction analyses in a pruned set of TwinGene SNPs produced using the—indep-pairwise 50 5 0 . 8 command in PLINK 21 ., Thereafter , we calculated the average rank for each SNP’s ranking on the Pint rank-ordered distributions from the SNP × smoking and SNP × physical activity interaction analyses and performed enrichment analysis using these average ranks with >95th centile instead of Pint<0 . 05 as the cut-off ., We simulated genetic data for 44 , 000 individuals from a pruned set of 50 , 335 SNPs with allele frequencies , effect estimates and Pm values drawn from the GIANT consortium ., We generated an outcome trait by summing the products of the simulated allele counts and effect estimates over all SNPs for each individual , and subsequently added a randomly generated non-normal error term such that the trait resembles the observed distribution of the transformed BMI trait used in the main ( real data ) analyses ., We also simulated a fixed binary interacting factor with 30% prevalence ., Using this simulated dataset , we calculated Pm , Pv and Pint values for each SNP and undertook, i ) pairwise Spearman correlation analyses between Pm , Pv and Pint values ( 5 , 000 simulations ) ,, ii ) enrichment analysis using binomial tests ( 2 , 500 simulations ) and, iii ) Mann-Whitney U tests to determine systematic differences in Pv and Pm ranks ( 2 , 500 simulations ) ., Following the same pipeline , we created additional simulated datasets narrowing down SNPs to, i ) those with Pm values from the lowest percentile ( n = 504; highest Pm = 5×10−3 ) and to, ii ) genome-wide significant SNPs ( n = 71; Pm<5×10−8 ) , and tested the pairwise Spearman correlation for Pm , Pv and Pint values ( 1 , 000 simulations for both sets ) ., Simulations were run using the statistical software R ( v . 3 . 3 . 2 ) ., 25

Introduction, Results, Discussion, Materials and methods

Phenotypic variance heterogeneity across genotypes at a single nucleotide polymorphism ( SNP ) may reflect underlying gene-environment ( G×E ) or gene-gene interactions ., We modeled variance heterogeneity for blood lipids and BMI in up to 44 , 211 participants and investigated relationships between variance effects ( Pv ) , G×E interaction effects ( with smoking and physical activity ) , and marginal genetic effects ( Pm ) ., Correlations between Pv and Pm were stronger for SNPs with established marginal effects ( Spearman’s ρ = 0 . 401 for triglycerides , and ρ = 0 . 236 for BMI ) compared to all SNPs ., When Pv and Pm were compared for all pruned SNPs , only BMI was statistically significant ( Spearman’s ρ = 0 . 010 ) ., Overall , SNPs with established marginal effects were overrepresented in the nominally significant part of the Pv distribution ( Pbinomial <0 . 05 ) ., SNPs from the top 1% of the Pm distribution for BMI had more significant Pv values ( PMann–Whitney = 1 . 46×10−5 ) , and the odds ratio of SNPs with nominally significant ( <0 . 05 ) Pm and Pv was 1 . 33 ( 95% CI: 1 . 12 , 1 . 57 ) for BMI ., Moreover , BMI SNPs with nominally significant G×E interaction P-values ( Pint<0 . 05 ) were enriched with nominally significant Pv values ( Pbinomial = 8 . 63×10−9 and 8 . 52×10−7 for SNP × smoking and SNP × physical activity , respectively ) ., We conclude that some loci with strong marginal effects may be good candidates for G×E , and variance-based prioritization can be used to identify them .

Most contemporary studies of gene-environment interactions focus on gene variants that are known to bear strong and reliable associations with the traits of interest ., The strategy is intuitive because it helps limit the number of tests performed by focusing on a relatively small number of gene variants ., However , this approach is predicated on an implicit assumption that these loci are strong candidates for interactions owing to their established relationships with the index traits ., The counter-argument is that , because these loci have highly consistent signals within and between populations that vary by environmental characteristics , the probability that these variants interact with other factors is low ., The current analysis tests whether variants with strong marginal effects signals ( i . e . , those prioritized through conventional genome-wide association analyses ) are strong or weak candidates for gene-environment interactions ., Here we describe analyses focused on lipids and BMI that test this hypothesis by comparing marginal effect signals with variance effect signals and those derived from explicit genome-wide , gene-environment interaction analyses ., We conclude that for BMI , there are features of the top-ranking marginal effect loci that render them stronger candidates for interactions than is true of variants with weaker marginal effects signals ., These findings are likely to help optimize the efficiency of future gene-environment interaction analyses by providing evidence-based rankings for strong candidate loci .

genome-wide association studies, medicine and health sciences, quantitative trait loci, sociology, social sciences, physical activity, mathematics, statistics (mathematics), genome analysis, research and analysis methods, public and occupational health, genomic signal processing, lipids, proteins, mathematical and statistical techniques, lipoproteins, statistical methods, consortia, genetic loci, cholesterol, biochemistry, signal transduction, cell biology, meta-analysis, genetics, biology and life sciences, physical sciences, genomics, cell signaling, computational biology, human genetics

journal.pbio.2002439

Dynamic patterning by the Drosophila pair-rule network reconciles long-germ and short-germ segmentation

Like other arthropods , the fruit fly Drosophila melanogaster has a segmented body plan ., This segmental pattern is laid down in the embryo during the first 3 hours of development ., During this time , the anteroposterior ( AP ) axis of the blastoderm is progressively patterned down to cellular-level resolution by an elaborate , multi-tiered network of genes and their encoded transcription factors 1 , 2 ., These genes were first identified in a landmark genetic screen 3 , 4 , and their regulatory interactions have subsequently been dissected by 3 decades of genetic experiments ., Along the way , this body of research has revealed many fundamental principles of transcriptional regulation 5 , and Drosophila segmentation remains a central model for developmental systems biology today ., Much of the “heavy lifting” of segment patterning is carried out by the so-called “pair-rule” genes , which make up the penultimate tier of the Drosophila segmentation cascade ., The pair-rule genes are the first genes to be expressed in spatially periodic patterns in the Drosophila embryo and are collectively responsible for patterning the expression of the “segment-polarity” genes , which organise and maintain segmentally reiterated compartment boundaries termed “parasegment boundaries” ., Notably , this involves transducing a double segment pattern of early pair-rule gene expression , in which each set of stripes is offset slightly from the others , into a single-segment pattern of segment-polarity gene expression , in which most genes are expressed in discrete , non-overlapping domains 6–8 ., There are 7 canonical pair-rule genes: hairy 9 , even-skipped ( eve ) 10 , runt 11 , fushi tarazu ( ftz ) 12 , odd-skipped ( odd ) 13 , paired ( prd ) 14 , and sloppy-paired ( slp ) 15 ., 5 of these genes ( hairy , eve , runt , ftz , and odd ) are known as the “primary” pair-rule genes because they are expressed earlier than the 2 “secondary” pair-rule genes , prd and slp 16 ., ( Note that these terms have a somewhat tortuous history , and older literature will classify the genes differently . ), Each of the primary pair-rule genes is initially patterned by spatial inputs from the upstream tier of transcription factors , encoded by the “gap” genes , which are expressed in broad , overlapping AP domains during cellularisation 17 ., This patterning occurs in an ad hoc manner , with specific combinations of gap factors regulating the expression of particular pair-rule stripes through discrete “stripe-specific” enhancer elements 18–21 , which act additively with one another ., For certain pair-rule genes , such as eve , this regulation is sufficient to generate an overall pattern of 7 equally spaced stripes along the future trunk of the embryo 16 , 22–24 ., For other pair-rule genes , such as odd , the gap-driven pattern is irregular and may have missing stripes 16 ., In these cases , the initial patterns are regularised by cross-regulatory “zebra” enhancer elements 25–27 , which take periodic inputs from other pair-rule factors and yield periodic outputs ., Similar zebra elements are responsible for driving the periodic expression of the secondary pair-rule genes , which turn on after the primary pair-rule patterns have refined 16 , 28 ., At gastrulation , the segment-polarity genes turn on , activated by a broadly expressed transcription factor , Odd-paired ( Opa ) 29 , and spatially regulated by the pair-rule genes 6 , 7 , 30 , 31 ., Opa activity also “rewires” the regulatory interactions between the pair-rule genes , causing several of their expression patterns to transition dynamically from double- to single-segment periodicity ( i . e . , from 7 stripes to 14 stripes ) 32 ., These pair-rule factors ( and/or their paralogs ) then play roles in the segment-polarity network , which also contains several components of the Wingless ( Wg ) and Hedgehog signalling pathways 33–37 ., The Drosophila gap gene network has been used frequently in recent years as a case study for the application of dynamical systems 38–40 and information theory 41–43 approaches to developmental patterning , but the pair-rule network has received little system-level attention ., Indeed , the most recent models of pair-rule patterning 8 , 44 date from more than 10 years ago ., Since these were published , 3 important discoveries have been made about segment patterning , all of which challenge established assumptions about the Drosophila segmentation cascade and all of which concern the pair-rule genes in some way ., So long as the pair-rule network remains poorly understood , key questions raised by these findings will go unanswered ., The first discovery is from comparative studies in other arthropod embryos ., Drosophila is a “long-germ” embryo , patterning almost all of its segments simultaneously in the blastoderm prior to germ-band extension 45 ., However , the ancestral mode of arthropod development is “short-germ” embryogenesis , in which segmentation is sequential and coordinated with germ-band elongation 46–48 ., Orthologs of the pair-rule genes play a key role in segment patterning in all arthropods studied thus far ( for example , 49–52 ) , but in short-germ embryos , their expression has been shown to oscillate in a posterior “segment addition zone” throughout germ-band extension 53–55 ., This periodic dynamic expression indicates that in these organisms , they are either components of or entrained by a segmentation “clock” 56 ., How the expression of the pair-rule genes in long-germ embryos such as Drosophila relates to their expression in short-germ embryos ( for example , the model beetle , Tribolium castaneum ) is unclear ., It is thus not understood how long-germ segmentation was derived from short-germ segmentation , an important evolutionary transition that has occurred multiple times independently within the higher insects 57 ., The second discovery stems from quantitative studies of Drosophila segmentation gene dynamics ., These studies have revealed that the domains of gap gene expression in the trunk of the embryo shift anteriorly across the blastoderm over the course of nuclear division cycle 14 ( cellularisation ) 58–60 ., The shifts are mirrored downstream in similarly shifting expression of the pair-rule stripes 59 , 61 , a finding that is at odds with existing models of segment patterning , which assume these stripes to be static domains 6 , 8 , 44 , 62–64 ., While we know that these subtle shifts are ultimately driven by feedback interactions within the gap gene network 38 , 39 , 65–67 , their functional role ( if any ) remains unclear ., The final key finding relates to the structure of the pair-rule network itself ., In a recent paper on the pair-rule network 32 , Michael Akam and I showed that many of the regulatory interactions between the pair-rule genes are temporally regulated ( by Opa , as described above ) ., We argued that the pair-rule network is best viewed as 2 distinct networks , 1 operating during cellularisation and 1 during gastrulation , each with a specific topology and resultant dynamics ., Analysing the “early” ( cellularisation-stage ) and “late” ( gastrulation-stage ) pair-rule networks separately should lead to a better understanding of pair-rule patterning and might also reveal why the network shows this bipartite organisation in the first place ., In this paper , I present a new model of the pair-rule system , which incorporates the stage-specific architecture of the pair-rule network ., I take the set of identified genetic interactions between the pair-rule genes as a starting assumption , formalise them in a Boolean logical model , and use dynamical simulations to analyse how they collectively lead to complex pattern formation ., I find that gap-mediated kinematic shifts are required for correctly phasing the pair-rule stripes , something that proves crucial for downstream segment patterning ., I also find that graded Eve activity is not strictly necessary for pair-rule patterning , and I explain the aetiology of the surprisingly severe eve null mutant phenotype ., Finally , I show that a slightly modified version of the Drosophila pair-rule network gains the capacity to pattern in both simultaneous and sequential modes , conceptually reconciling long- and short-germ segmentation ., Fig 1A summarises the inferred regulatory interactions between the pair-rule genes ., Following Clark and Akam ( 2016 ) 32 , individual interactions are assigned to distinct “early” and “late” networks , which operate during mid-cellularisation or late cellularisation/gastrulation , respectively ., Note that a few regulatory interactions ( e . g . , repression of ftz by Eve ) are common to both networks , but the majority are restricted to a single phase of patterning ., Gene regulatory network models represent “intellectual syntheses” of diverse experimental data 70 ., I arrived at the topologies in Fig 1A by carefully analysing relative expression data in tightly staged wild-type embryos and cross-referencing these observations with the large number of mutant and misexpression experiments recorded in the genetic literature ( for example , 8 , 30 , 31 , 64 , 71–78 ) ., In places where the data were particularly ambiguous , I also re-characterised pair-rule gene expression in select pair-rule mutants in order to pick apart direct versus indirect regulatory interactions ., In almost all cases , the interactions in the network diagrams have been previously inferred by multiple sets of researchers; my contribution has been ( 1 ) to bring this body of work together into something consistent and relatively complete and ( 2 ) to recognise the distinction between the early and late phases of regulation , rather than pooling all interactions into a single network ., Most of the evidence and reasoning behind the inferred interactions ( and interactions inferred to be absent ) in Fig 1A are described in Appendix 1 of Clark and Akam ( 2016 ) 32 ., Additional evidence in favour of the “early” cross-regulatory interactions between the primary pair-rule genes ( boxed yellow area in Fig 1A , left ) is presented in S1 Text , based on patterns of pair-rule gene expression in hairy , eve , and runt mutants ., Two things are immediately clear from the network diagrams ., First , the direct regulatory interactions between the pair-rule genes are overwhelmingly repressive ., This is consistent with a mode of patterning consisting of spatially ubiquitous activation ( by maternally provided factors , for example ) combined with precisely positioned repression from other segmentation genes 76 , 79–81 ., While certain of the pair-rule factors ( e . g . , Ftz and Prd ) have been shown to quantitatively up-regulate the expression of other pair-rule genes and thus contribute to this background activation in a spatially modulated way 82–84 , these effects do not , for the most part , seem to be important for qualitatively determining the spatial pattern of pair-rule gene expression and so have been omitted from the diagram ., Most described incidences of one pair-rule gene genetically activating another pair-rule gene are instead indirect ( i . e . , mediated by the direct repression of another repressor ) ., Second , the 2 networks have very different structures , presumably reflecting the different patterning function each must perform ., During mid-cellularisation , pair-rule gene cross-regulation is responsible for refining many of the pair-rule stripes and standardising their phasing relative to other pair-rule stripes , resulting in a regular repeating pattern of double-segment periodicity ., This is carried out by the early network , which is sparse , composed of unidirectional regulatory interactions , and has no feedback loops ., Two of the pair-rule genes , eve and hairy , are patterned by gap factors rather than other pair-rule factors and so represent “input-only” factors to the network 16 ., The remaining primary pair-rule genes ( runt , ftz , and odd ) do receive extensive gap inputs at early stages of cellularisation , but , by mid-cellularisation , their patterns are largely specified by other pair-rule genes ( see S1 Text ) ., ( Note , however , that some aspects of the ftz pattern cannot be explained by pair-rule inputs alone , see Appendix 2 of 32 . ), The secondary pair-rule genes turn on later ( prd at mid-cellularisation and slp towards the end of cellularisation ) and are patterned by primary pair-rule genes ., Overall , the early network has a hierarchical structure , in which Eve and Hairy convey positional information derived from the gap factors to the remaining primary pair-rule genes and eventually to the secondary pair-rule genes ., The late network , on the other hand , is extremely dense and consists largely of mutually repressive pairs of interactions ., It is responsible for converting a double-segmental pattern of overlapping stripes into a segmental pattern of discrete segment-polarity fates ., This is the final step in the Drosophila “segmentation cascade” and completes the transition from the analog ( graded ) positional information carried by the maternal and gap gene products to the essentially digital positional information carried by the segment-polarity genes 64 ., The numerous positive ( i . e . , double-negative ) feedback loops within the late network are consistent with it acting like a multi-stable switch , individual segment-polarity fates representing attractor states towards which the system will rapidly converge ., As described above , gap inputs and the early pair-rule network combine to establish a repeating double-segmental pattern of pair-rule gene expression ., The positional information within this pattern is then converted into a stable output pattern of segment-polarity states by the late network ., Each initial double-segment repeat is about 7–8 nuclei wide , and each specified segment will consist of at least 3 distinct states characterised by the expression of engrailed ( en ) , odd , and slp , respectively ( Fig 1B ) ., The en and odd stripes are about 1 nucleus wide , while the slp stripes are about 1–2 nuclei wide ., Parasegment boundaries form wherever En and Slp domains abut , while Odd provides a buffer zone that preserves the AP polarity of each segment ., ( This tripartite segment pattern conforms to prescient theoretical predictions made by Hans Meinhardt in the early 1980s 62 , 85 , 86 . ), It is crucial that all 3 domains are specified within each segment—and that they are in the correct order—because patterning defects such as boundary losses , ectopic boundaries , and/or polarity reversals arise when the pattern is perturbed 8 , 31 , 33 , 76 , 87 ., The extremely high resolution of the final segmental output pattern implies that the initial double segment pattern established by the early pair-rule network must contain sufficient positional information to allow almost every nucleus to be distinguished from its immediate neighbours ., We are thus left with 2 questions ., First , how does the early network establish a situation in which the different nuclei within a double-segment repeat each expresses a unique combination of pair-rule factors ?, Second , how exactly is this code “read” by the late network ?, ( Or , in other words , which sets of initial conditions will result in a cell following an expression trajectory that ends at , for example , stable en expression , rather than stable odd or stable slp ? ), In later sections , I address these questions by simulating and analysing the networks shown in Fig 1A ., However , before getting into specifics of how particular genes are regulated and expressed , it is worth considering a more fundamental question: where is the positional information coming from in the first place ?, The topology of the early network ( Fig 1A , left ) implies that , to a first approximation , all the positional information in the final pattern must trace back to the expression patterns of just 2 factors , Eve and Hairy ., Boolean ( ON/OFF ) combinations of Eve and Hairy would only be sufficient to specify 4 different domains within each double-segment repeat ( Fig 1C , top ) , whereas the real output pattern consists of at least 6 distinct domains ( i . e . , En , Odd , Slp , En , Odd , Slp ) ., How is it possible that just 2 independent spatial signals are able to give rise to such a high-resolution final output ?, One potential answer is that stripes of Eve and/or Hairy might carry quantitative information that permits them to convey more than 2 “states” within the positional code ., Since the early 1990s , this idea has been applied to the graded margins of the early Eve stripes 30 , 75 , 88 ., These stripes have been proposed to act as local morphogen gradients , repressing different target genes at different concentration thresholds and thus differentially positioning their respective expression boundaries ., Current models of pair-rule patterning rely on the assumption that there are 4 functionally distinct levels of Eve activity across an Eve stripe ( from the centre to the edge: HIGH , MEDIUM , LOW , and OFF ) 8 , 44 ., These different levels would provide cellular-level resolution within each double-segment repeat and , combined with information from the Hairy stripe , allow each nucleus to be uniquely specified ( Fig 1C , bottom ) ., While a given concentration of Eve protein may well repress its various targets with different efficacies , it is unlikely that segment patterning relies significantly upon this mechanism , for 3 main reasons ., First , for the model to be viable , the Eve stripes would need to provide an accurate and precise set of positional signals within each double-parasegment repeat , i . e . , the Eve stripes would have to be extremely regular and all share the same shape and amplitude ., However , more posterior Eve stripes show significantly lower expression levels than more anterior Eve stripes throughout most of cellularisation 59 ., Furthermore , pair-rule transcripts are apically localised , and therefore pair-rule gene expression becomes effectively cell autonomous soon after membrane invagination begins 89 , 90 ., This means that , unlike for the gap genes ( whose transcripts remain free to diffuse between neighbouring nuclei ) , for eve , there is little or no spatial averaging to buffer the high intrinsic noise of transcription 91 , 92 ., This reduces the precision of the Eve signal and thus its capacity to reliably convey analog information ., Second , the morphogen model also requires the readout of the Eve signal to be very sensitive; i . e . , eve target genes would have to reliably discriminate between different Eve expression levels and pattern their expression boundaries accordingly ., However , it is not clear that this actually occurs within the embryo—for example , the model proposes that graded Eve stripes result in offset boundaries of the Eve targets odd and ftz , but recent observations indicate that these offsets are in fact produced by other mechanisms 32 ., Third , the morphogen model does not explain the full severity of the eve null mutant phenotype , in which aberrant expression patterns are seen even in regions that would be outside the Eve stripes in wild-type embryos ., Neither does the morphogen model explain the patterning robustness of eve heterozygotes , in which halving Eve expression levels fails to perturb the overall pattern of segment-polarity domains ., How , then , might the spatial resolution of the segment pattern be explained if not by an Eve morphogen gradient ?, Traditional models of Drosophila segmentation are essentially static: each tier of segmentation gene expression provides a single set of spatial signals , which is transduced into a new set of spatial signals by the tier below ., This simplifies the real situation in the embryo , in which both gap and pair-rule expression domains shift subtly from posterior to anterior over time 59 , 93 ., Explicitly considering these temporal aspects of segmentation gene expression suggests an alternative segment patterning mechanism: using the temporal dynamics of a relatively coarse pair-rule signal to provide high-resolution spatial information across each pattern repeat ., A signal that varies over time can be used to convey an arbitrary quantity of information , even if each reading of that signal provides very little ( think of Morse code or binary storage ) ., The eve and hairy stripes continue to be regulated by gap inputs throughout most of cellularisation and therefore shift across nuclei in concert with the gap domains ., This means that , rather than each nucleus having to deduce its position from a single level of , e . g . , Eve protein ( as in the morphogen model ) , the nucleus actually experiences a temporal sequence of Eve protein levels ., Strikingly , an overall shift of just 2 nuclei would be theoretically sufficient for a Boolean Eve stripe to , on its own , specify the positions of all 6 segment-polarity domains within a double-segment repeat ( Fig 1D ) ., This kind of mechanism would , however , rely on the downstream targets of Eve and Hairy being able to decode a temporal sequence of Eve/Hairy expression and convert it into an appropriate segment-polarity fate ., In the following sections , I carry out simulations and analysis of the network shown in Fig 1A and , based on the results , argue that the cross-regulatory interactions between the pair-rule genes function to achieve exactly this task ., In order to investigate how pair-rule patterning works , I used the networks shown in Fig 1A to create a toy model of the pair-rule system and then simulated pair-rule gene expression across an idealised 1-dimensional tissue ., In this section , I briefly describe the structure and assumptions of the modelling approach; a full description of the model plus details of all simulations are given in S2 Text ., ( Source code for running the simulations is available in S1 and S2 Files , while pair-rule networks in SBML-qual format are available in S3 and S4 Files ) ., The genes whose regulation I model explicitly are the 7 pair-rule genes , plus en , whose product plays a key role in regulating late pair-rule gene expression ., I have also included 4 inputs that are extrinsic to the system: 2 temporal signals , Caudal ( Cad ) 94 and Opa , and 2 signals to represent the positional information provided by the gap system , “G1” and “G2” ., Cad represses the secondary pair-rule genes during early stages of patterning 87 , while Opa turns on midway through patterning and triggers the switch from the early network to the late network 32 ., G1 is responsible for patterning the hairy pair-rule stripes while G2 is responsible for patterning the eve pair-rule stripes ., G1 and G2 do not represent specific gap factors but are instead an abstraction of the spatial inputs ( i . e . , stripe boundary locations ) provided by the gap system as a whole ., Each gene in the system is represented by a Boolean variable , and its control logic is formalised using logical rules ( essentially equivalent to the “logical equations” used in Sanchez and Thieffry 2003 44 or the “vector equations” used in Peter et al . 2012 70 ) ., For example , if Opa is OFF ( early network ) , odd is expressed only if both Hairy and Eve are also OFF , while if Opa is ON ( late network ) , odd expression relies on all of Runt , En , and Slp being OFF ( compare Fig 1A ) ., In most cases ( apart from , e . g . , activation of en by Ftz or Prd ) , gene activation is assumed to be driven by some ubiquitous background factor ( s ) and is not explicitly included in the model ., The network simulation proceeds by discrete time steps , with expression output at time point t + 1 calculated from the state of the system at time point t ., Because of the speed and dynamicity of segment patterning , time delays associated with protein synthesis and protein decay imply that protein and transcript expression domains for a given gene will often be non-congruent within the Drosophila embryo ., This is likely to be significant for patterning , and I therefore approximate this effect by adding simple time delay rules into the simulation ., Each gene has associated “synthesis delay” and “decay delay” parameter values s and d , both of which are integers representing a certain number of time steps ., ( Once a gene turns on , transcript will be present immediately , but protein will only appear s time steps later . Similarly , once a gene turns off , transcript will disappear immediately , but protein will only disappear after another d timesteps have elapsed . ), For parsimony ( and consistent with real expression kinetics 82 ) all pair-rule genes and en are assigned the same delay value , which applies to both s and d ., The specific value of this delay is fairly arbitrary , because the ratio between different delays is what affects how the system behaves , but I have chosen this value to be 6 ., Given that the half-life of ftz RNA during cellularisation is 7 minutes 79 , this means that each time step in the simulation can be thought of as representing on the order of 1 minute of real developmental time ., The time delays of the other components in the network ( Cad , Opa , G1 , and G2 ) are assigned appropriate values relative to this timescale , so that their simulated behaviour roughly approximates their spatiotemporal expression in a real embryo ., The simulation is set up to occur across a row of 20 “cells” , an idealised representation of the AP axis ., This row of cells is not meant to correspond to a specific region of the Drosophila embryo but rather to be generally representative of patterning within the main trunk ( i . e . , pair-rule stripes 3–6 ) , in which pair-rule genes are not additionally affected by cephalic or terminal factors ., Each cell within this “tissue” is simulated independently , starting from a specific set of initial conditions ., ( As mentioned above , pair-rule transcripts are apically localised , and therefore the cross-regulation between the pair-rule genes is likely to be effectively cell autonomous from roughly mid-cellularisation onwards . ), The starting conditions for each cell usually involve specifying the appropriate expression of G1 and G2 , setting Cad to ON , and setting all other genes to OFF ., G1 and G2 are initialised with patterns that are offset by 2 cells and repeat every 8 cells , meaning that the hairy and eve stripes specified by these inputs will partially overlap and exhibit a double-segment periodicity , as in real embryos ., Gap inputs into runt , ftz , and odd are omitted , meaning that their early expression is organised entirely by the spatial inputs from Hairy and Eve ., As the simulation proceeds , Cad protein will disappear , allowing prd to turn on 87 , followed by slp ., ( Note that we do not currently know how exactly the timing of slp expression is controlled , so in order to reproduce the timing observed in the embryo , slp expression in the simulation requires Prd expression to already be present . ), Shortly afterwards , Opa protein will turn on , switching the control logic of pair-rule gene expression to the late network and causing pair-rule gene expression to eventually reach a final , stable state ., After the switch to the late network , the gap factors and Hairy cease to regulate the pair-rule genes and then fade away , as in real embryos ., Note that the model just described , which is Boolean , deterministic , and uses discrete time steps , is not designed to capture the full complexity of the embryo ( in which gene expression is , of course , quantitative , stochastic , and continuous ) ., Rather , it represents a tool to expose the key mechanisms of patterning—and to delineate how much of what is observed in the embryo follows simply from the qualitative structure of the regulatory network ., It also provides an important sanity check of the inferences that led to that structure being proposed in the first place ., Using the model described above , I first simulated a scenario in which gap gene inputs and hence the pair-rule stripes of Hairy and Eve are completely static ., The results are shown in S9 Movie and are summarised in Fig 2A ., Under these conditions , the positional information provided by Hairy and Eve is essentially equivalent to the situation diagrammed in Fig 1C ( bottom ) and thus has no possibility of generating the correct segmental output ., Unsurprisingly , the simulation does a bad job of recapitulating the patterns of pair-rule gene expression seen in real embryos ( see below ) ., In particular , at the end of the simulation , there is no en expression anywhere at all , and neither odd nor slp is expressed in a segmental pattern ., I then simulated a scenario in which the gap gene inputs and hence the Hairy and Eve stripes shift anteriorly over time ., Given that the shift rate in real embryos is of the same order as the synthesis and decay rates of the segmentation gene products , I set the rate of these shifts to be such that the time taken for an expression domain to shift anteriorly by 1 cell is equal to the synthesis/decay delay parameter value of the pair-rule genes , i . e . , 6 time steps ( see S2 Text ) ., The simulation output for this scenario is shown in S10 Movie and summarised in Fig 2B ., Even though the pair-rule network is unchanged and the Hairy and Eve stripes retain the same pattern and relative phasing as for the static simulation , the model now performs completely differently ., Qualitative aspects of actual pair-rule gene expression ( i . e . , whether the expression domains of each pair of genes are congruent , overlapping , abutting , or separate , and the way this changes over time ) are recapitulated remarkably well ., For all pair-rule genes except prd , the match between the model output and the real spatiotemporal dynamics of gene expression is about as close as could be achieved by a simple , Boolean model—a few examples are highlighted in Fig 3 , and the full set of comparisons is shown in S1 Fig . For prd , the real spatiotemporal expression profile is only partially recovered: the early pair-rule stripes are positioned correctly but do not refine correctly at later stages—they narrow rather than split , meaning that alternate segmental stripes are missing from the final pattern ( S3 Fig ) ., However , the prd domains missing from the simulation are not actually required for segment boundary patterning in real embryos ( they are not reflected in the larval cuticles of prd mutants , although they do have minor effects on wg expression 6 , 31 , 95 ) ., Accordingly , the simulation still generates the correct final segmental output: a repeating pattern of En , Odd , Slp x2 , En , Odd , Slp x2 ., These results tell us a number of things ., First , it is not strictly necessary to invoke morphogen gradients in order to account for Drosophila pair-rule patterning ., Second , posterior-to-anterior shifts of the Hairy and/or Eve stripes appear to be crucial for properly patterning the other pair-rule genes , and analysing the different behaviour of the static and shifting simulations should reveal exactly why ., Third , the model as formulated is too simple to explain important aspects of the prd expression profile ., Additional complexities that influence prd expression in real embryos could include ( 1 ) additional spatial or temporal regulatory inputs missing from the model , ( 2 ) quantitative information from existing spatial or temporal inputs that is not captured by the use of Boolean variables , or ( 3 ) differential synthesis/degradation rates of particular segmentation gene products not accounted for by the equal time delays assumed by the model ., At least the first option seems to apply , as I have recently discovered that the Sox transcription factor Dichaete 96 , 97 also affects prd regulation 87 ., Above , I described how the final segmental output consists of the pattern En , Odd , Slp , En , Odd , Slp across each double-parasegment repeat ., I then used a dynamical model of the pair-rule system to sho

Introduction, Results, Discussion, Materials and methods

Drosophila segmentation is a well-established paradigm for developmental pattern formation ., However , the later stages of segment patterning , regulated by the “pair-rule” genes , are still not well understood at the system level ., Building on established genetic interactions , I construct a logical model of the Drosophila pair-rule system that takes into account the demonstrated stage-specific architecture of the pair-rule gene network ., Simulation of this model can accurately recapitulate the observed spatiotemporal expression of the pair-rule genes , but only when the system is provided with dynamic “gap” inputs ., This result suggests that dynamic shifts of pair-rule stripes are essential for segment patterning in the trunk and provides a functional role for observed posterior-to-anterior gap domain shifts that occur during cellularisation ., The model also suggests revised patterning mechanisms for the parasegment boundaries and explains the aetiology of the even-skipped null mutant phenotype ., Strikingly , a slightly modified version of the model is able to pattern segments in either simultaneous or sequential modes , depending only on initial conditions ., This suggests that fundamentally similar mechanisms may underlie segmentation in short-germ and long-germ arthropods .

Segmentation in insects involves the division of the body into several repetitive units ., In Drosophila embryos , all segments are patterned rapidly and simultaneously during early development , in a process known as “long-germ” embryogenesis ., In contrast , many insect embryos retain an ancestral or “short-germ” mode of development , in which segments are patterned sequentially , from head to tail , over a period of time ., In both types of embryo , the patterning of segment boundaries is regulated by a network of so-called “pair-rule” genes ., These networks are thought to be quite divergent due to the different expression patterns observed for the pair-rule genes in each case: regularly spaced arrays of transient stripes in Drosophila , and dynamic expression within a posterior “segment addition zone” in short-germ insects ., However , even in Drosophila , a clear understanding of pair-rule patterning has been lacking ., Here , I make a computational model of the Drosophila pair-rule network and use simulations to explore how segmentation works ., Surprisingly , I find that Drosophila segment patterning relies on pair-rule gene expression moving across cells over time ., This conclusion differs from older models of pair-rule patterning but is consistent with the subtly dynamic nature of pair-rule stripes in real embryos , previously described in quantitative studies ., I conclude that long-germ and short-germ segmentation involve similar expression dynamics at the level of individual cells , even though they look very different at the level of whole tissues ., This suggests that the gene networks involved may be much more conserved than previously thought .

invertebrates, genetic networks, morphogenic segmentation, molecular probe techniques, gene regulation, animals, simulation and modeling, animal models, developmental biology, drosophila melanogaster, model organisms, network analysis, experimental organism systems, molecular biology techniques, embryos, morphogenesis, drosophila, research and analysis methods, embryology, computer and information sciences, probe hybridization, gene expression, molecular biology, insects, arthropoda, fluorescent in situ hybridization, eukaryota, cytogenetic techniques, gene identification and analysis, genetics, biology and life sciences, organisms

journal.pbio.2005264

Peripherally derived macrophages modulate microglial function to reduce inflammation after CNS injury

The immune system plays a pivotal role in development and homeostatic functions of the central nervous system ( CNS ) 1 ., Immune system dysfunction can give rise to CNS disease 2 and its response to injury shapes recovery 3–5 ., The cellular response to CNS injuries is stereotyped and involves the rapid reaction of tissue-resident microglia 6 , 7 and the recruitment of myeloid cells such as neutrophils and monocyte-derived macrophages ( MDMs ) within days 5 , and the activation of lymphocytes from the blood , meninges , and choroid plexus 8 ., Two cell types that dominate CNS lesions are resident microglia and infiltrating MDMs ., It is known that microglia and MDMs are ontogenetically distinct 9 , 10 , express cell type–specific transcripts and proteins 11 , 12 , and can thus potentially perform different functions at the site of injury 13–16 ., However , their relative contribution to the injury response , and subsequent recovery , remains unclear ., It is not known if these two cell types interact to specifically modulate each other’s function ., Two of the primary functions of microglia and MDMs during CNS injury are phagocytosis and propagation of inflammation 17 ., We have shown previously that after traumatic spinal cord injury ( SCI ) , cessation of microglia phagocytosis coincides with the infiltration of MDMs 14 ., In animal models of stroke and CNS autoimmune disease , expression profiling of microglia after injury or at the onset of disease ( at the point of MDM infiltration ) shows that pathways involving core functions of microglia , such as inflammation , RNA transcription , and phagocytosis , are significantly down-regulated 16 , 18 , 19 ., We therefore hypothesized that MDMs entering the CNS signal to resident microglia and modulate their function ., Three to five days after CNS injury , infiltrating macrophages are distributed across lesion sites and are therefore potentially able to interact with microglia 14 , 19 , 20 ., After SCI , resident microglia and MDMs both increase in number around the lesion site 21 ., We observed that MDMs and microglia are often in close proximity to one another ( S1A Fig ) ; however , whether they can regulate each other’s functions is not known ., To assess whether the cessation of microglial phagocytosis seen after SCI correlates with the entry of MDMs into the CNS , as reported previously 14 , we used antibody-conjugated magnetic-bead sorting to isolate CD11b+ cells from spinal cord lesions in LysM-eGFP knock-in mice ., LysM-eGFP reporter mice strongly express eGFP in myelomonocytic cells 22 but in less than 3% of microglia after SCI and other CNS injuries 19 , 20 , 23 ., Therefore , isolated CD11b+ positive cells could be defined as resident microglia ( CD11b+/eGFP−ve ) or infiltrating MDMs ( CD11b+/eGFP+ve/Ly6G−ve ) after extraction from the injured spinal cord ., To assess the effect of infiltrating cells on resident microglia , cells were isolated from spinal cord lesions prior to significant MDM infiltration ( one day ) and after infiltration ( three days ) ., Immediately after isolation , cells were placed in vitro and incubated with pHrodo-labeled myelin for four hours ., Flow cytometric analysis revealed that increasing numbers of MDMs at the lesion site correlates with a significant reduction in microglial phagocytosis ( Fig 1A–1E ) ., We therefore hypothesized that MDMs entering the CNS signal to resident microglia and modulate their function ., As an additional note , we detected significantly more MDMs one day after injury , as compared to naïve spinal cord ( S1B Fig ) ., This is likely to represent macrophages in the process of entering the tissue , either through the vasculature or the meninges but before they have been reported to be seen in the parenchyma ., It is possible that they could release signaling molecules from these locations that influence microglia , as seen in the reduction of phagocytosis at day one after SCI , compared to uninjured ( naïve ) mice ( Fig 1E ) ., There is a further significant increase in infiltration of MDMs by day three , compared to one day after SCI ( S1B Fig ) ., To directly test the hypothesis that MDMs modulate microglial function , we created a bilaminar culture system by plating bone marrow–derived macrophages ( BMDMs ) on coverslips with small paraffin spacers , which were then placed into wells containing adult microglia ( S2D Fig ) ., Primary adult mouse microglia were cultured under conditions that retain a transcriptional profile more similar to their in vivo counterparts , as compared to other media conditions , primary microglial cultures from neonates or microglial cell lines 11 ( S2A and S2B Fig ) ., Other features , such as genes that reflect region-specific factors or function , may be altered when cells are placed in culture ., The addition of Transforming growth factor ( TGF ) -β was not only necessary for a gene expression profile more similar to freshly isolated microglia but also showed greater ramified morphology in culture at seven days ( S2C Fig ) ., Microglia “signature” genes were down-regulated during lipopolysaccharide ( LPS ) -induced inflammation , which supports their description as homeostatic 11 ., There is no further modulation of these microglial genes in the presence of macrophages ( S2E Fig ) ., As a control for cell numbers , we assessed modulation of these microglial genes by coculturing with microglia instead of BMDMs ( S2F Fig ) ., Suppression of inflammatory genes in adult microglia does not occur when cocultured with adult microglia ., For these experiments , adult mouse microglia were cultured with or without adult microglia and stimulated with LPS ( 100 ng/mL ) ( S2F Fig ) ., Using this bilaminar in vitro system , we assessed if soluble factors released by these two cell types affect phagocytic function in one another ., Microglia and BMDMs were cocultured in the bilaminar system for 24 hours , separated , and incubated with pHrodo-labeled myelin for 90 minutes ., Phagocytic uptake was assessed with flow cytometry ., Uptake of pHrodo-labeled myelin was significantly decreased in microglia cultured in the presence of BMDMs compared with microglia cultured alone ( Fig 1F and 1G ) ., Surprisingly , myelin phagocytosis by BMDMs was significantly increased after coculture with adult microglia ( Fig 1H and 1I ) ., These findings reveal direct communication between the two cell types divergently affecting phagocytic function ., We next investigated macrophage effects on inflammatory gene expression in microglia and macrophages from adult mice and humans ., We recently described a mathematical model of cytokine signaling , which found four inflammatory cytokines , interleukin ( IL ) -1β , tumor necrosis factor ( TNF ) , IL-6 , and IL-10 , to be key nodes in the inflammatory network 24 ., After LPS stimulation , coculture with BMDMs significantly down-regulated these genes in mouse and human microglia ( Fig 1J and 1L ) ., In contrast , IL-1β expression was increased in LPS-stimulated macrophages cocultured with mouse microglia ( Fig 1K ) ., There were nonsignificant trends towards increases in inflammatory cytokines in human macrophages in the combined presence of LPS and microglial cells ( Fig 1M ) , contrasting with the significant suppression of these genes in microglia in the same conditions ( Fig 1L ) ., These experiments show the direct suppressive effects of macrophages on microglia , and reciprocal but divergent effects of microglia on macrophages ., In the bilaminar culture system in mouse and human microglia , we found significant suppression of pro-inflammatory cytokines IL-1β , TNF , and IL-6 in the presence of macrophages ( Fig 1J and 1L ) ., We also found a significant reduction in human microglia of IL-10 , a canonical brake on inflammation 25 , 26 ., Despite IL-10’s opposing function to the pro-inflammatory cytokines , all four cytokines are increased with LPS 24 hours after stimulation ., We have recently described the complexities of cytokine networks over time and shown how the modulation of one cytokine in the system may result in varying temporal kinetics of the others 26 ., Here , we examined a single time point , but as all four cytokines were suppressed by the presence of macrophages , we sought to examine whether microglial transcription , in general , became globally suppressed by macrophages ., To understand the global effects of macrophage suppression on microglia , we transcriptionally profiled LPS-activated adult mouse microglia in the presence or absence of macrophages ., A total of 1 , 076 genes were significantly differentially regulated in activated microglia in presence of macrophages , with approximately 50% up-regulated and 50% down-regulated ., Ingenuity pathway analysis ( IPA ) revealed that the most dysregulated canonical signaling pathways were those related to nuclear factor ( NF ) -κB signaling , a master regulator of inflammation ( Fig 2A ) and apoptosis and cell death ( Fig 2B ) ., Network analysis revealed three major clusters of genes distributed across two distinct regions reflecting distinct gene co-expression patterns ( Fig 2C ) ., In the major gene cluster 1 ( 185 genes ) , which comprised genes down-regulated within microglia in the presence of macrophages , we found that the top upstream regulators , predicted to be inhibited with high confidence , included MyD88 , IL-1β , and TNF ( Fig 2D ) ., These analyses support our findings that gene expression within the major inflammatory cascades in microglia is suppressed in the presence of macrophages ., As IL-1β was one of the key cytokines to be differentially regulated in microglia and macrophages in the coculture experiments ( Fig 1J and 1K ) , we searched for factors known to regulate IL-1β in microglia ., Prostaglandin E2 ( PGE2 ) signaling via the EP2 receptor has been reported to reduce IL-1β expression in microglia 27 ., In addition , EP2 receptor signaling has also been reported to reduce phagocytosis 28–30 ., We therefore hypothesized that PGE2 signaling via microglial EP2 receptors could be responsible for the suppressive effects of macrophages ., Inducible microsomal prostaglandin E synthase-1 ( mPGES ) and EP2 receptor were up-regulated during inflammation in mouse and human microglia and macrophages in vitro ( Fig 3A–3D and S3A Fig ) and in vivo in mice after SCI ( Fig 3E ) ., EP2 receptor expression was also up-regulated 22-fold when assessed by transcriptional array ( S3A Fig ) ., Transcript levels for EP1 and 4 were down-regulated in mouse microglia in vitro , suggesting they are not involved ( S3A Fig ) ., mPGES and hydroxyprostaglandin dehydrogenase ( HPGD ) work in concert to regulate PGE2 production and release as HPGD converts PGE2 to its biologically inactive metabolite 31 ., In human macrophages , mPGES was significantly increased and HPGD was significantly down-regulated during inflammation ( LPS ) when cocultured with microglia , compared with macrophages cultured alone ( without LPS ) ( Fig 3B ) , suggesting greater PGE2 production in stimulated macrophages in the presence of microglia ., The expression of EP2 receptor in human microglia was not significantly up-regulated upon stimulation with LPS ( Fig 3D ) ., However , the data show that it is expressed , allowing the cells to detect PGE2 ., Taken together , these experiments show that the components needed to allow PGE2 signaling at the EP2 receptor in microglia are up-regulated during inflammation and may be utilized for macrophage–microglia communication ., To functionally assess the role of the EP2 receptor , we treated adult mouse microglia with the EP2 specific agonist , Butaprost ., Treatment of microglia significantly reduced TNF , IL-6 , and IL-10 in the same manner as the macrophage-mediated suppression of these genes ( Fig 3F ) ., This contrasted with the effects of Butaprost on BMDMs , which only reduced TNF expression ( S3E Fig ) ., Butaprost also significantly reduced phagocytosis by microglia ( Fig 3G ) ., Macrophage suppression of microglial phagocytosis was rescued by the selective EP2 antagonist , PF-04418948 32 ( Fig 3H ) ., In addition , unlike wild-type ( WT ) BMDMs , macrophages that lack mPGES ( mpges −/− ) do not suppress microglial phagocytosis ( Fig 3I ) ., Taken together , these results show that PGE2 produced by peripherally derived macrophages plays a major role in suppression of microglial phagocytic function via EP2 receptors ., To investigate this mechanism in vivo , we performed SCI in WT and mpges −/− mice and assessed the phagocytic microglial response ., In addition , as our in vitro data show that PGE2 derived from macrophages suppresses microglial phagocytosis , we performed SCI in C–C chemokine receptor type 2 ( CCR2 ) null mice to compare the microglial response in a lesion that contains very few MDMs ( see Fig 5A and 5B ) 33 ., Three days after SCI in WT , mpges −/− , and CCR2 null mice , microglia were isolated from the lesion and ex vivo phagocytosis of pHrodo ( Green ) -myelin was quantified by flow cytometry ( Fig 4A and 4B ) ., Microglia from WT mice after SCI ( i . e . , with macrophages and PGE2 present in the lesion ) , showed low levels of phagocytosis ( Fig 4A and 4B ) ., In contrast , microglia from mpges −/− and CCR2 null mice showed significantly increased phagocytosis ( Fig 4A and 4B ) , indicating that in the absence of PGE2 , or the absence of macrophages in the lesion , microglial phagocytic function is increased ., To assess the effect of blocking the EP2 receptor , in vivo , on microglial phagocytosis , we injected pHrodo-myelin into the brain ( corpus callosum ) of WT mice , together with vehicle or the EP2 receptor antagonist ( PF-04418948 ) ., We assessed brain tissue three days after injection with immunofluorescence and confocal microscopy ., In the corpus callosum of mice injected with pHrodo-myelin and vehicle , the area containing pHrodo-myelin is mainly populated with CD11b+ , Tmem119-negative cells ( MDMs ) ( Fig 4C ) ., In mice injected with pHrodo-myelin and EP2 antagonist , there is a significant increase in Tmem119+ microglial cells in the area containing pHrodo-myelin ( Fig 4D and 4E ) ., Many of these Tmem119+ microglia contain or are closely associated with the fluorescently tagged myelin ( Fig 4D and 4F ) ., Importantly , the percentage of total microglial cells in the corpus callosum that are in contact with or contain pHrodo-myelin is significantly increased in EP2 antagonist–injected mice compared with controls ( Fig 4C , 4D and 4F ) ., These data indicate that blocking the EP2 receptor pathway in vivo promotes microglial phagocytic activity and increases recruitment of microglia to the site of injury ., In summary , in vivo and in vitro studies suggest macrophage production of PGE2 acts at the EP2 receptor to mediate suppression of microglial phagocytosis ., We also sought to investigate macrophage effects on microglial cell death and proliferation , as microglia proliferate at the sites of CNS injury 14 , 34 ., In vitro , the presence of macrophages did not significantly reduce microglial viability ( S4A Fig ) ., However , when combined with inflammation ( LPS stimulation ) , macrophages significantly reduced viability , suggesting an increase in microglial cell death , compared with untreated microglia ( S4A Fig ) ., This fits with our transcriptional profiling data , which show that apoptotic and cell death pathways are significantly dysregulated under similar conditions ( Fig 2B ) ., These data suggest that macrophages can affect microglial apoptosis under inflammatory conditions in vitro and warrant further analysis ., To investigate macrophage effects on microglial proliferation , we used Click-iT EdU assay ( Invitrogen ) , which incorporates 5-ethynyl-2′-deoxyuridine ( EdU; a nucleoside analog of thymidine ) to DNA during active DNA synthesis ., We found no evidence that macrophages affect the proliferation of microglia in vitro with bilaminar cultures ( S4B and S4C Fig ) ., We also depleted circulating macrophages prior to their infiltration after SCI using clodronate liposomes in LysM-eGFP mice to assess microglial proliferation with Ki67 ., Clodronate significantly depleted eGFP+ infiltrating cells at the lesion site when assessed five days after injury ( S4D and S4E Fig ) , but this had no effect on microglial proliferation ( S4F and S4G Fig ) ., To investigate whether macrophages modulated another important microglial function , namely rapid process extension towards microlesions , we induced laser lesions in organotypic hippocampal slice cultures ( OHSCs ) , as done previously 35 , in the presence or absence of BMDMs ., We also investigated the initial reaction of microglial morphologies in an in vivo model of traumatic brain injury ( TBI ) 36 with local administration of Butaprost versus vehicle control ., Initial process extension and acute morphological changes that are dependent on purinergic receptor signaling 6 , 36 are not affected in OHSCs by the presence of macrophages ( S4H–S4M Fig ) or altered by the EP2 agonist in TBI ( S4N–S4O Fig ) ., In summary , coupled with our transcriptional profiling data , these results highlight that macrophages target specific pathways and functions in microglia , such as inflammation and apoptosis , but do not affect microglial proliferation or their rapid process extension response to injury ., These findings may be of significance , as the rapid reaction of microglia in the early phases of injury are thought to be protective 35 , 37 , 38 ., Our results suggest that peripheral macrophages do not interfere with this response ., Mice that lack CCR2 cannot successfully recruit MDMs to traumatic CNS lesions 33 ., We showed that this leads to increased microglial phagocytic activity ( Fig 4A and 4B ) ., Therefore , we next assessed whether lack of MDM infiltration affects activation of microglia and functional recovery in vivo after SCI using CCR2 null mice 33 ., Although CCR2 has been reported to be expressed in injured neurons , expression appears variable between species and between investigators 39–41 ., We are not aware of other genetic evidence for CCR2 protein expressed in neurons 13 , 42 ., CCR2 is also reported to be expressed in a subset of T-regulatory cells ( Tregs ) 43 ., Although it is unknown what role CCR2+ Tregs play after SCI , it has been shown that T cells may play a beneficial role in CNS repair 44 ., Despite this , a major phenotype five days after SCI was that MDMs were almost absent in the lesioned spinal cord of CCR2rfp/rfp ( CCR2 KO ) mice , as compared with WT mice ( Fig 5A and 5B ) ., Neutrophil infiltration was not affected five days after injury ( Fig 5C ) ., Three days after injury , there was a trend to an increase , but this was not statistically significant ( S4P Fig ) ., To study the impact of the absence of MDMs on microglial-mediated inflammation , we isolated microglia from SCI lesions four and seven days after injury and assessed the expression profiles of 86 inflammatory genes using a PCR array ., Four days after SCI in WT mice , approximately half of these genes were significantly down-regulated in microglia ., Seven of the top 20 most down-regulated genes in WT mice were significantly less down-regulated in CCR2 KO mice , indicating that the absence of MDMs at the lesion resulted in less suppression of microglial inflammation ( Fig 5D ) ., Moreover , pathways identified as being suppressed by macrophages in our in vitro bilaminar system , such as inflammation driven by MyD88 and NF-κB and apoptotic pathways driven by Trp53 and Bcl2 were also significantly more suppressed when MDMs were present at the lesion ( Fig 5D ) ., Seven days after SCI , microglial inflammatory genes continued to be dysregulated ( Fig 5E ) ., Components of important inflammatory pathways continued to be significantly less suppressed in the absence of macrophages , such as MyD88 , Il17a , and Cxcl2 ., However , dysregulation of microglial gene expression was less unidirectional than factors associated with pro-inflammatory response , such as Irf1 , Tlr9 , and Il12b , and growth factors such as Egf and Tgfb1 associated with recovery were significantly more down-regulated in microglia , in the absence of infiltrating macrophages ( Fig 5E ) ., Our previous work shows that initial perturbation to inflammatory networks is likely to cause unpredictable patterns of expression at later time points 24 , 26 ., Therefore , to investigate the long-term consequence of the initial loss of microglial suppression and subsequent dysregulation , we assessed long-term activation of microglia and its impact on functional recovery after SCI in CCR2 KO versus WT mice ( Fig 6A–6G ) ., Up-regulation of CD11b ( αM integrin ) is well established as a readout of microglial/macrophage activation 45–47 , and CD86 is a costimulatory receptor up-regulated during inflammation in microglia in vivo 48 , 49 ., CD11b expression in microglial cells was already increased at a cellular level in mice lacking macrophage infiltration ( CCR2 KO ) seven days after SCI versus controls ( Fig 6A ) ., There was a trend to an increase in CD86+ microglia but it did not reach statistical significance ( S4Q Fig ) ., To assess microglial activation 28 days after injury , we quantified CD11b and CD86 expression by immunofluorescence of tissue sections caudal to the lesion epicenter ( Fig 6C and 6D ) ., Importantly , 28 days after SCI , CD11b and CD86 immunoreactivity was greater in area and intensity in CCR2 KO mice despite the lack of infiltrating MDMs , which also express CD11b and CD86 ( Fig 6C and 6D ) ., In other words , CD11b and CD86 expression is markedly increased in microglia in CCR2 KO mice 28 days after SCI ., These results indicate that preventing the communication between MDMs and resident microglia contribute to long-term microglial activation after CNS injury ., We also investigated whether increased microglial inflammation in the absence of infiltrating macrophages in CCR2 KO mice influences functional recovery and histopathology ., CCR2 KO mice showed greater myelin loss , an indicator of secondary tissue damage , caudal to the lesion 28 days after SCI , compared with controls ( Fig 6E and 6F ) ., The increased microglial activation associated with the absence of macrophage influx after SCI is associated with worse locomotor recovery in CCR2 KO mice compared with WT controls , as measured by the Basso Mouse Scale ( BMS ) ( Fig 6g ) ., The role of microglia in CNS injury and disease is now considered critical to the pathological process 50 ., Our work suggests a novel concept that macrophages from the peripheral circulation , which enter the CNS after injury , may act to modulate microglial activation , thus preventing microglial-mediated acute and chronic inflammation ., These findings support previous work that shows blocking CCR2-dependent macrophage infiltration with an anti-CCR2 antibody worsens locomotor recovery after CNS injury 51 , 52 ., However , these earlier papers 51 did not show how macrophages mediate these effects ., Our work now shows that infiltrating macrophages suppress microglial activation by reducing their expression of inflammatory molecules and ability to phagocytose , thus preventing chronic microglia-mediated inflammation in the CNS ., Other work has suggested that subsets of infiltrating macrophages are detrimental to SCI 53 , and it has been reported that CCR2 antagonism , producing a 50% reduction in infiltrating macrophages , is beneficial after TBI 54 ., Also , CCR2 KO mice showed acute and transient behavioral improvement after intracerebral hemorrhage ( one and three days ) , but this was not sustained at seven days 55 ., Here , our finding that inhibition of macrophage entry to the CNS results in a worse outcome after SCI is supported by work that defines specific beneficial macrophage populations in multiple CNS injury and disease contexts 56–59 ., Our results now suggest a new mechanism by which infiltrating macrophages mediate their beneficial actions via the regulation of microglial activation ., Such a mechanism will operate alongside macrophage-intrinsic mechanisms ., We observed that macrophages regulate microglia in both mouse and human cells ., This is important as it represents an independent replication of the concept in a different laboratory ., It shows that macrophages derived from the blood ( human ) or bone marrow ( mouse ) appear to have similar effects on microglia and that the findings may be relevant to human disease ., It is still controversial as to whether the net effect of microglia is beneficial or detrimental to CNS injury 4 , 5 , 60; however , the kinetics of the microglial response must be considered ., There is evidence that the initial responses of microglia , which occur in the first few minutes to several hours after injury , are beneficial and limit the expansion of CNS lesions 35 , 37 , 38 ., Conversely , prolonged microglial dysregulation and neuroinflammation are deleterious to the CNS 61 ., Therefore , the initial microglial response to injury may be beneficial , but prolonged inflammation and activation are potentially detrimental to recovery ., Our data suggest that macrophages play a role in mitigating this detrimental response by infiltrating the injury site and reducing microglial-mediated inflammation and chronic microglial activation ., The absence of this protective mechanism may contribute to a worse outcome when infiltrating macrophages do not enter the CNS after SCI in CCR2 KO mice ., To our knowledge , this is the first description of such a cellular mechanism to reduce deleterious consequences of CNS injury ., Microglial cells are now prime targets in drug discovery for CNS injuries and neurodegenerative diseases 62 ., To properly assess the roles of microglia in CNS injury , our data suggest that the context , timing , and interaction with macrophages should also be considered ., Attempts to target either of these two cell populations should be approached with caution and a better understanding is needed of their divergent and complex roles in injury and disease ., The heterogeneity and region-specific differences in microglia 63 and macrophage populations 64 will also need to be considered ., Recent work has shown that peripherally derived macrophages can engraft the brain and maintain an identity distinct from microglia 65 , thus opening the possibility for therapeutic engraftment of MDMs to the CNS and allowing macrophage–microglia cross talk in disease contexts ., Cell-to-cell interactions between different brain resident cell types are now becoming evident 66–69 ., During inflammation , microglia have also been shown to drive astrocyte-mediated toxicity 66 , which , subject to the context , is dependent on microglial NF-κB signaling 67 ., Our data show that peripheral macrophages regulate the NF-κB signaling pathway in microglia that , in turn , reduce inflammatory mediators , such as TNF , which can drive astrocyte-mediated toxicity 66 , 67 ., This raises the possibility that macrophage signaling to microglia may have subsequent effects in other CNS cells , such as astrocytes ., In summary , we suggest that infiltrating macrophages provide a natural control mechanism against detrimental acute and long-term microglial-mediated inflammation ., Manipulation of peripherally derived infiltrating cells may provide a therapeutic treatment option to target microglial-mediated mechanisms that cause or exacerbate CNS injury and disease ., All animal procedures were approved by the Animal Care Committee of the Research Institute of the McGill University Health Centre and followed the guidelines of the Canadian Council on Animal Care and the ARRIVE guidelines for reporting animal research 70 ., Before surgical interventions and cardiac perfusions , mice were deeply anesthetized by intraperitoneal injection of ketamine ( 50 mg/kg ) , xylazine ( 5 mg/kg ) , and acepromazine ( 1 mg/kg ) ., Human brain tissue was collected during clinical practice , fully anonymized , and therefore available for use under the legislation of the Tri-Council Policy Statement two and Plan daction ministériel en éthique de la reserche et en intégrité scientifique of Quebec and Canada ., This study was carried out in accordance with the guidelines set by the Biomedical Ethics Unit of McGill University , approved under reference ANTJ2001/1 , and conducted in accordance with the Helsinki Declaration ., C57BL/6 ( Charles River , St-Constant , QC ) , heterozygote lysM+/EGFP mice ( kindly provided by Dr . Thomas Graf and obtained from Dr . Steve Lacroix ) ; homozygote CCR2RFP/RFP and their C57BL/6J controls ( Jackson ) ; heterozygote Cx3CR1+/gfp ( Jackson ) and Ptges−/− mice 71 ( obtained from Dr . Maziar Divangahi , McGill University ) , aged 8–14 weeks , were kept under a 12-hour light/dark cycle with ad libitum access to food and water ., The LysM-eGFP mouse was originally generated by Faust and colleagues , 2000 ., EGFP is expressed specifically in the myelomonocytic lineage by using homologous recombination ., This was achieved by knocking the enhanced GFP ( EGFP ) gene into the murine lysozyme M ( lys ) locus and using a targeting vector , which contains a neomycin resistant ( neo ) gene flanked by LoxP sites and “splinked” ends , to increase the frequency of homologous recombination ., Removal of the neo gene through breeding of the mice with the Cre-deleter strain led to an increased fluorescence intensity 22 ., Female mice were anesthetized by intraperitoneal injection of ketamine ( 50 mg/kg ) , xylazine ( 5 mg/kg ) , and acepromazine ( 1 mg/kg ) and a moderate contusion injury ( 50 kDa force; 500–600-μm tissue displacement ) was made at the T11 thoracic vertebral level using the Infinite Horizon Impactor device ( Precision Scientific Instrumentation , Lexington , KY ) , as previously described 72 ., Male C57BL/6 mice ( 8–12 weeks ) were , anesthetized , transcardially perfused and brains removed and kept in ice-cold Hanks Balanced Salt Solution ( HBSS ) ., Cerebellum and meninges were removed , and brain was cut into small pieces ., Tissue was enzymatically dissociated using Neural Tissue Dissociation Kit ( P ) ( Miltenyi cat # 130-092-628 ) according to the manufacturer’s instructions , with modifications ., Following digestion , tissue was transferred to a 15-mL Dounce on ice and homogenized with 20× passes of a large clearance pestle ., Tissue was resuspended in 35% isotonic percoll and overlaid with HBSS ., Following centrifugation ( 400g; 45 minutes ) , myelin was removed , and pure populations of microglial cells were selected using CD11b microbeads ( Miltenyi #130-093-634 ) , as previously described 63 , 73 ., Pure ( >95% CD11b-positive ) adult microglia were resuspended at 8×105 cell−mL ( approximately two brains per mL ) in media ( DMEM F12 , 10% fetal bovine serum FBS , 1% penicillin/streptomycin P/S ) , with 10% L-cell conditioned media , a source of macrophage colony-stimulating factor ( M-CSF ) , or 10 ng/mL recombinant mouse M-CSF ( R and D cat no 416-ML-010/CF ) , and 50 ng/mL recombinant human TGF-β1 ( Miltenyi cat no: 130-095-067 ) to maintain their transcriptional profile , as previously described 11 ., Cells were plated in pre-coated poly-L-lysine plates , media was changed at three days , and experiments were performed at seven days ., At seven days , microglia were collected and microglial “signature” genes assessed by qPCR ., Network analysis and Markov clustering ( see below ) were performed to assess conditions driving cells to a similar phenotype of their freshly isolated counterparts ., CD11b+ cells ( myeloid cells ) were collected by magnetic bead cell sorting , as above , from SCI lesions of lys-EGFP-ki mice ( 2 . 5 mm either side of the epicenter ) at 1 or 3 days after injury , or from uninjured controls ., The CD11b+ fraction was immediately plated into 96 well plates ( pre-coated poly-L-lysine , one animal per well ) in DMEM F12 media containing 10% FBS ., Cells were incubated for four hours with pHrodo ( Invitrogen

Introduction, Results, Discussion, Materials and methods

Infiltrating monocyte-derived macrophages ( MDMs ) and resident microglia dominate central nervous system ( CNS ) injury sites ., Differential roles for these cell populations after injury are beginning to be uncovered ., Here , we show evidence that MDMs and microglia directly communicate with one another and differentially modulate each other’s functions ., Importantly , microglia-mediated phagocytosis and inflammation are suppressed by infiltrating macrophages ., In the context of spinal cord injury ( SCI ) , preventing such communication increases microglial activation and worsens functional recovery ., We suggest that macrophages entering the CNS provide a regulatory mechanism that controls acute and long-term microglia-mediated inflammation , which may drive damage in a variety of CNS conditions .

The immune and the central nervous systems are now thought to be inextricably linked ., In response to injury , the immune system shapes CNS recovery through a complex of molecular and cellular mediators ., However , it is unclear how the kinetics , magnitude , and components of this response can be harnessed to improve CNS restoration ., The two immune cells that dominate CNS lesions are resident microglia—already present before the injury—and infiltrating macrophages , which enter from the blood after injury ., Both cells are thought to be critical to the outcome , yet it is unknown if , or how , they interact ., To investigate this , we used mouse and human cells in microglia–macrophage coculture systems and an in vivo model of traumatic spinal cord injury ., We show that infiltrating macrophages suppress key functions of microglia , such as removal of tissue debris and propagation of inflammation ., Preventing macrophage–microglia communication increases microglial activation and worsens recovery ., We suggest that infiltrating macrophages from the blood provide a natural control mechanism against detrimental acute and long-term microglial-mediated inflammation ., Manipulation of the peripheral macrophages may provide a therapeutic treatment option to target microglial-mediated mechanisms that cause or exacerbate CNS injury and disease .

blood cells, traumatic injury, medicine and health sciences, innate immune system, immune cells, immune physiology, pathology and laboratory medicine, cytokines, nervous system, immunology, cell processes, microglial cells, age groups, developmental biology, adults, signs and symptoms, molecular development, white blood cells, inflammation, animal cells, neurotrauma, glial cells, critical care and emergency medicine, phagocytosis, immune response, trauma medicine, immune system, people and places, diagnostic medicine, cell biology, anatomy, central nervous system, spinal cord injury, neurology, physiology, biology and life sciences, cellular types, population groupings, macrophages

journal.pcbi.1000130

Optimal Compensation for Temporal Uncertainty in Movement Planning

In the execution of any movement , there is always timing uncertainty ., This uncertainty has two major consequences ., First , it limits performance on any task for which there are costs associated with temporal imprecision ., Second , it has implications for how the motor system should plan movements when the costs of temporal imprecision are asymmetric ., In hurrying to catch a subway train , for example , the cost of arriving early is usually small compared to the cost of arriving late and missing the train ., An optimal movement planner must take into account temporal reward asymmetries in forming movement plans ., The complexity of movement planning under risk is further increased because temporal uncertainty in the motor system changes constantly ., Two major sources of variation in temporal uncertainty occur over different time courses and have different properties: One is a uniform , global shift in temporal uncertainty possibly due to aging , fatigue , injury or disease 1–9 ., The second is a linear increase in the standard deviation of movement duration with increases in mean movement duration 10 ., Here we use a model of optimal temporal movement planning to investigate the control of movement duration in the face of these two types of temporal uncertainty while human subjects attempted to touch a computer screen within a specified temporal window ., We introduced asymmetries in the penalties imposed for early vs . late movement timing ( Figure 1A ) , while at the same time increasing subjects temporal uncertainty by adding Gaussian noise with 25 ms standard deviation ( see Methods ) ., As in all models of motor planning and motor control based on decision theory , we are concerned with the interplay of three elements: possible decisions ( here planned movement time , τ ) , uncertainty in the mapping of motor decisions to motor outcomes ( represented by the family of probability distributions pt|τ ) , and the costs/benefits resulting from those motor outcomes , G ( t ) ., The mathematical models considered here are part of a growing literature on Bayesian decision models of motor phenomena , such as models of motor adaptation 11–13 and motor planning/control e . g . , 14–21 , including the use of prior information in spatial 16 , 18 and temporal 17 motor planning , the use of asymmetric cost functions in spatial motor planning 14–15 , 19 and when selecting a speed-accuracy tradeoff 20–21 ., The neural computation of decision variables such as those considered here and in previous work has also begun to be investigated 22–25 ., Figure 1B illustrates the computations needed to maximize expected gain with temporally asymmetric penalties ., When discussing movement duration , we must distinguish between the planned arrival time , denoted τ , and the actual arrival time , t ., When movements are executed , the actual arrival time will be unpredictably earlier or later than τ ., In Figure 1B we show four possible choices of τ and outline the calculation of expected gain for each ., Note that the optimal planned arrival time need not fall within the temporal reward window ., Human performance will be optimal if the CNS learns its linear temporal uncertainty function , ( 1 ) as it relates to planned movement time ( τ ) , and uses this information ( ασ and βσ ) to plan reach times that maximize expected gain ., Human performance in our task could be sub-optimal in several ways , each depending on the type of information the CNS maintains about Equation 1 ., We consider 5 such sub-optimal models , denoted M1 , … , M5 ., In the first three of these , subjects fail to take account of ασ , βσ , or both when planning reaches ., In model M1 , subjects fail to compensate for the experimentally imposed static increase in temporal uncertainty due to the added Gaussian noise ( SD\u200a=\u200a25 ms ) ; in M2 subjects fail to compensate for the linear increase in temporal uncertainty with increasing reach duration; and in M3 subjects fail in both respects ( for details , see Methods: Data Analysis and Model Comparison ) ., Models M4 and M5 were analogous to models M2 and M3 , respectively , but assumed the offset or slope were unknown and hence not fixed to match the training data or added 25 ms timing uncertainty ., We compare subjects performance to each of these sub-optimal movement strategies , and to the optimal strategy ( M0 ) that results in maximum expected gain ., During training trials , subjects attempted to produce reaches with an experimenter-specified temporal duration; no rewards or penalties were imposed ., In Figure 2A , we plot the mean movement duration as a function of the target duration for subject HT ., The points lie near the identity line , indicating that the subject could accurately produce a wide range of movement times on command ., Figure 2B shows the temporal uncertainty function ( the standard deviation of arrival times as a function of target duration , with and without the added noise ) measured during training for the same subject ., As expected , unperturbed standard deviations ( dot-dashed line , open symbols ) increase linearly across this range ., Estimated Weber-noise parameters ( ασ ) for all subjects temporal uncertainty functions , and verification of the stationarity of those functions ( across the training trials and the subsequent main experiment ) , are provided in Figure 3 ., Note that fitted functions obtained from training data ( lines ) and the standard deviations measured during main-experiment reaches ( filled diamonds ) were well-matched , consistent with the idea that subject performance did not change during the experimental reaches ., Each of the models makes predictions of reach durations that are based on the aspects of the temporal uncertainty function it incorporates ., Because the optimal model ( M0 ) incorporates both components of the temporal uncertainty function , it can take account of the temporal noise actually experienced by each subject when planning reaches , in turn allowing it to predict optimal movement times ., Three of the sub-optimal models ( M1–M3 ) each specify only a portion of the actual temporal noise experienced by subjects ., Because these models cannot account for the full temporal uncertainty function , their predicted ‘best’ movement times are sub-optimal ., For each subject and model , we derived predictions of the mean duration in each of the four conditions that would maximize expected gain in the task given that temporal uncertainty function ( see Methods: Model Predictions; Figure 4 illustrates these calculations for an example subject ) ., These predictions allow us to compare observed performance in the task to the theoretical performance of subjects who maximize expected gain under the constraints imposed by each of the four models ., In addition to these four models , we considered two sub-optimal models that did not have fixed parameters ( M4 and M5 ) ., In models of this type , the model likelihood ( see Method: Data Analysis and Model Comparison ) is calculated by integrating over the possible values of the unknown parameters ( e . g . , overall noise level ) ., The results of a Bayesian comparison of the performance of the four models ( see Methods: Data Analysis and Model Comparison ) favored the optimal model M0 over the sub-optimal models; yielding 11 . 5 dB in favor of M0 , but −60 . 5 dB , −11 . 5 dB and −41 . 4 dB of evidence for M1 , M2 and M3 , respectively ., Models M4 and M5 are less constrained , resulting in evidence below −100 dB ., Negative evidence is evidence against a model relative to the other possible models ., In our previous work 26 we have used 3 dB evidence , corresponding to odds of nearly 2∶1 , as a minimal guideline for inferring an advantage for a model over its competitors ., The 11 . 5 dB evidence for M0 is strong , corresponding to nearly 15∶1 odds in favor of the optimal model over the set of alternatives ., To assess inter-subject variability , we recomputed the evidence values for 5 subgroups of subjects , with each subgroup consisting of all subjects but one ., The change in evidence that occurred as we left each subject out is a measure of how much the conclusions we draw are based on one subject alone ., While the evidence decreases somewhat when each subject is removed ( and it should since we are basing our conclusion on fewer data ) , it always favored M0 , and always by at least 7 . 5 dB , consistent with the conclusions based on all subjects taken together ., We note , in particular , that removing the non-naive subject who was an author ( TEH ) still resulted in evidence of 9 dB in favor of M0 ., In addition , we plotted , for all subjects and conditions , the mean observed movement duration as a function of the duration predicted by each of the four models ( Figure 5 plots the deviations of the actual from the predicted movement times ) ., In such a plot , consistency of the data with the model corresponds to the data falling along the identity line ., We computed linear regressions of observed mean duration as a function of predicted mean duration for each of the four models ., Only M0 had a best-fit slope and intercept whose confidence intervals contained those of the identity line ( Table 1 ) , corroborating the result of the Bayesian model comparison ., We conclude that the evidence favoring M0 over any of the competing models is overwhelming , implying that subjects compensated for their increased uncertainty at longer durations and also for the 25 ms added uncertainty imposed experimentally ., To investigate how the suboptimal models fail , we present differences between observed average temporal endpoints and model predictions for each of the four models ( Figure 5 ) ., For each of the sub-optimal models , we describe how the pattern would appear if data were fit with that model ., Model M1 compensates for increased temporal uncertainty with increased movement duration but fails to compensate for the σ\u200a=\u200a25 ms temporal noise added experimentally ., Subjects conforming to this model will have temporal aim points closer to the center of the target region than they should be since they are based on an erroneously small estimate of temporal uncertainty ., That is , compared with the optimal model ( M0 ) , model M1 predicts longer durations for predictions of durations shorter than the target duration ( 650 ms ) , and shorter durations for predictions longer than the target duration ., Thus , we predict the left-hand cloud of residuals to move down and right and the right-hand cloud to move up and left , which is precisely what happened ( upper-right panel , Figure 5 ) ., Subjects employing model M2 ( lower-left panel , Figure 5 ) would fail to take duration-dependent noise into account , but compensate for the s\u200a=\u200a25 ms temporal noise added experimentally ., Such subjects overestimate noise for short durations and underestimate it for long durations ., Intuitively , the residuals should move up and left ., This is true of most data points , but not all ., The intuitive pattern is occasionally broken due to the complex , nonlinear calculation of expected gain ( Figure 1B ) and the switch from the veridical uncertainty function ( M0 ) to an incorrect , flat function ( M2 ) ., As expected , the predictions of M3 combine the shifts of the other two suboptimal models ., In summary , based on the comparison of the optimal and three suboptimal models , we conclude that subjects delayed or advanced their temporal endpoints in accordance with the calculated optimal times defined by M0 ., The Bayesian model comparison employed is novel and correct for comparison of non-nested models ( see Method: Data Analysis and Model Comparison ) ., We also carried out a set of statistical tests based on linear regression of actual versus predicted times ., The conclusions based on these regressions tests are identical to those just reported: we reject models M1 , M2 and M3 but not M0 ( Table 1 ) ., The gains earned by subjects potentially provide an additional dimension for testing the models ., We have compared actual gains to expected gains predicted by each of the models ., However , the gain functions are flat relative to the sampling variability of observed points earned , so that this analysis does not serve to differentiate the models ., To investigate the possibility that subjects used a hill-climbing strategy during the main experiment , instead of maximizing expected gain by taking account of their own temporal uncertainty function and experimentally imposed gain function , we performed a hill-climbing simulation using each subjects temporal uncertainty function ., In the simulation , intended duration was moved away from the penalty region by 3Δt ms after each penalty and towards the center of the target region by Δt ms for each miss of the target that occurred on the opposite side from the penalty ( corresponding to the 3∶1 ratio of penalty to reward ) ., The value of Δt was initially set to be relatively large ., With each change of direction of step , Δt was reduced by 25% to a minimum step size of 1 . 5 ms . While this simulation approximately reproduced the final average reach times observed experimentally , it does not provide a good model of subject performance ., First , there were significant autocorrelations of reach durations beyond lag zero in the simulation data but not in the experimental data ., Second , a learning algorithm would be expected to produce substantially higher σ values during test than those observed during training ., This is what we found with our hill-climbing simulation ., Using subjects training σ values to produce the simulated data , the simulation produced 17 out of 20 main-experiment σ values that were above the training values , whereas our subjects main-experiment σ values ( Figure 3 ) were entirely consistent with temporal uncertainty functions measured during training ., To move accurately , an organisms motor system must generate an intricate series of precisely timed neural commands ., The exact nature of these commands is not known ., Whatever the format of the command signals 27–32 , movement controlled by any physical controller-actuator system , including biological motor systems , will always exhibit some motor uncertainty ., Nevertheless , it is possible to plan movements that will maximize expected gain in the face of that uncertainty ., To do so , an organism must be capable of assessing both the probabilities of possible movement outcomes and their consequences ., One of the most thoroughly studied cases in which humans integrate the probabilities of possible movement outcomes and their consequences is the tradeoff between movement speed and spatial accuracy 20–21 , 33–34 ., However , in our experiment we were concerned with temporal accuracy , and faster movements are typically more temporally accurate ( the opposite of the spatial speed-accuracy tradeoff ) ., By imposing costs for early/late arrivals , we were able to determine whether the motor system is capable of picking movement times that maximize expected gain , taking into account temporal uncertainty ., We conclude that , in the timing task we examined , the motor system estimates and compensates almost perfectly for its own temporal uncertainty and correctly anticipates how that uncertainty interacts with the asymmetric reward structure of the environment ., This outcome is plausible given the close neurophysiological links between motor timing and the assessment of probabilities and consequences 22–25 , 35–37 ., We note however that it has been argued that a representation of time plays no role in one of the most basic forms of motor learning: motor adaptation 38 ., The current study provides evidence that the motor system is capable of using a representation of time in at least some circumstances where the consequences of the movement are unambiguously linked to the timing of the movement , and in addition that it does so optimally ., Several models of spatio-temporal movement control are based on optimizing an internal cost function that either includes or predicts movement timing ., One such model of trajectory formation , the minimum variance model 39 , assumes that the CNS selects a spatio-temporal reach trajectory by optimizing a cost function based on the movements endpoint variance ., In particular , the minimum variance model selects “…the temporal profile of the neural command … so as to minimize the final positional variance for a specified movement duration…” 39 , p . 782 ., More recently the minimum-time model of trajectory formation has been proposed 40 based on the assumption that , subject to a constraint on movement accuracy , the CNS attempts to minimize movement duration ., In both models , the speed-accuracy tradeoff is modeled by scaling the spatial variance of the reach with the amplitude of the motor control signal; that is , they assume signal-dependent spatial motor noise ., In the absence of signal-dependent noise , both models would predict a ‘bang-bang’ control scheme , where the control signal takes first a maximum positive and then maximum negative value producing alternating maximum forward and reverse accelerations leading to maximum movement speed and hence minimum duration ., However , bang-bang control predicts trajectories that are inconsistent with typical motor behavior ., By modeling spatial noise as signal-dependent , it is possible to predict a range of important behavioral results with both the minimum-variance and minimum-time models , such as the smooth variation in spatial and temporal reach profiles e . g . , 41–42 , Fitts law 33 , and the spatio-temporal details of saccadic trajectories 43 ., Unlike these previous studies , here the emphasis is on accuracy of movement duration ., This results in a reverse speed-accuracy tradeoff; slower movements have lower temporal accuracy ( even though they have higher spatial accuracy ) ., We show that , in a task where spatial uncertainty ( and therefore signal-dependent spatial noise ) plays essentially no role , reach durations are selected to nearly maximize expected gain in the presence of duration-dependent temporal uncertainty ., Duration-dependent temporal uncertainty constitutes a constraint on the temporal aspects of movement planning that is similar in many respects to the planning constraint imposed by signal-dependent spatial noise ., Simultaneously minimizing temporal and spatial noise provides a method of solving the underconstrained problem of trajectory selection ., Although several previous studies have proposed multiply-constrained models of movement planning 44–45 and the duration-dependence of temporal uncertainty is well known e . g . , 10; 46–47 , we provide the first demonstration of the CNS making use of its own temporal uncertainty in movement planning ., While selecting the movement trajectory that minimizes spatial and/or temporal noise is a possible method of movement planning , the optimal movement planner carefully separates the constraints imposed on spatial and temporal accuracy ( duration-dependent temporal noise and signal-dependent spatial noise ) with the costs of spatial and temporal errors , which we discuss next ., In both the minimum-time and minimum-variance models 39–40 , a trajectory is selected so as to optimize an internal cost for spatial variance or movement duration ( respectively ) in the presence of signal-dependent spatial noise ., The cost is internal in the sense that it does not make reference to any externally imposed costs on movement errors , such as monetary rewards and penalties that may be imposed due to ones spatial precision or movement duration ., There have been a large number of models of movement based on the optimization of internal cost functions that identify movement cost with an invariant kinematic or dynamic variable ( time 48 , spatial precision 39 , torque-change 49–50 , jerk 51 , etc . ) ., However , there are pitfalls inherent in identifying movement cost with an aspect of the movement itself , despite the current movement goals ., For example , the minimum-variance model always chooses a movement with the best possible spatial precision , even when that level of precision is unnecessary for the task ., Similarly , the minimum-time model always chooses the shortest duration movement that satisfies the constraint on spatial precision even when , as in some conditions of the current study , an external temporal cost function rewards longer-duration movements ., Recent models of optimal movement planning e . g . , 14 , 18 , 26 , 44 approach the problem somewhat differently ., In these models , which have previously been used to predict spatial movement endpoints 14 , 18 and movement trajectories 44 , the difference between a constraint on movement planning and a cost incurred from movement error must be recognized ., While duration-dependent temporal noise , signal-dependent spatial noise , energy consumption , biomechanics , etc . constitute constraints on movement planning and control , they are not properly costs ., A cost essentially imposes a weighting on the available constraints , and is task dependent ., By experimentally imposing costs 14–15 , 18–21 , 26 on spatial or temporal inaccuracy , it is possible to predict flexible movement strategies that incorporate task-relevant constraints ( e . g . , duration-dependent temporal uncertainty ) while effectively ignoring ( down-weighting ) constraints that are not as important to the task at hand ( signal-dependent spatial uncertainty ) ., In the present study , we manipulated the temporal cost function by imposing penalties on too-short reach durations in some conditions , and too-long durations in other conditions , and determined whether subjects responded appropriately to these different cost functions ., We have modeled movement planning as minimizing an external gain function in the presence of task-relevant internal temporal noise ., By identifying the to-be-minimized cost with the movement goal we have separated fixed kinematic/dynamic variables from the purpose of the movement ., This allows us to predict flexible movement plans that may minimize spatial or temporal uncertainty , but only when that is relevant to the task at hand ., A deeper understanding of movement planning and execution will result from models that similarly separate cost functions from fixed sets of kinematic/dynamic variables while simultaneously taking account of task-relevant spatial and/or temporal uncertainty ., Subjects were first given a training session in which temporal targets ( width: 3 ms , no adjacent penalty region ) were presented at six target durations ( 565 , 595 , 625 , 655 , 685 and 715 ms; 8 repetitions each , in separate blocks , followed by 50 repetitions each , in separate blocks ) spanning the range of temporal aim points observed during pilot work ., Although this window was too narrow for subjects to reliably hit , subjects were not scored during training , and were told simply to time their reaches as closely to each target time as possible ., This session allowed us to estimate the standard deviation of each subjects movement durations for a set of precisely known target durations , and also allowed subjects to learn their own ( noise-added ) temporal uncertainties in the task ., Standard deviations at each target time ( Figures 2B and 3 ) were measured from the final 40 repetitions to avoid possible initial practice effects ., Immediately following training , subjects were given a temporal target centered at 650 ms , with a half-width of 0 . 6σ650 , where σ650 was the estimated SD of movement duration for a mean duration of 650 ms . In this way , we equated the difficulty of the task across subjects based on their training performance ., Subjects were paid a bonus for touching the spatial target within the temporal target window ( Figure 1A , green , cross-hatched bars ) , and penalized for touching the spatial target within a temporal penalty window ( Figure 1A , red , striped bars ) or for failing to touch the spatial target ., Four blocked conditions were employed ( Figure 1A ) , two early temporal penalty conditions and two late penalty conditions ( 64 trials each ) ., The two early temporal penalty regions began at 0 ms and ended either 0 . 6σ650 or 1 . 35σ650 ms prior to 650 ms . The two late temporal penalty regions began either 0 . 6σ650 or 1 . 35σ650 ms following 650 ms , and were open-ended ., The outcome of each trial was signaled by distinct auditory tones notifying the subject that a reward was earned or a penalty assessed ., The possible reward earned on any trial was $0 . 12 and the penalty was −$0 . 36 ( or −$0 . 60 for missed spatial targets ) ., Note that the ratio of penalty to bonus magnitudes was 3∶1 ., Trials in which the spatial target was not touched were re-run ( fewer than 1% of all trials ) to equate the number of touched-target trials in each condition ., The untouched-target trials were not analyzed ., Subjects were four students at New York University who were not aware of the purpose of the experiment and one author ( TEH ) ., All subjects gave informed consent before the experiment ., The experimental protocol had been approved by the Institutional Review Board at New York University ., As described in the Introduction , decision theoretic models of motor behavior are concerned with the interplay of three elements: movement strategy , uncertainty , and the gain or loss from possible movement outcomes ., The interplay of these three elements is represented graphically in Figure 1B for the optimal model , M0 ., Calculation of the temporal endpoints predicted by each of the models to be considered required that the expected gain , in terms of average bonus earned per reach , be computed based on the constraints supplied by the hypothetical system ., For example , the optimal neuromotor controller would make use of information concerning both Weber-like increases in temporal uncertainty with increasing reach time , and the experimentally increased overall temporal uncertainty ., A given motor strategy or plan , s , determines the critical states of the system ., Although motor plans are complex sequences of control signals in time , the only consequence of the choice of motor plan in our task is to select an expected temporal endpoint , τs ., The expected gain from s is then given by ( Figure 1B ) : ( 2 ) where G, ( t ) describes the gain or loss associated with a particular temporal endpoint ( Figure 1A and Figure 1B , middle panel ) ., The term p ( t | τs ) describes the probability density of temporal endpoints expected from any chosen movement strategy s ., Note that these are planned durations , not reaction times , and hence we have no a priori expectation that these distributions will be skewed ., We model the duration distribution as a Gaussian with mean arrival time τs and a standard deviation σ ( τs ) ( 3 ) ( QQ plots of these distributions confirm that the Gaussian distribution models the data well ) ., The temporal uncertainty function , σ ( τs ) is able to capture the well-known Weber-like scaling of temporal standard deviation with mean arrival time τs ( Figure 1B , top panel ) ., We used values estimated from each subjects training data to compute individual σ ( τs ) functions for models M0–M3 ., In Figure 1B ( bottom panel ) , for the rightmost choice of τ , the probability of arrival in the penalty zone is nearly as high as that of arrival in the reward zone ., This choice of τ is likely to lead to nearly as many penalties as rewards ., Given that the penalty/reward ratio was 3∶1 , expected gain is negative for this choice of τ ., The distribution associated with the leftmost choice of τ is primarily in the uncolored time zone where the subject earns nothing ., This choice of τ is likely to lead to rare rewards and extremely rare penalties , resulting in only a small total reward across many trials ., Interestingly , a third choice of τ , centered on the temporal reward region , earns even less than the previous choice of τ because of a combination of its proximity to the temporal penalty , the magnitude of temporal movement noise , and the ratio of the reward to penalty magnitudes ., The best of the four choices shown is therefore the τ located at the left edge of the rewarded temporal region ., Of the four shown , it makes the best compromise between the width of the probability distribution for t and its distance from the centers of the reward and penalty regions , given the widths of those regions and the ratio of gains to losses ., Of course , there are infinitely many possible choices of τ ., The lower panel shows the expected gain as a function of τ , with the maximum expected gain ( MEG ) point indicated with a circle at the peak of the expected gain function ., If observers select this value τopt , they maximize their expected gain ., We computed τopt for each of the four penalty conditions and each subject based on an estimated temporal uncertainty function σ ( τs ) that was specific to each subject ., In all cases the optimal ( maximum expected gain ) value of τs was shifted away from the penalty region ., The optimal Bayesian model ( M0 ) makes full use of the temporal uncertainty function σ ( τs ) from each subjects training session ., The five sub-optimal models use less information ., M1 uses the σ ( τs ) calculated from each subjects training data without the experimentally added σ\u200a=\u200a25 ms noise ., M2 uses each subjects constant σ for all τs that includes the overall added σ\u200a=\u200a25 ms noise; it uses the square root of the average of perturbed variances about the target durations measured during training ., M3 uses the subjects constant σ without the experimentally added noise ., M4 and M5 use a constant offset and constant offset and slope , respectively , but assume that the values of these parameters are unknown ., Of course , some subjects are more accurate than others but this is explicitly taken account of in our analysis ., Each models predictions are defined in terms of performance relative to an individuals temporal uncertainty function ., Subjects who are inherently poorer timers are being compared to a standard ( defined by each model ) that is tailored to ( defined in terms of ) the limits of that subjects abilities ., So while there are in fact individual differences between subjects , these were removed in the design and analysis of the experiment ., Because we equated subjects in this way we could analyze group data ., The predicted movement strategy , s , is therefore a function of the type, ( s ) of temporal uncertainty information incorporated by each model Mm , the reward structure defined by the jth experimental condition ( j\u200a=\u200a1 to 4 ) , and the temporal uncertainties measured during training for the kth subject ( k\u200a=\u200a1 to 5 ) ., Let denote the value of τ predicted by model Mm based on an estimate of timing uncertainty calculated from the assumptions of each model ., For convenience , we denote the temporal uncertainty for an attempt to produce a movement duration of ( using the full temporal uncertainty function based on the training trials ) , , as ., The models we considered are not all nested and consequently we chose a method of model comparison for non-nested models 52–54 that we describe next ., Let denote the ith arrival time ( of the 64 trials per condition ) in condition j for the kth subject ., The likelihood of model Mm is given by: ( 4 ) where ., ( 5 ) Note however that for M4 and M5 , the model likelihood must be calculated by integrating over the unknown parameters: the constant offset , , and constant offset and slope , , of the temporal uncertainty function , respectively , where the prior probability distributions over the parameters are taken to be bounded Jeffreys ( uninformative ) priors 55 ., Let π ( Mm ) denote the prior probability of the mth model ., Then the posterior probability of the mth m

Introduction, Results, Discussion, Materials and Methods

Motor control requires the generation of a precise temporal sequence of control signals sent to the skeletal musculature ., We describe an experiment that , for good performance , requires human subjects to plan movements taking into account uncertainty in their movement duration and the increase in that uncertainty with increasing movement duration ., We do this by rewarding movements performed within a specified time window , and penalizing slower movements in some conditions and faster movements in others ., Our results indicate that subjects compensated for their natural duration-dependent temporal uncertainty as well as an overall increase in temporal uncertainty that was imposed experimentally ., Their compensation for temporal uncertainty , both the natural duration-dependent and imposed overall components , was nearly optimal in the sense of maximizing expected gain in the task ., The motor system is able to model its temporal uncertainty and compensate for that uncertainty so as to optimize the consequences of movement .

Many recent models of motor planning are based on the idea that the CNS plans movements to minimize “costs” intrinsic to motor performance ., A minimum variance model would predict that the motor system plans movements that minimize motor error ( as measured by the variance in movement ) subject to the constraint that the movement be completed within a specified time limit ., A complementary model would predict that the motor system minimizes movement time subject to the constraint that movement variance not exceed a certain fixed threshold ., But neither of these models is adequate to predict performance in everyday tasks that include external costs imposed by the environment where good performance requires that the motor system select a tradeoff between speed and accuracy ., In driving to the airport to catch a plane , for example , there are very real costs associated with driving too fast and also with being just a bit too late ., But the “optimal” tradeoff depends on road conditions and also on how important it is to catch the plane ., We examine motor performance in analogous experimental tasks where we impose arbitrary monetary costs on movements that are “late” or “early” and show that humans systematically trade off risk and reward so as to maximize their expected monetary gain .

neuroscience/behavioral neuroscience, neuroscience/sensory systems, neuroscience/motor systems, computational biology/computational neuroscience, neuroscience/theoretical neuroscience

journal.pbio.1000475

Multi-Platform Next-Generation Sequencing of the Domestic Turkey (Meleagris gallopavo): Genome Assembly and Analysis

The rapid and continuing development of next-generation sequencing ( NGS ) technologies has made it feasible to contemplate sequencing the genomes of hundreds—if not thousands—of species of agronomic , evolutionary , and ecological importance , as well as biomedical interest 1 ., Recently , a draft genome of the giant panda was described , based solely on Illumina short read sequences 2 ., Below , we describe the genome sequence of the turkey ( Meleagris gallopavo ) determined using primarily NGS platforms ., In this case , however , a combination of Roche 454 and Illumina GAII sequencing was employed ., While this approach presented unique assembly challenges , the turkey sequence benefits from the particular advantages of both platforms ., In addition , unlike the case for the panda , this novel approach allowed us to use a BAC contig-based physical and comparative map , along with the turkey genetic map 3 and the chicken genome sequence 4 , to align the turkey sequence contigs and scaffolds to most of the turkey chromosomes ., Such an alignment is essential for making long range evolutionary comparisons and employing the sequence to improve breeding practices using , for example , genome-based selection approaches , where chromosome locations are critical ., The high throughput and low cost of NGS technologies allowed sequencing the turkey genome at a fraction of the cost of other recently reported genomes of agricultural interest ( bovine and equine ) 5 , 6 ., The draft turkey genome sequence represents the second domestic avian genome to be sequenced , and this permits a genome-level comparison of the two most economically important poultry species ., When added to the recently published zebra finch genome 7 , analysis of the three avian genomes reveals new insights into the evolutionary relationships among avian species and their relationships to mammals ., Turkeys , like chickens , are members of the Phasianidae within the order Galliformes ., One estimate 8 is that the last common ancestor of turkeys and chickens lived about 40 million ( M ) years ago; however , other estimates are more recent 9 , 10 ., Comparison of the turkey genome to that of the chicken provides the opportunity for high resolution analysis of genome evolution within the Galliformes ., The turkey has 2n\u200a=\u200a80 chromosomes ( chicken has 2n\u200a=\u200a78 ) and , as for most avian species , the majority of these are small “microchromosomes” that cannot be distinguished by size alone ., Although most turkey chromosomes are syntenic to their chicken orthologues , the chicken chromosome GGA2 is orthologous to two turkey chromosomes , MGA3 ( GGA2q ) and MGA6 ( GGA2p ) , due to fission at or near the centromere , while GGA4 is orthologous to MGA4 ( GGA4q ) and MGA9 ( GGA4p ) 10 , 11 ., Generally , DNA from a single inbred animal is preferential for sequencing to minimize polymorphism ., For the turkey , however , such an option is not available , and thus we sequenced DNA from “Nici” ( Nicholas Inbred ) , a female turkey , which is also the source DNA for the two BAC libraries that have been characterized 12 ., Nici is from a subline ( sib-mating for nine generations ) originally derived from a commercially significant breeding line , but her genome is still extensively heterozygous ., A side benefit of this approach was the concomitant identification of extensive and commercially relevant single nucleotide polymorphism ( SNP ) data , as discussed below ., With the exception of the BAC end sequences ( BES ) used only for chromosome alignment , the sequence data used for this assembly came solely from two sequencing platforms: the Roche/454 GS-FLX Titanium platform ( 454 Life Sciences/Roche Diagnostics , Branford , CT ) and the Illumina Genome Analyzer II ( GAII; Illumina , Inc . , San Diego , CA ) ., The 454 data were generated using the latest “Titanium” protocol at Roche and the Virginia Bioinformatics Institute ( Virginia Tech ) and included both unpaired shotgun reads and paired-end reads produced from two libraries with estimated 3 kilobase pair ( Kbp ) and 20 Kbp fragment sizes ., The 454 runs yielded approximately 3 M read pairs from the 3 Kbp library ( average usable read length 180 bases ) , 1 M read pairs from the 20 Kbp library ( average length 195 bases ) , and 13 M shotgun reads ( average length 366 bases ) ., The Illumina sequencing data were generated at the USDA Beltsville Agricultural Research Center and the NIH National Institute on Aging from both single and paired-end read libraries with a 180 bp fragment size for the paired reads ., Details on the sequence data are presented in Table 1 ., These data represent approximate 5× genome coverage in 454 reads and 25× coverage in GAII reads , assuming a genome size similar to that of the chicken at 1 . 1 billion bases 4 ., In addition , BACs used to generate the 40 , 000 BES alignment markers by traditional Sanger sequencing spanned ∼6× clone coverage of the genome ., Since female DNA was used , coverage of the Z and W sex chromosomes was half that of autosomes; therefore the assembly of both these chromosomes was poor ., A modified version of the Celera Assembler 5 . 3 13 , 14 was used to produce the contigs and scaffolds in the assembly ( see Methods for details ) ., The initial assembly contained 931 Mbp of sequence in 27 , 007 scaffolds with N50 size of 1 . 5 Mbp ., The span of the scaffolds was 1 . 038 Gbp ., The scaffolds contained 145 , 663 contigs with N50 size of 12 . 6 Kbp ., The assembled scaffolds were then ordered and oriented on turkey chromosomes using a combination of two linkage maps and a comparative BAC contig physical map ., The first turkey linkage map 3 had 405 chicken and turkey microsatellite sequences that mapped to the assembled scaffolds ., The second linkage map , based on segregation of SNPs in a different population 15 , had 442 SNP markers mapped to the scaffolds ., The comparative chicken-turkey physical map 16 provided turkey chromosome positions for 30 , 922 BES found in scaffolds ., Comparison of scaffolds to the marker map resulted in splitting only 39 scaffolds due to inconsistencies between the assembled scaffolds and marker positions on the chromosomes ., A total of 28 , 261 scaffolds containing 917 Mb of sequence were assigned to chromosomes ( Table 2 ) ., Included in this number were 7 , 080 single-contig scaffolds that represented repetitive sequences but that could be linked to non-repetitive scaffolds ., The remaining 5 , 858 scaffolds were pooled to form ChrUn ( unassigned ) which contains 19 Mb of sequence in comparison to about 64 Mb on the current chicken chr_Un ., Analysis of the assembled contigs showed that 4 . 6% of the sequence was covered only by reads from a single sequencing platform , with 2 . 3% covered exclusively by each ., If the reads covered the genome uniformly , one would expect to have missed only 0 . 67% of the genome with Roche/454 and 0 . 0006% with Illumina ., The distribution of regions of exclusive coverage for both platforms ( Figure S1 ) shows there was a large number of short ( <20 bp ) gaps in coverage by Illumina sequencing , whereas the Roche/454 coverage gaps tended to be larger ., Mean sequencing gaps were 46 bases for Illumina reads and 72 for the Roche/454 coverage ., Coverage biases previously have been shown for both platforms 17 , but fortunately , the biases are relatively orthogonal ., Therefore , it is definitely beneficial to use data from both platforms in de novo assemblies ., The draft turkey assembly was compared to the chicken genome assembly ( 2 . 1 ) , which was sequenced and assembled using traditional Sanger sequencing 4 ., Table 3 illustrates that assembly of NGS sequence data , although feasible , does not produce contigs and scaffolds as large as those expected from an assembly based on Sanger sequencing ., However , the relatively low cost of NGS sequencing ( <$250 , 000 for the turkey ) makes such projects feasible for species with more focused interest groups and facilitates for resources to be directed toward genome analysis and interpretation as opposed to generating raw sequence data ., However , chromosome assemblies currently still require the integration of multiple data types including shotgun reads and contigs , genetic linkage maps , BAC maps and BES , and cytogenetic assignments ., The challenge was to develop databases and software to achieve this goal ., Integrity of the assembly was validated by mapping the assembled turkey scaffolds to 197 Kbp of finished BAC sequence containing part of the MHC B-locus , GenBank accession DQ993255 . 2 ., The average sequence similarity was over 99 . 5% and no inconsistencies in the 21 scaffolds that mapped to that region were observed ., The extent of the genome coverage could be estimated both from the total span of the assembled scaffolds and from portions of the chicken genome with syntenic matches to the turkey scaffolds ., Both methods produced consistent estimates of the size of the euchromatic portion of the turkey genome at about 1 . 05 Gbp ., With 936 Mbp of sequence in the final chromosomes , including ChrUn , the assembly encompasses an estimated 89% of the total sequence of the genome ., One of the striking observations in the chicken genome sequencing project was the difficulty obtaining sequences for specific regions , including the 10 smallest microchromosomes 4 ., For example , the chicken genome lacks sequence orthologous to human chromosome 19q ., Remarkably , these sequences appeared to be absent not only from the shotgun clone libraries used to generate the whole genome shotgun ( WGS ) reads but also from all available BAC libraries 18 ., Although these regions have high GC content , it is unclear why this region of the genome is resistant to cloning in E . coli ., In general , BAC coverage of microchromosomes is less than macrochromosomes in both chicken and turkey BAC libraries , although the HSA19q orthologues are an extreme example of a missing syntenic region ., Since the turkey genome was sequenced without any cloning step , the assembly was tested for representation of HSA19q orthologous sequence ., Presence of sequences was verified by performing a BLAT analysis of the complete HSA19q sequence against the turkey and chicken genomes ( Table S1 ) ., Surprisingly , regions orthologous to HSA19q were not represented at a higher frequency in the turkey assembly versus the chicken assembly ., As was observed in the chicken , regions orthologous to HSA19p and a small syntenic region from HSA19q are covered well in the turkey assembly ( MGA30 and 13 , respectively ) ., These results suggest that absence of HSA19q orthologous sequences is not due to the high GC content , in that Illumina sequences show a bias towards higher coverage of GC rich regions 19 , 20 ., The identification of a single BAC clone that hybridizes across the entire length of a single microchromosome in chicken 21 suggests that the occurrence of microchromosome-specific repeats might be a more likely explanation for the absence of these sequences using both traditional Sanger sequencing as well as NGS technologies ., Heterozygous alleles , including both SNPs and single nucleotide insertions and deletions ( indels ) , were detected by scanning the assembled contigs for positions where the underlying reads significantly disagreed with the consensus base 22 ., A previous study cataloging heterozygous alleles from assembled shotgun reads within an individual human genome used a similar approach , augmented with a set of quality criteria used to distinguish genuine biological variations from sequencing error 23 ., Following this approach , a set of quality criteria was developed and implemented within the assembly forensics toolkit 24 ., Two classes of SNVs were catalogued: ( 1 ) those with abundant evidence , called strong SNVs ( 601 , 490 SNVs ) , and ( 2 ) a more inclusive set called weak SNVs ( 920 , 126 SNVs total ) ., In the turkey genome , transitions were roughly 2 . 4× more common than transversions: 295 , 055:122 , 731 for strong SNVs and 466 , 629:200 , 743 for all SNVs ., Many single base indel positions were detected: 183 , 215 of 601 , 490 strong SNVs , and 249 , 512 out of all 920 , 126 SNVs ., A very small number of SNVs ( 489 strong , and 3 , 242 all ) were detected with more than two well-supported variants , suggestive of unfiltered sequencing errors or collapsed repeats ., The depth of coverage for strong SNVs ranged between 6 and 30 with mean and standard deviation of 15 . 3±5 . 3 , while the depth of coverage for all SNVs ranged between 4 and 5 , 319 with mean and standard deviation of 41 . 4±134 . 6 ., The very high coverage regions are highly likely to be due to collapsed near-identical repeats ., Annotation of the turkey genome sequence identified a total of 15 , 704 genes ( Table S2 ) of which 15 , 093 were distinct protein coding loci and 611 non-coding RNA genes ., In addition , multiple distinct proteins produced by alternative splicing were identified for some loci , giving a total of 16 , 217 distinct protein sequences ., Orthologs between turkey , chicken , and human proteins were defined using sequence homology , phylogenetic trees , and conservation of synteny ., All gene annotations are available from the Ensembl genome browser version 57 ( http://e57 . ensembl . org ) ., The draft turkey genome assembly was used to test the distribution of nucleotide diversity across the turkey genome by aligning SNPs covering ∼3 . 97% of the genome identified through resequencing a reduced representation library from commercial turkeys 15 ., Substantial deviations were observed between regions in the genome ., Chromosome Z showed the lowest nucleotide diversity , about half ( θ\u200a=\u200a0 . 000273 ) that of the autosomes , which is likely the result of a lower effective population size of this chromosome and lower recombination rate ( Figure S2 ) 25 ., The five largest chromosomes had similar nucleotide diversities as the microchromosomes ., Given the higher recombination rate on the microchromosomes , the ensuing higher mutation rate 26 , and lower susceptibility to hitchhiking effects , equal rates of nucleotide diversity between micro- and macrochromosomes may seem unexpected ., However , these findings are in line with observations in the chicken 27 and may be explained by higher gene density and higher purifying selection on the microchromosomes ., Within chromosomes , extended regions of low nucleotide diversity were detected , many of which coincided with centromeres ., Comparisons of gene family assignment statistics for the turkey and chicken genome assemblies are shown in Table S4 ., Although the draft turkey sequence has fewer genes than the current chicken genome build ( 2 . 1 ) , part of the difference may be due to cutoff values used by annotation groups resulting in variation in gene number ., Even with this caveat , more than half of the gene families show no change in copy number between them ( Table S5a–d ) ., Overall , most families exhibiting variation have general regulatory functions related to transcription , metabolism , cation transport , cell-cell signaling , and cell development or differentiation ( Figure 3 ) ., Distinct keratin families , encoding major structural proteins of chicken feathers , claws , and scales , have undergone uneven expansion or contraction with considerable variation in number among species ., More than half of the innovation families ( found in turkey but not chicken ) have unknown functions , are singletons , and were annotated by mapping to the zebra finch protein prediction ., Species-specific gene families in birds and mammals are summarized in Tables S6 , S7 ., Of these , 881 are specific to turkeys and chickens and 271 specific to birds ., The inference for bird-specific functions is of relatively high quality since the likelihood that a bird gene is not simultaneously found in all 13 non-bird species is low ., Most of the rapidly evolving gene families in birds have unknown functions ., Approximately 83% of the turkey/chicken-specific families and 71% of the bird-specific families have unknown functions ., For the remaining families , most have well-defined roles ( Table 5 ) ., Families related to egg formation ( such as avidin , ovocalyxin , and vitellogenin ) and scavenger receptors were identified as avian specific in the present and previous analyses 4 ., Examination of gene family sizes between the avian species and the platypus , an egg laying mammal , found two egg-related gene families egg envelop protein ( ENSFM00500000271806 ) and vitellogenin , an egg yolk precursor protein ( ENSFM00250000000813 ) to be conserved among the four egg-laying species ., Both of these gene families are absent from eutherians ., Other gene families specific to egg-laying species ( birds and platypus ) are mainly related to protein metabolism , cell-cell communication , and regulatory functions ., Several other proteins related to egg formation , such as avidin and ovocalyxin , are found in birds but not in platypus ., In contrast to unique gene families , only 70 families were completely absent in both the turkey and chicken ( 33 in all birds ) compared to the non-avian species ., These include the gene family associated with enamel formation ( ENSFM00250000008876 , an enamelin precursor related to teeth ) , which is completely lost in the three avian species ., Genes encoding the vomeronasal receptors and several casein related families are also completely absent in the avian species ., Several olfactory receptor families specific to mammals are either absent or dramatically reduced in birds ., Interestingly , the olfactory receptor 5U1 and 5BF1 gene families , reported to be dramatically expanded in chicken as compared to humans and flies 4 , is contracted in turkey ., Lineage events in the turkey , chicken , and zebra finch genomes reveal significantly higher synonymous substitution rates on microchromosomes than macrochromosomes ( Figure 4a ) , with a clear inverse relationship with chromosome size ., This suggests that genes on the microchromosomes are exposed to more germ-line mutations than those on other chromosomes 38 ., However non-synonymous mutation rates do not seem to vary so widely and when combined show the dN/dS ratio ( a measure of selection ) to increase with chromosome size ., These results are consistent with the prediction that the higher synonymous substitution rates of microchromosomes combined with the “Hill-Robertson” effect 39 of higher recombination rates on these smaller chromosomes increases purifying selection 40 on the microchromosomes ( Figure 4b and 4c ) ., Theory predicts natural selection to be more efficient in the fixation of beneficial mutations in mammalian X-linked genes than in autosomal genes , where hemizygous exposure of beneficial non-dominant mutations increases the rate of fixation ., This “fast-X effect” should be evident by an increased ratio of non-synonymous to synonymous substitutions ( dN/dS ) for sex-linked genes ., As shown in Figure 5 , there is solid confirmation of the predicted rapid evolution in the sex-linked genes based on turkey , chicken , and zebra finch genome-wide data ., These results confirm that evolution proceeds more quickly on the Z chromosome 41 , where hemizygous exposure of beneficial non-dominant mutations increases the rate of fixation ., Based on the analysis of differentially evolved genes , 428 and 257 genes were identified as being under accelerated evolution in the turkey and chicken lineages , respectively ., Most of the accelerated genes in the turkey lineage have gene ontology ( GO ) terms related to DNA packaging and regulation of transcription ( Figure 6a ) ., In contrast , a large proportion of the accelerated genes in the chicken lineage have GO terms related to negative regulation of cellular component organization and biogenesis , proteolysis , interphase , and cell cycle arrest ( Figure 6b ) ., The enrichment of KEGG pathways using DAVID supports the GO term analysis ( Table S8 ) ., These results suggest that genes with a role in transcriptional regulation are key in the evolution of the turkey , whereas genes involved in protein turnover and cell proliferation have been more important in the evolution of the chicken ., For genes classified as innate immune loci by InnateDB ( www . innatedb . ca ) , dN/dS ratios were calculated for each pair of species ( turkey-chicken , turkey-zebra finch , etc . ) and then compared with ratios for non-immune genes ., Innate immune genes showed lower dN/dS ratios than other genes in all species-pairs of mammals and birds , except between turkey and chicken where the values are essentially equal ( Figure 7 ) ., Using Wilcoxon rank sum test , it is obvious from the comparisons that the innate immune-related genes have been under more purifying selection than non-immune-related genes except between turkey and chicken ( Table S9 ) ., Evolution of genes of the innate immunity system is thought to be continuous and under balancing selection 42 ., However , purifying selection under the same conditions may be the dominant force acting on the vast majority of genes that function within the innate immune system 43 ., Although only innate immune genes are under purifying selection by functional constraints , they are also more constrained than other genes ., This relationship supports the view that the ancient innate immune system has a highly specialized function , critical for the recognition of pathogens and thus should be under purifying selection ., However , unlike other species , the dN/dS ratios for innate immune genes between turkey and chicken are similar to other genes ., Perhaps the adaptation of turkey and chicken to different ecological niches has exposed them to new pathogenic environments with potentially lethal pathogens having exerted selective pressures on their genomes ., This thesis would suggest that there was a period of accelerated evolution of the innate immunity system after the divergence of these species 30–40 M years ago ., The availability of the turkey genome for comparison to the chicken 4 and zebra finch 7 allows for interrogation of the immune gene repertoire ., In general , homologs for all the innate immune gene families were found ( Table 6 ) , with smaller gene families present in birds ., This finding is consistent with earlier comparisons of mammalian with the chicken genome 44 and provides greater evidence of an avian-wide phenomenon ., Examples include the chemokines , TNF superfamily , and pattern recognition receptors ., Inflammatory CCL chemokines , which occur in all avian and mammalian species , fall into two multigene families ( MIP and MCP; Figure S4 ) ., There are four MIP family members in the chicken and the zebra finch ( CCLi1–4 ) , yet only three family members in the turkey genome build ( CCLi2–4 ) ., For the MCP family , there are six ( CCLi5–10 ) , three ( CCLi5–7 ) , and five ( CCLi5–7 and 9–10 ) members in the chicken , zebra finch , and turkey genomes , respectively ., The chicken genome sequence lacks TNFSF-family members TNFSF1 and TNFSF3 44 ., Presence of these lymphotoxins controls lymph node formation in mammals 45; however , lymph nodes are absent in birds 46 ., Therefore , it was not surprising that these genes were not found in any of the three avian genomes ., In contrast , lack of TNFSF2 ( TNFA ) was unexpected , since it is found in many fish species 47 , and there are several reports of TNF-alpha-like activity in chickens 48 ., A sequence homology search in the three avian species only detected TNFSF15 , a close relative of TNFSF2 ., Loss of TNFSF1 , 2 , and 3 ( as well as TNFSF14 ) in the avian lineage could explain these observations ( Figure S5 ) ., Absence of specific genes from the three avian genomes further implies that particular genomic regions are intrinsically difficult to clone and/or sequence with the traditional Sanger and NGS methods ., Finally , clear differences between birds and mammals exist in the size of the pattern recognition receptor families ., For example , there are only six NODLR family members in each of the three avian species , in contrast to 22 and 32 in human and mouse , respectively ( Table 6 and Figure S6 ) ., These are cytoplasmic receptors that recognize a range of ligands that activate caspases , and elicit an inflammatory response ., A recent analysis revealed hundreds of NODLR genes in fish 49 with homologs of all mammalian genes ., It is therefore clear that NODLR genes were lost during the evolution of the avian genomes ., In contrast , while similar numbers of TLRs are found in birds and mammals , evolutionary histories of gene gain , loss , and conversion are complex ( Figure S7 ) 50–52 ., The avian TLR1A/B and TLR2A/B genes are orthologs of mammalian TLR1/6/10 and TLR2 , respectively ., All three birds have lost TLR8 and 9 but retained TLR7 ., The avian TLR21 is the ortholog of mouse TLR13 , which was lost in the human lineage , and TLR15 appears to be unique to the avian lineage ., Approximately 6 . 94% of the turkey genome consists of interspersed repeats , most of which belong to three groups of TEs , the CR1-type non-LTR retrotransposons , the LTR retrotransposons , and the mariner-type DNA transposons ( Table 7 and Dataset S1 ) ., The CR1 group of TEs is the most abundant , occupying 4 . 81% of the genome , which is likely an underestimate because a number of highly degenerate and low copy number CR1-type elements remain to be characterized ., Overall , the turkey and chicken genomes are very similar with respect to repeat content and the types of predominant TEs 4 , 53 with high sequence similarities between major TEs ., For example , CR1_B in turkey and chicken share ∼91% nucleotide identity over a 2 Kbp region , the Birddawg_I LTR retrotransposons share ∼89% identity over a 3 . 6 Kbp region , and the mariner transposon Galluhop shares ∼91% identity over the entire 1 . 2 Kbp of the full-length element ., Similar to the chicken , the Galluhop repeat in turkey is associated with a deletion derivative of ∼550 bp ., Repetitive sequences are among the fastest evolving sequences in the genome ., Therefore , the conservation of the repeat elements and sequences between the turkey and chicken is indicative of very stable genomes given a divergence time of 30–40 M years ., Genome projects enable the collection of large supermatrices of alignable nucleotide sequences for phylogenetic analysis ., Galliform phylogeny was re-examined by collecting sequences from the turkey and chicken genomes for 42 loci ., These sequences were assembled into the largest supermatrix available for the order , containing 83 galliform species representing 73 genera , with three anseriform outgroup species ., With several whole mitochondrial sequences , two genomes , and repeated use in multiple studies , 37 taxa were represented by 11 or more loci , and 12 taxa by more than 20 loci , providing data-rich anchor points that bridged locus sets throughout the tree ., For the turkey , the main finding was its close relationship with the Central American ocellated turkey Agriocharis ( Meleagris ) ocellata ( 94% bootstrap support ) and the relation to the grouses within the phasianids ( Figure S10 ) ., The turkey-grouse clade has been recovered in several 59–61 but not all previous multi-locus studies ., The average bootstrap support for the nodes was high and the topology reproduced many features of previous studies , with monophyly of the megapodes , cracids , numidids and odontophorids , and polyphyly of the Perdicinae and Phasianinae within the phasianids ., Grouping of an African bird ( Ptilopachus petrosus ) traditionally classified as a phasianid with the New World quails as recently observed 59 is supported , with the three loci independently reproducing this clustering ., The same was true when P . nahanii was used instead of P . petrosus ., Polyphyly of Francolinus was expected 62; however , the implied polyphyly of Lophura was not ., Increased throughput and decreased costs of NGS technologies facilitate cost- and time-effective sequencing of genomes ., The turkey genome sequence described herein represents the first eukaryotic genome completely sequenced and assembled de novo from data produced by a combination of two NGS platforms , Roche-454 and Illumina-GAII ., This genome project is a first where the majority of the production cost was invested in analysis and interpretation rather than generating sequence , and that the assembly is comparable in genome coverage to the predominantly Sanger-based sequences of the chicken and zebra finch ., The sequence assigned to the chromosomes covers approximately 93% of the turkey genome ., The quality of this sequence makes it a valuable resource for comparative genomics including identification of thousands of SNVs amenable to whole genome analyses ., The turkey sequence confirms and extends the previously known high synteny between the turkey and chicken genomes 3 ., These two avian species are remarkably similar with only 30 predicted rearrangements ( mainly small inversions ) distinguishing their genomes , despite last sharing a common ancestor about twice as long ago as the common ancestor of mice and rats or humans and gibbons ., Chromosome rearrangements that occurred show a trend towards more acrocentric chromosomes in the turkey than in the chicken ., The stability of galliform genomes is further confirmed by the overall conservation of gene sequences and repeat families ., At less than a third the size of mammalian genomes , a greater proportion of the turkey genome ( ∼10% ) is under selective constraint versus mammals where the fraction of conserved nucleotides is approximately 5% ., This also reflects the reduced percentage of the turkey genome comprised of interspersed repeats ( 7% ) ., Whereas genomes of close relatives allow for analysis of rapidly changing sequence , those of distant species help elucidate regions conserved during vertebrate evolution ., Gene families present only in birds provide a broad perspective on lineage-specific evolution ., For example , variation in gene content between birds and an egg-laying mammal ( platypus ) shows functions shared by egg-laying animals in general as well as those unique to egg-laying birds ., Likewise , genes specific to mammalian characteristics such as tooth formation have been lost in avian species ., Some gene families such as TLRs of the innate immune system show complex evolutionary histories of gene gain , loss , and gene conversion between mammalian and avian species ., The adaptive immune system is a relatively recent innovation peculiar to the vertebrates and provides a valuable framework for genome comparisons 63 ., Genes involved in the control and regulation of the immune response towards invading pathogens are subject to strong selective pressures: the so-called “arms race” between pathogen and host ., The result has been exceptional sequence divergence between the immune genes of vertebrate species , in particular those between birds and mammals 64 ., Additionally , many immune genes belong to gene families that have been subject to lineage specific expansions and contractions , facilitating the evolution of new functions to combat pathogenic challenges ., There are many fundamental differences between the immune systems of birds and mammals , including the major histocompatibility complex ( MHC ) structure 65 , absence of lymph nodes in birds 46 , and different mechanisms of somatic recombination in the generation of antibody diversity 66 ., From an evolutionary perspective , the turkey and chicken provide an interesting case for comparative study ., These two genomes have undergone intense artificial selection in recent decades for similar production traits , yet their differentially evolved genes included more functioning in transcriptional regulation in turkey , and more functioning in protein turnover and cell proliferation in chicken .,

Introduction, Results and Discussion, Methods

A synergistic combination of two next-generation sequencing platforms with a detailed comparative BAC physical contig map provided a cost-effective assembly of the genome sequence of the domestic turkey ( Meleagris gallopavo ) ., Heterozygosity of the sequenced source genome allowed discovery of more than 600 , 000 high quality single nucleotide variants ., Despite this heterozygosity , the current genome assembly ( ∼1 . 1 Gb ) includes 917 Mb of sequence assigned to specific turkey chromosomes ., Annotation identified nearly 16 , 000 genes , with 15 , 093 recognized as protein coding and 611 as non-coding RNA genes ., Comparative analysis of the turkey , chicken , and zebra finch genomes , and comparing avian to mammalian species , supports the characteristic stability of avian genomes and identifies genes unique to the avian lineage ., Clear differences are seen in number and variety of genes of the avian immune system where expansions and novel genes are less frequent than examples of gene loss ., The turkey genome sequence provides resources to further understand the evolution of vertebrate genomes and genetic variation underlying economically important quantitative traits in poultry ., This integrated approach may be a model for providing both gene and chromosome level assemblies of other species with agricultural , ecological , and evolutionary interest .

In contrast to the compact sequence of viruses and bacteria , determining the complete genome sequence of complex vertebrate genomes can be a daunting task ., With the advent of “next-generation” sequencing platforms , it is now possible to rapidly sequence and assemble a vertebrate genome , especially for species for which genomic resources—genetic maps and markers—are currently available ., We used a combination of two next-generation sequencing platforms , Roche 454 and Illumina GAII , and unique assembly tools to sequence the genome of the agriculturally important turkey , Meleagris gallopavo ., Our draft assembly comprises approximately 1 . 1 gigabases of which 917 megabytes are assigned to specific chromosomes ., Comparisons of the turkey genome sequence with those of the chicken , Gallus gallus , and the zebra finch , Taeniopygia guttata , provide insights into the evolution of the avian lineage ., This genome sequence will facilitate discovery of agriculturally important genetic variants .

genetics and genomics/comparative genomics, genetics and genomics/genetics of the immune system, computational biology/comparative sequence analysis, evolutionary biology/genomics, genetics and genomics/genome projects, computational biology/genomics, genetics and genomics/bioinformatics

The combined application of next-generation sequencing platforms has provided an economical approach to unlocking the potential of the turkey genome.

journal.pcbi.1004724

A Biophysical Model of CRISPR/Cas9 Activity for Rational Design of Genome Editing and Gene Regulation

The RNA-mediated Cas9 adaptive immunity system ( CRISPR type II ) has revolutionized genome engineering by enabling the precision cutting of DNA that can be customized to target any sequence 1 , 2 , 3 , 4 , 5 , 6 , while being functional in a broad range of prokaryotes and eukaryotes , including bacteria , yeast , flies , fish , plants , worms , monkeys , mice , rats , rabbits , frogs , and human cell lines 3 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 ., By forcing the host to repair these precision DNA cuts , the CRISPR/Cas9 system allows recombinant DNA to be inserted at desired genome locations , and therefore can be used for performing high-throughput gene knockouts , loss-of-function screening , artificial immunization , removal of latent genome-encoded viruses , and site-specific gene therapy applications 19 , 20 , 21 , 22 ., A nuclease-deficient version of Cas9 , called dCas9 , retains its RNA-guided DNA binding activity and has been used as a transcription factor to tightly control gene expression levels and rewire a hosts transcriptional regulatory network 23 ., Multiple dCas9-based repression and activation devices , including within layered genetic circuits , have been developed in bacteria , yeast , and mammalian cells; these genetic circuits can regulate a targeted promoters transcription rate by up to 1000-fold 5 , 24 , 25 , 26 , 27 ., In principle , the expression of multiple guide RNAs , working with dCas9 , enables the regulation of many promoters simultaneously , and provides an almost limitless source of programmable transcription factors ., Based on recent observations , the CRISPR/Cas9/dCas9 system is highly versatile , but has imperfect specificity and activity under a wide range of environmental and genotypic conditions 25 , 28 , 29 , motivating a study of its mechanisms and the development of a model to rationally design its guide RNAs 21 ., One major challenge has been binding to off-target DNA sites , resulting in off-target mis-cutting of genomic DNA by Cas9 or gene mis-regulation by dCas9 28 , 30 , 31 , 32 , 33 ., Several strategies have been shown to reduce Cas9 off-target behavior by manipulating its cleavage activity 33 , 34 , 35 , 36 , 37 , 38 , 39 , 40 ., For example , two guide RNAs expressed together with a partially nuclease-deficient Cas9 nickase have been used to make two single-strand cuts at adjacent locations , increasing the rate of on-target repair by homologous recombination 40 ., Further , fusing dCas9 to the FokI nuclease increased the specificity of its nuclease activity to a 20 bp recognition sequence 39 ., These strategies address off-target cutting , but not off-target binding and gene regulation ., A system-wide understanding of how guide RNAs work together with Cas9/dCas9 to control off- and on-targeting binding would enable the rational design of guide RNAs , and other controllable factors , to improve Cas9/dCas9 specificity and activity ., In particular , when engineering dCas9-based genetic circuits , it will be desirable to modulate dCas9s ability to regulate gene expression through the introduction of guide RNA mismatches 8 ., However , the quantitative relationship between guide RNA sequence and dCas9s binding affinity is currently unknown ., In this work , we develop a comprehensive , mechanistic model of CRISPR/Cas9 that predicts how experimental conditions and guide RNA sequences ( crRNAs ) control target site selection and cleavage activity ., To initially parameterize this model , we analyze the large amount of structural , biochemical , and next-generation sequencing data that has recently measured several aspects of CRISPR/Cas9s function with different crRNAs under varied experimental conditions 4 , 29 , 33 , 35 , 37 , 38 , 41 , 42 , 43 , 44 ., We formulate a single system-wide model that explains how these disparate observations can originate from the same CRISPR/Cas9 mechanism of function ., We also present quantitative criteria for designing guide RNA sequences with targeted binding and cleavage activities ., By accounting for several important factors beyond the guide RNA sequence , our design rules are a significant improvement over existing , and somewhat contradictory , sequence design rules whose outcomes have also depended on the selected experimental conditions 8 , 21 , 33 , 37 , 42 ., To develop this model , we employed statistical thermodynamics and the law of mass action to formulate a five-step mechanism that accounts for concentration-dependent , cell volume-dependent , host genome-dependent , and crRNA-dependent changes to Cas9 complex formation , diffusion , target specificity , and target activity ( Fig 1 ) ., Kinetic and thermodynamic constants were estimated by analyzing six studies of Cas9/dCas9 function ( Table 1 ) ., We validated this model using in vitro Cas9-dependent cleavage rate data ( Fig 2 ) , obtained by Sternberg et al . 38 , together with new data collected in this study , measuring in vivo dCas9-dependent transcriptional repression in synthetic genetic circuits within bacterial cells ( Fig 3 ) ., Further , to predict how a guide RNA controls target specificity , we used deep sequencing data 3 , 33 , 37 , 41 to compile a position-dependent , nearest neighbor binding model that accounts for canonical and non-canonical PAM recognition sites , R-loop formation , and mismatches with DNA target sites ( Fig 4 ) ., We then employ the model to predict the binding occupancies of dCas9 to the lambda phage genome , mirroring a recent experimental study utilizing DNA curtains , to illustrate the differing dynamics between on-target and off-target DNA sites ( Fig 5 ) ., Finally , we applied the model to predict the frequency and location of off-target cleavage sites in a medically relevant example , where Cas9 was used to remove latent HIV viral DNA segments from a human cell line 45 ( Fig 6 ) ., Finally , by performing a sensitivity analysis on the model , we show the optimal experimental conditions to maximize on-target ( d ) Cas9 activity and minimize ( d ) Cas9 off-target binding ( Fig 7 ) ., The activity of Cas9-mediated cleavage is dictated by a 5-step mechanism that includes the expression of Cas9 and crRNA , the formation of active Cas9:crRNA complex , a random intracellular walk to search for DNA target sites , the formation of a Cas9:crRNA:DNA complex ( an R-loop ) at DNA sites , and finally DNA site cleavage ( Fig 1 ) ., We developed a dynamical mechanistic model that incorporates all known biomolecular interactions and processes that control the rates of these steps ( Materials and Methods ) ., The mechanistic model accounts for how several factors control all the DNA sites cleavage rates , including changing Cas9 and crRNA expression levels , different crRNA protospacer ( guide ) sequences , different DNA site sequences , both canonical and non-canonical PAM recognition DNA site sequences , and the effects of DNA site supercoiling ., The model also explicitly accounts for the hosts specifications , including its genome sequence , genome length , cell size , and growth rate ., Moreover , the model allows for the expression of multiple crRNA guide strands , and it will determine how the competitive binding of crRNAs to Cas9 will also affect the DNA sites cleavage rates ., When expressing Cas9 , the model calculates the numbers of all free , bound , and cleaved DNA sites that contain a canonical or non-canonical PAM site , encoded within the host genome or on plasmids ., When expressing nuclease-deficient dCas9 , the model calculates the occupancy of stably bound dCas9:crRNA complexes to all DNA sites ., Overall , the formulated model contained eight unknown parameters quantifying the binding interactions between Cas9 and crRNA as well as the effects of DNA site supercoiling on Cas9 binding affinity ., In addition , the model also utilized a multi-parameter free energy model quantifying crRNA-DNA site interactions ., We first utilized the in vitro measurements obtained by Sternberg et . al . to determine the kinetic parameter values that quantify Cas9:crRNA complex formation , pre-cleavage dissociation , and Cas9-dependent cleavage 38 ., In this study , the binding locations and cleavage rates of Cas9 using a plasmid DNA substrate were measured to characterize the multi-step process by which Cas9 finds DNA targets , initiates R-loop formation , and cleaves DNA sites ., Here , we utilized the authors dynamic measurements of DNA site cleavage at different concentrations of Cas9 and crRNA , using either an on-target site on plasmid DNA ( Fig 2B in 38 ) or an on-target site on a double-stranded DNA fragment ( Extended Data Fig 5 in 38 ) ., We also analyzed Cas9’s protein structure and its motility to estimate that Cas9s characteristic length is λCas9≈150°A 43 , 46 and its diffusivity in a cytoplasmic-like buffer is 45 μm2/s 47 ., Therefore , we determined that Cas9 performs an isotropic random walk with a diffusive specific flow rate of 4 . 05 x10-10 1/sec ., In the presence of 25 nM plasmid DNA , these calculations indicate that a Cas9:crRNA complex collides with a DNA site 61 times per second ., We then determined the kinetic parameter values controlling Cas9:crRNA association ( kf ) , isomerization ( kI ) , pre-cleavage dissociation ( kd ) , and cleavage activity ( kC ) by calculating the rate of cleavage ( rC ) across a range of Cas9 and crRNA concentrations , mirroring the experimental conditions , and comparing to experimental cleavage measurements ( 56 experiments; R2 = 0 . 97; S1 Fig ) using 25 nM plasmid DNA 38 ., The model solution was evaluated for an initial 10 minute time period , followed by in silico addition of the DNA substrate and an additional 30 minute time period ., The best-fit kinetic parameter values were then determined through optimization to minimize the relative error between calculated and measured cleavage rates ( Materials and Methods ) ., Based on our analysis ( S2 Fig ) , we could uniquely parameterize kf , kI , and the ratio kc/kd ( Table 2 ) ., Surprisingly , the rate of cleavage was found to be less than the rate of pre-cleavage dissociation ( kc/kd << 1 ) , suggesting that ( d ) Cas9 must engage in multiple aborted rounds of binding and R-loop formation before successfully cleaving the DNA site ., Using the best-fit parameter values , the model was able to accurately capture the experimentally observed time-dependent cleavage rates while varying the Cas9 and crRNA concentrations ( Fig 2 ) ., The best-fit parameter values are reported in Table, 2 . As expected , when the Cas9 concentration is limiting , the calculated amount of cleaved DNA is almost equal to the Cas9 concentration because Cas9 does not turn-over ., However , when non-supercoiled , short ( 55 bp ) DNA fragments were used as template , Sternberg et . al . found that Cas9’s total cleavage activity dropped by 5-fold even though the apparent cleavage rate of DNA increased ( S3A Fig ) ., The authors hypothesized that the reduced cleavage activity originated from a batch of partially active Cas9 enzyme ., To test this possibility , we first reduced the concentration of Cas9 in silico to 20% of the reported concentration ., The model reproduced the measured amount of cleaved DNA after the 10 minute incubation period , however , the model-calculated rise to steady-state was slower than the experimentally observed rise ( S3A Fig ) ., Instead , if we also accounted for the much smaller number of DNA sites and the lack of negative supercoiling of the short DNA fragments , then the model correctly explains the experimentally observed fast rise time ( S3B and S3C Fig ) ., Specifically , there were 5482 total possible DNA sites ( N ) when plasmid DNA template ( 2741 bp ) was used in the in vitro measurements , compared to only 110 possible DNA sites when short DNA fragments were used ( 55 bp ) , resulting in about 50-fold higher rise time ., The difference in DNA site supercoiling partly counteracted this much higher model-calculated rise time by requiring an additional 0 . 43 kcal/mol energy for the Cas9:crRNA complex to successfully form an R-loop , lowering the model-calculated rise time to about 25-fold higher than when using the plasmid DNA as template , which is close to the experimental measurement ., When using dCas9 to implement genetic forms of computing , we anticipated the need to introduce several adjacent crRNA binding sites to differentially regulate gene expression ., However , according to the biophysics of R-loop formation , it was possible that the binding of a ( d ) Cas9:crRNA complex to one target DNA site could actually lower the affinity of ( d ) Cas9:crRNA complexes to adjacent DNA sites ., Specifically , when a dCas9:crRNA complex binds to a DNA site , the creation of an R-loop will negatively supercoil the site’s DNA , for example , by untwisting it ., Because DNA’s linking number is conserved , the negative supercoiling of one DNA site will increase the positive supercoiling of adjacent DNA sites ., According to model Eq 13 , a higher superhelical density will make it less likely for another dCas9:crRNA complex to bind to adjacent DNA sites by requiring a higher free energy input to stably form an R-loop 48 ., To investigate this effect , we constructed a three plasmid system that expresses dCas9 using a constitutive promoter , a single crRNA using an IPTG-inducible PTAC promoter , and a YFP reporter protein using a constitutive promoter containing a fully complementary ( on-target ) crRNA binding site ( Fig 3A ) ., Using dCas9:crRNA as a transcriptional repressor , we measured steady-state YFP expression levels as the transcription rate of the crRNA was steadily increased via IPTG induction ., We then introduced either one , two , four , or eight additional on-target crRNA binding sites at a distal location on the high-copy reporter plasmid , upstream of the YFP promoter , separated by a transcriptional terminator , and performed the same YFP fluorescence measurements ., These auxiliary on-target crRNA binding sites were separated by 60 to 80 bp of non-repetitive DNA ., The presence of the many additional crRNA binding sites in a non-regulatory location had the expected effect of sequestering dCas9:crRNA , resulting in lower amounts of transcriptional repression at YFP’s promoter and higher YFP expression levels ( Fig 3B ) ., In light of this data , we consider two distinct hypotheses relating the number of artificially added crRNA binding sites to the apparent increase in YFP expression level ., First , if dCas9-mediated R-loop formation has no effect on the superhelical density of surrounding crRNA binding sites , then we should expect that adding more crRNA binding sites will proportionally sequester more dCas9:CrRNA , resulting in greater YFP expression levels as more crRNA binding sites are added ., Second , if dCas9-mediated R-loop formation does increase the supercoiling of adjacent DNA sites , then we should expect that many additional crRNA binding sites will not proportionally sequester more dCas9:crRNA , resulting in a sub-linear increase in YFP expression as more sites are added ., To quantify the extent that R-loop formation increases the superhelical density of surrounding crRNA binding sites , we added a single parameter to our model ( Δσ ) ., When n copies of dCas9:crRNA are bound to nearby DNA sites , the initial superhelical density of the remaining nearby DNA sites is increased by nΔσ , which increases the sites’ ΔGsupercoiling according to Eq 13 , and lowers the probability that they will be bound by additional dCas9:crRNA ., If Δσ is zero , model calculations show that adding 8 crRNA binding sites to the plasmid will yield greater amounts of dCas9:crRNA sequestration , resulting in 300-fold more YFP expression ( Fig 3B , right ) ., However , if Δσ is positive , adding more crRNA binding sites will yield diminishing amounts of dCas9:crRNA sequestration and sub-linear increases in YFP expression ( Fig 3B , left ) ., Using this data-set to evaluate these two hypotheses , we found that adding 2 , 4 , or 8 additional crRNA binding sites increased dCas9:crRNA sequestration and YFP expression , but with lower-than-proportional amounts , suggesting that there is indeed a anti-cooperative mechanism affecting site occupancies ( Fig 3B ) ., We found that a moderate site-to-site superhelical density penalty ( Δσ = 0 . 0065 ) was sufficient to explain how adding more crRNA binding sites sublinearly increased dCas9:crRNA sequestration and YFP expression level ( Fig 3B , left ) with a high degree of confidence ( R2 = 0 . 97 , p < 10−8; S4 Fig ) ., The apparent site-to-site changes in superhelical density appear to be additive; for dCas9 to stably bind to the 8 binding site array , it would be necessary to untwist over 160 bp of the 900 bp region , equivalent to about 6 kcal/mol of free energy input , which would greatly destabilize R-loop formation and lower dCas9:crRNA occupancy ., To compare , a model that ignores changes in superhelical density , and its effect on dCas9:crRNA occupancy , was not able to explain the measurements ( Fig 3B , right ) ., Additionally , according to this data-set , it appears that crRNA concentration , and not dCas9 concentration , was limiting the total amount of dCas9:crRNA that could bind these additional crRNA bind sites or the promoter to repress YFP expression , discounting an alternative hypothesis ., Cas9 requires the presence of a protospacer adjacent motif ( PAM ) sequence to bind to a DNA site , form an R-loop , and cleave DNA ., While the consensus PAM sequence for the Cas9 from S . pyogenes is NGG , it was previously observed that R-loop formation could take place at non-canonical PAM sites , resulting in a considerable amount of off-target activity 3 , 35 , 42 ., To quantify Cas9’s binding free energy to DNA sites that use either canonical and non-canonical PAM sites , we utilized data from a recent study that measured Cas9’s cleavage activity when bound to DNA sites with identical PAM-proximal sequences , but randomized PAM sequences , using a homolog of Cas9 from S . pneumonia 3 ., We compared cleavage activities to a reference PAM site , which we defined by the four nucleotide sequence 5’-CGGT-3’ , with a corresponding reference free energy ( ΔGPAM , ref = -9 . 9 kcal/mol ) ., This reference free energy was consistent with our in vivo measurements shown in Fig, 3 . Importantly , we found that the first nucleotide ( N in NGG ) did not significantly contribute to Cas9’s cleavage activity , but that the fourth nucleotide did significantly alter cleavage activity ., We then employed model Eqs 6 and 15 to calculate the change in ΔGtarget , and therefore the change in ΔGPAM , corresponding to each four nucleotide PAM sequence ., As only the PAM sequences vary , the free energies ΔGexchange and ΔGsupercoiling were not expected to change significantly ., To eliminate background noise , we excluded any PAM sequence that resulted in less than 1% cleavage ., Further , we found that averaging cleavage activities over the first nucleotide position of each PAM sequence resulted in apparent free energies with a low coefficient of variation of 9% ., Overall , we quantified the apparent ΔGPAM free energies of 26 PAM sequences and found that they vary by 4 kcal/mol ( Table 3 ) , which is equivalent to about 700-fold change in instantaneous cleavage activity ( all other factors being equal ) ., As expected , the canonical PAM site NGGN binds with the highest affinity to Cas9 with ΔGPAM energies exceeding -9 kcal/mol ., However , there are several non-canonical PAM sites with sufficiently high affinities to contribute to off-target cleavage activity , including NAGN and NGWN ., Further , the presence of a gap between a fully complementary protospacer and a PAM site does not fully ablate Cas9’s binding affinity; a single nucleotide gap ( NNGG ) penalized binding by 2 . 2 kcal/mol , while a single nucleotide bulge ( GGNN ) had a larger effect ( a 3 kcal/mol penalty ) ., Recent studies have demonstrated that Cas9 can bind well to several non-canonical PAM site such as NAG , NGA , NAA , NTG , NGC , NCG , and NGT , though the extent of its promiscuity does depend on the Cas9 species origin 49 , 50 ., Using the ΔGPAM free energies in Table 3 and an estimate of the DNA site’s superhelical density , the model can now calculate the binding free energy ( ΔGtarget ) of Cas9:crRNA when the crRNA’s guide sequence perfectly matches the DNA site’s sequence ., To quantify the effects of mismatches , we next developed a free energy model ( ΔΔGexchange ) that accounts for changes in the crRNAs guide sequence ., A mismatch between the crRNA guide sequence and a DNA site destabilizes the formation of the Cas9:crRNA:DNA R-loop and increases the likelihood that the Cas9:crRNA complex dissociates prior to cleaving the DNA site 35 , 38 , 44 ., In our model , we quantify the thermodynamics of the R-loop strand displacement process , comparing the free energy of the initial double-stranded DNA state to the free energy of the Cas9:crRNA:RNA R-loop , resulting in a free energy change ( ΔΔGexchange ) ., ΔΔGexchange will change whenever a mismatch is introduced , though the magnitude of the change will depend on both the position of the mismatch and the surrounding sequence composition ., As the last step in developing our model , we utilized three next-generation sequencing datasets ( Table 1 ) to parameterize position- and sequence-dependent free energy models quantifying the Cas9:crRNA:DNA interactions during R-loop formation ., Three types of free energy models were created and compared to investigate whether Cas9 plays a role in mediating these interactions , and whether these interactions varied across different host genomes ., In the Pattanayak et al . , the on-target and off-target cleavage activities from four sgRNAs were measured via deep sequencing across a degenerate library of DNA sites within an in vitro reaction 33 ., In Hsu et . al . and Mali et . al . , respectively , the amounts and locations of Cas9-based cleavage and dCas9-based transcriptional activation were recorded in vivo via deep sequencing 33 , 37 , 41 ., We categorized these measurements into two data-sets , dataset I and II ( Table 1 ) ., To analyze these data-sets , we first identified all DNA sites that utilized a canonical PAM sequence similar to the PAM sequence adjacent to the targeted sequences and yielded greater than 50 read counts , finding 3671 sites in data-set I and 5979 sites in data-set II ., Further , the superhelical densities of DNA sites are the same within the in vitro data-set , and largely similar across the E . coli genome , enabling us to disregard changes in ΔΔGsupercoiling for this analysis ., We then compared sequencing read counts between Cas9 cleavage at the perfectly complementary ( on-target ) site and all off-target sites , obtaining a direct relationship between changes in sequencing read count and changes in ΔΔGexchange , according to our model Eqs 7 and 8 ., When analyzing dCas9-based transcriptional activation measurements , we assumed that the dCas9 binding probability was proportional to the transcription rate of the target promoters ., For each sequence , this rate was also proportional to the ratio of the background-subtracted read counts from the samples and the background-subtracted read counts from the positive controls ., We then utilized Eqs 7 and 8 to convert the normalized RNA-Seq read counts into changes in ΔΔGexchange 51 ., By excluding alternative PAM sites , we were able to more precisely quantify the energetic effects of introducing mismatches into DNA site sequences ., Comparing the Pattanayak et . al . and Hsu et . al . datasets , the overall average energetic penalty for a single mismatch was 0 . 14 and 0 . 78 kcal/mol , equivalent to a 1 . 26-fold and 3 . 7-fold drop in Cas9 activity , respectively , which suggests that the differences between in vivo and in vitro measurements and characterization protocol had an influence on off-target cleavage activities ., However , some single mismatches were found to penalize ΔGexchange by 4 kcal/mol , equivalent to a 785-fold drop in Cas9 activity ., Therefore , we next formulated position-dependent and sequence-dependent models to quantify how the introduction of mismatches in either the crRNA guide sequence or DNA site sequence affected Cas9 activity ., In the first free energy model , we employed Eq 9 to calculate ΔΔGexchange , which quantifies the thermodynamic stability of the RNA-DNA and DNA-DNA complexes responsible for R-loop formation , together with 21 unknown position-dependent coefficients ., While the free energies of DNA-DNA and DNA-RNA complementary duplexes have been measured 52 53 , there has been limited measurements of DNA-RNA mismatch free energies ., Using a dinucleotide nearest-neighbor model , there are 240 types of RNA-DNA mismatches; however , the free energies of only about 72 of them have been experimentally measured 54 , 55 , 56 , 57 , 58 ., After incorporating the known complementary and mismatch DNA-DNA and RNA-DNA free energies into Eq 9 , and utilizing either dataset I or dataset II to parameterize the position-dependent coefficients , the resulting model was not able to predict Cas9 binding or cleavage activity ( R2 = 0 . 32 and 0 . 07 for dataset I and dataset II , respectively; S4 Fig ) ., Consequently , we anticipate that additional measurements of RNA-DNA mismatch free energies and kinetic modeling will improve the development of accurate first principles models of R-loop formation ., We then developed an alternative free energy model ( Eq 10 ) that does not rely on previous thermodynamic measurements of nucleic acid interactions , but instead uses measured Cas9 activities at thousands of DNA sites to determine unknown model parameters ., The free energy model accounts for all possible guide RNA guide sequences and DNA site sequences , employing a dinucleotide nearest-neighbor model ( 256 unknown coefficients ) together with 21 position-dependent coefficients ., We determined the unknown parameters using either dataset I ( 3671 measurements ) or dataset II ( 5979 measurements ) , utilizing nonlinear least-squares to minimize the error between the apparent and calculated ΔΔGexchange free energies ( Materials and Methods ) ., This parameterization determined values for 86% and 80% of the unknown parameters , using dataset I and II , respectively ., In particular , these datasets lacked DNA sites with two consecutive mismatches , resulting in several unidentified parameters ., The resulting free energy models for ΔΔGexchange were qualitatively consistent with anecdotal observations; for example , the first eight position-dependent coefficients have the highest values , accounting for about 67% ( dataset II ) to 81% ( dataset I ) of ΔΔGexchange variation , quantifying the impact of PAM-proximal mismatches on Cas9 activity ( Fig 4A ) ., As a comparison , in a recent in vivo study 29 , 87% of sequences with high binding affinities to a Cas9:crRNA complex have at most 1 mismatch within the first 8 nucleotides ( Fig 4B ) ., The apparent mismatch free energies also varied up to 5 kcal/mol , suggesting that mismatch sequence composition is an additional factor that affects Cas9 activity ., However , the energetic penalties of specific mismatched RNA:DNA sequences were not necessarily the same across the two models ., When parameterized with in vitro Cas9 cleavage measurements ( dataset I ) , the most energetically unfavorable mismatches were found at dAG , dGG , and dCG dinucleotides that were positioned over rAC , rAG/rGA , or rGT/rTG dinucleotides ., In contrast , when parameterized with in vivo Cas9 activity measurements ( dataset II ) , the mismatch free energy penalties were more evenly distributed , potentially due to confounding interactions arising from the DNA sites chromatin states ., Overall , the empirically parameterized free energy models were able to sufficiently account for the sequence- and position-dependent effects on Cas9 activity across the thousands of DNA sites ( R2 = 0 . 74 and 0 . 61 for datasets I and II , respectively; Fig 4C ) ., However , the maximum uncertainty in a free energy parameter was 2 kcal/mol , indicating that there is significant opportunity for improving both the breadth and precision of Cas9 activity measurements with the objective of developing more accurate free energy models ., Next , we applied the parameterized mechanistic model to calculate dCas9 binding occupancies across the lambda bacteriophage genome when using a crRNA guide sequence that targets a specific genomic location , designated λ2 ., Our calculations mirror recently conducted experiments that monitored the dynamics of fluorescently labeled dCas9:crRNAλ2 as it interacted with an array of λ-phage genomic DNA within a flow chamber , called a DNA curtain 38 ., Using these calculations , we examine how the sequence composition and PAM density of a genome affects the partitioning of dCas9 and its binding dynamics ., Overall , the λ-phage genome contains 3179 and 2497 canonical PAM sites on its forward and reverse strands , respectively , together with 17933 and 16445 non-canonical PAM sites with a density of about one PAM site per 2 . 4 bp ., To calculate the dCas9 binding free energies at all PAM sites , we identified their corresponding ΔGPAM binding free energies ( Table, 3 ) and used both the λ2 guide and DNA site sequences to calculate the free energy change during R-loop formation ( ΔΔGexchange ) ., Here , we utilized the previously parameterized distance-dependent coefficients ( Fig, 4 ) and a DNA:RNA mismatch penalty of 0 . 78 kcal/mol , which was the overall average energetic penalty observed in the Hsu et . al . data-set ., We also assumed that all λ-phage genomic sites are equally supercoiled ( ΔΔGsupercoiling = 0 ) ., Model parameters are listed in S1 Table ., We found that the binding free energies of dCas9:crRNAλ2 varied by 25 kcal/mol across the 40054 PAM sites , and only 3880 of them had negative dCas9:crRNAλ2 binding free energies ( ΔGtarget < 0 ) ( Fig 5A ) ., Most PAM-proximal DNA sites had large numbers of mismatches with the crRNAλ2 guide sequence , causing ΔΔGexchange to be more positive than ΔGPAM ( Fig 5B and 5C ) ., In particular , there were only 25 DNA sites that had highly negative binding free energies ( ΔGtarget < -6 kcal/mol ) ., As expected , the λ2 DNA site formed a perfect DNA:RNA duplex with crRNAλ2 , resulting in a zero model-calculated ΔΔGexchange penalty and a ΔGtarget of -9 . 9 kcal/mol ., However , a second off-target DNA site , designated OS1 , had a canonical PAM ( GGGA , ΔGPAM = -9 . 4 kcal/mol ) , only two mismatches within the 8 most PAM-proximal nucleotides , and an additional six mismatches in the remaining 12 nucleotides , yielding a ΔGtarget of -6 . 3 kcal/mol ., Interestingly , fluorescently labeled dCas9 was observed to transiently bind to OS1’s position in the λ-genome 38 ., By enumerating and calculating the dCas9 binding free energies for all PAM sites , we can then calculate the system’s overall partition function to determine their binding occupancies under several scenarios ., The canonical partition function quantifies the amount of dCas9:crRNA that will be sequestered under equilibrium conditions ., It is also used in Eq 7 to determine the instantaneous binding probabilities to all DNA sites ., When using dCas9:crRNAλ2 , a fully accessible λ-genome has an overall partition function value of 162 . 6 ., The λ2 DNA site contributes the largest amount ( 151 . 04 ) , indicating that it has the largest probability of being bound first ., The off-target OS1 site contributes only 0 . 37 to the partition function summation , and therefore has a 408-fold lower probability of being bound first , compared to λ2 ., However , the additional 3879 off-target sites provide a significant contribution to the partition function summation , which will affect the binding occupancies at all PAM sites; sites with canonical PAMs contribute 7 . 83 , while those with non-canonical PAMs contribute 3 . 36 ., As a result , it is 30-fold more likely that dCas9:crRNAλ2 will initially bin

Introduction, Results, Discussion, Materials and Methods

The ability to precisely modify genomes and regulate specific genes will greatly accelerate several medical and engineering applications ., The CRISPR/Cas9 ( Type II ) system binds and cuts DNA using guide RNAs , though the variables that control its on-target and off-target activity remain poorly characterized ., Here , we develop and parameterize a system-wide biophysical model of Cas9-based genome editing and gene regulation to predict how changing guide RNA sequences , DNA superhelical densities , Cas9 and crRNA expression levels , organisms and growth conditions , and experimental conditions collectively control the dynamics of dCas9-based binding and Cas9-based cleavage at all DNA sites with both canonical and non-canonical PAMs ., We combine statistical thermodynamics and kinetics to model Cas9:crRNA complex formation , diffusion , site selection , reversible R-loop formation , and cleavage , using large amounts of structural , biochemical , expression , and next-generation sequencing data to determine kinetic parameters and develop free energy models ., Our results identify DNA supercoiling as a novel mechanism controlling Cas9 binding ., Using the model , we predict Cas9 off-target binding frequencies across the lambdaphage and human genomes , and explain why Cas9’s off-target activity can be so high ., With this improved understanding , we propose several rules for designing experiments for minimizing off-target activity ., We also discuss the implications for engineering dCas9-based genetic circuits .

The CRISPR/Cas9 immunity system has the potential to revolutionize medicine and biotechnology by enabling researchers to cut an organism’s genomic DNA at precise locations ., While Cas9 is perhaps the most versatile and easy-to-use technique for gene therapy developed yet , it is not perfect; the enzyme can also cut DNA at unwanted locations in an organism’s genome ., Cas9’s off-target activity must be greatly minimized to further improve its utility ., Here , we develop a system-wide , quantitative , physical model to better understand all the factors that collectively control Cas9’s off-target cleavage ., We solve for the unknown parameters using gene regulation data from our laboratory as well as structural , biochemical , and next-generation sequencing data from other laboratories ., Using the model in several examples , we explain how Cas9 identifies on-target versus off-target DNA sites , depending on the guide RNA sequence , the Cas9 and crRNA expression levels , the organism’s genome , and the organism’s cellular growth rate ., We then propose several rules for designing experiments with minimal off-target activity .

sequencing techniques, gene regulation, human genomics, dna transcription, luminescent proteins, yellow fluorescent protein, sequence motif analysis, dna, molecular biology techniques, thermodynamics, research and analysis methods, sequence analysis, proteins, gene expression, molecular biology, free energy, physics, biochemistry, nucleic acids, genetics, biology and life sciences, physical sciences, genomics, dna cleavage

journal.pbio.1002409

Hyperexpansion of RNA Bacteriophage Diversity

Bacteria play key roles in metabolic and immunological processes; however , at this time many of the factors that define the composition of a given microbial population are still unknown 1–4 ., Bacteriophages are abundant in many environments , and because they can lyse bacteria or transfer genes , bacteriophages likely play a role in shaping the specific composition of microbial populations ., The currently recognized bacteriophages employ highly diverse lifestyles , especially in regards to host range specificity and potential to induce cell lysis , and therefore , bacteriophages from different taxa may uniquely impact the microbial composition of a given niche 5 , 6 ., One particularly understudied area of bacteriophage diversity is that of RNA bacteriophages ., While many recent studies have aimed to characterize DNA bacteriophage communities in microbial populations , the RNA bacteriophage component of these communities is poorly defined 7–10 ., DNA bacteriophages are currently classified by the International Committee for the Taxonomy of Viruses ( ICTV ) into eight separate families with a total of 494 species , 55 single-stranded DNA ( ssDNA ) , and 439 double-stranded DNA ( dsDNA ) bacteriophage species ., These species derive from a diverse group of host bacteria; additionally , there are over 1 , 000 genomic sequences of DNA bacteriophage species in GenBank ., By contrast , according to the latest ( 2014 ) report of the ICTV , only two official families of RNA bacteriophages are recognized: the single-stranded RNA ( ssRNA ) bacteriophage family Leviviridae that includes four recognized species ( Enterobacteria phage Qβ , Enterobacteria phage F1 , Enterobacteria phage MS2 , and Enterobacteria phage GA ) and the segmented , double-stranded RNA ( dsRNA ) family Cystoviridae that contains a single recognized species ( Pseudomonas phage ϕ6 ) 11 , 12 ., There are complete sequences of 11 ssRNA and five dsRNA bacteriophages in the GenBank “Genomes” database as of 20 October 2015 , inclusive of the five ICTV-recognized RNA bacteriophage species ., In contrast to the DNA bacteriophages , in which bacteriophages have been characterized from a variety of bacterial phyla , all 16 of these bacteriophages are thought to infect hosts within the phylum Proteobacteria , with 15 that infect hosts within the class γ-proteobacteria ., In addition , three highly divergent , sewage-derived ssRNA bacteriophage genomes , with unknown host tropisms , were recently deposited in Genbank 13 ., For the analyses in this paper , we will refer to these 14 ssRNA bacteriophage sequences and five dsRNA bacteriophages sequences as the “reference RNA bacteriophages . ”, For some of these RNA bacteriophages , there are additional partial and/or full genomic sequences of closely related variants ( share > 66% nucleotide identity to the reference sequences ) also available in Genbank ., Bacteriophage identification has historically relied on culture-based methods 14–18 ., Given that the majority of bacterial species cannot be cultured in the laboratory , alternative culture-independent methods are necessary to describe bacteriophage diversity 19 ., In recent years , metagenomic sequencing has been applied to define bacteriophage populations in the human gut 20–25 , skin 26 , serum 27 , and in the environment 7 ., Additionally , computational mining of metagenomic datasets has been valuable for identifying additional novel taxa of DNA bacteriophages 28 , 29 ., However , the vast majority of these studies focused on sequencing and analysis of DNA only and therefore could not evaluate known or novel RNA bacteriophages that may be present ., Of the studies that did examine RNA viruses in the environment , only one recent metagenomic study of sewage reported the presence of two novel RNA bacteriophages related to leviviruses 13 ., Here , by mining multiple metagenomic datasets that were generated such that RNA could be evaluated , we identify partial genomes of over 120 highly diverse RNA bacteriophage phylotypes that are highly divergent from each other and all of the known RNA bacteriophage genomes ., This expansive diversity enabled us to identify new dimensions of RNA bacteriophage biology , including bacteriophages with novel genome organizations , numerous open reading frames ( ORFs ) that contain novel genes with no detectable homology to known bacteriophage genes , presence in novel ecological niches , and the first data in support of a RNA bacteriophage infection of a gram-positive bacterium ., We additionally assess the prevalence of two novel RNA bacteriophages in a cohort of macaques , presenting the first description of the ecological dynamics of these novel RNA bacteriophages ., Our results critically illuminate an unexamined dimension of molecular and ecological bacteriophage diversity and fundamentally establish a necessary framework that enables a more accurate dissection of RNA bacteriophage modulation of microbial populations ., To detect RNA bacteriophages , we initially queried multiple metagenomic nucleotide sequence datasets with protein sequences from the known leviviruses and cystoviruses ., We focused on datasets generated by our laboratory that contained cDNA sequences derived from RNA in the original material and that represented ecological niches known to support DNA bacteriophages , such as the vertebrate gastrointestinal tract and sewage ., Cystovirus protein queries yielded no significant alignments ( e-value < 10−4 ) ., In contrast , multiple nucleotide sequences in datasets from stool-associated and sewage specimens were identified that , following translation to amino acid sequences , aligned to leviviral proteins ., The four studies of relevance , which were previously generated by our laboratory , included a study of raw sewage 30 , two distinct studies of simian immunodeficiency virus ( SIV ) infection in nonhuman primates 31 , 32 , and a study of astrovirus infection in mice ., Any single dataset that had at least ten sequence reads that yielded significant alignments ( e-value < 10−4 ) was selected for assembly ., Using a National Center for Biotechnology Information ( NCBI ) conserved-domain search ( e-value < 10−3 ) or Phyre2 ( confidence > 90% ) , partial genomes of RNA bacteriophage phylotypes were defined as any assembled sequence greater than 750 nucleotides in length that contained a translated frame with a recognizable RNA bacteriophage-specific domain , such as a bacteriophage-specific RNA-dependent RNA polymerase ( RdRp ) , capsid , maturation protein , or packaging nucleoside triphosphatase ( NTPase ) 33 , 34 ., In order to focus our analyses on truly unique RNA bacteriophage phylotypes , any partial genomes that shared > 70% nucleotide identity in either the RdRp or the maturation gene were defined as belonging to a single phylotype ., The longest partial genome for a given phylotype was selected as the representative sequence for that phylotype in all downstream analyses ., By these criteria , partial genomes of 20 unique RNA bacteriophage phylotypes were identified in 17 distinct specimens ., Five partial genomes were assembled from metagenomic data from sewage specimens , 14 were from rhesus macaque stool data , and one was from mouse stool data ., Additional partial genomes that shared 85%–97% nucleotide identity to these 20 unique partial genomes were also identified in multiple other specimens in these studies , but they did not represent novel phylotypes by our criteria and therefore were not analyzed further ., Based on the sequence diversity of each of these assembled partial genomes within a single phylotype , we believe it is unlikely that these RNA bacteriophages originate from laboratory contamination ., The 20 unique bacteriophage phylotypes were sequentially named based on whether it was identified from an environmental or animal specimen , followed by a two-letter descriptor of the ecological niche ., To confirm the partial genome assemblies , the eight longest partial genomes ( range 3 . 5–5 . 0 kb ) out of the 20 identified were experimentally validated by generating multiple overlapping reverse transcription PCR ( RT-PCR ) amplicons followed by Sanger sequencing ( S1 Table ) ., The average length of the amplicons was ~1 . 8 kb; primers used to generate these amplicons are available in S3 Table ., In addition , the 3ʹ ends of AVE000 , AVE001 , and AVE003 were extended using rapid amplification of cDNA ends ( RACE ) ., To expand our search space , we analyzed publicly deposited sequencing datasets—generated by other laboratories—that sequenced RNA ( >10 , 000 Sequence Read Archive SRA datasets associated with >2 , 000 publications ) ., These included transcriptomic and RNA-inclusive metagenomic studies ., The metagenomic data analyzed were derived from environmental sources , such as oceans , sewage , and soil , and animal-associated sources , including stool ., We aligned amino acid sequences from the 20 novel and 19 reference RNA bacteriophages against sequences in these datasets , following six-frame translation , using tBLASTn ., Out of 2 , 765 RNA-inclusive metagenomes and 7 , 309 transcriptomic datasets examined , 115 contained at least ten sequences with significant alignments ( e-value < 10−4 ) ., The complete sequencing data from each of these 115 datasets were assembled , and RNA bacteriophage partial genomes were defined as above ( length > 750 nt , <70% identity to any other partial or complete genome ) ., We identified 138 unique partial genomes that contained ssRNA bacteriophage domains and five unique partial genomes that contained characteristic dsRNA bacteriophage motifs ( S1 and S2 Tables; S1 Dataset ) ., Thus , including the initial identification of the 20 novel ssRNA bacteriophages , we identified a total of 158 unique ssRNA bacteriophage motif-containing partial genomes ., For the partial genomes that contained ssRNA bacteriophage-associated domains , 119 contained RdRp domains , and 81 contained maturation domains ( 42 contained both maturation and RdRp domains ) ., Three partial genomes contained dsRNA bacteriophage RdRp domains ., As RNA viruses are not known to encode multiple RdRp genes , we conservatively estimated the number of novel RNA bacteriophage phylotypes based on the number of partial genomes that contain unique RdRp domains ., Based on this criterion , we have identified at least 122 novel RNA bacteriophage phylotypes , greatly increasing the known RNA bacteriophage diversity ., Furthermore , it is possible that some of the partial genomes that contained only maturation domains may derive from additional novel RNA bacteriophages , so this is likely an underestimation ., To elucidate the evolutionary relationships between the novel and known RNA bacteriophages , we next performed phylogenetic analysis ., Of the 119 novel ssRNA RdRp-domain-containing partial genomes , we limited the analysis to the 71 partial genomes that encompassed all five conserved motifs of the RdRp palm domain 35 ., In addition , we included the 14 “reference ssRNA bacteriophages . ”, We included an outgroup containing the RdRp palm domains of the two type species of the family Narnaviridae as their polymerases are most closely related to those of leviviruses 36 ., While bootstrap support for some portions of the tree is limited , it nonetheless demonstrated that the partial genomes were highly divergent from each other and from the known RNA bacteriophages ( Fig 1A ) ., For the dsRNA bacteriophage-domain-containing partial genomes , two unique partial genomes contained the entire RdRp gene , and both were clearly distinct ( Fig 1B ) ., There are currently no official ICTV criteria for defining species among RNA bacteriophages ., For many other viral taxa , strictly molecular criteria are used 37 , 38 ., For example , marine picornaviruses have previously been identified by sequence alignment and classified based on phylogenetic distance 38 ., Taxonomy of DNA bacteriophages has traditionally relied on bacteriophage morphology but now is largely determined using sequence-based criteria , in that phages that share a certain percentage of genes are considered the same species 39 , 40 ., One possible classification strategy for RNA bacteriophage would be to infer sequence-based criteria based on the current ICTV-recognized species ., Distinct ssRNA bacteriophage species within ICTV-recognized genera ( levivirus and allolevivirus ) share <60% amino acid identity ( 51% and 55% amino acid identity in the RdRp , respectively ) ., By extension , if membership in a species is defined as sharing ≥60% amino acid identity in the RdRp , the 158 ssRNA RNA bacteriophage phylotypes would represent 111 novel RNA bacteriophage species ., Even using a 50% amino acid identity threshold ( which would collapse currently recognized distinct species into one species ) would still result in 53 novel ssRNA bacteriophage species ( Table 1 ) ., Regardless of the final criteria used for classification by the ICTV , the RNA bacteriophage phylotypes identified in this study dramatically expand the known sequence diversity of RNA bacteriophages ., The family Cystoviridae has a single genus , the only ICTV-recognized species of which is Pseudomonas phage ϕ6 12 ., The four additional fully sequenced dsRNA bacteriophage species in Genbank , which are not officially classified in the genus Cystovirus , encode polymerase proteins that share 20%–51% amino acid identity to that of Pseudomonas phage ϕ6 ., A species defining criterion of 50% amino acid identity would classify the three novel dsRNA bacteriophage phylotypes as three species ., A threshold of 40% amino acid identity , which would collapse four distinctly recognized species into two , would result in two novel dsRNA species ., Following gene prediction and annotation of all the novel RNA bacteriophages , multiple novel genome organizations were identified ., Two of the novel RNA bacteriophage partial genomes , both of which were confirmed by RT-PCR and Sanger sequencing , were much longer than all sequenced leviviruses , which range from 3 . 73–4 . 27 kb in length 41 ., The genome of AVE000 had at least a 4 . 95 kb genome; this longer genome can be attributed to the presence of a novel >1 . 20 kb ORF of unknown function that is 5ʹ to and partially overlaps the maturation protein by 259 nucleotides ( Fig 2A ) ., AVE001 also has an expanded genome of at least 5 . 02 kb , due to the presence of a strikingly large 2 . 39 kb ORF containing the maturation domain , which is larger than all of the reference ssRNA bacteriophages maturation genes , which on average are 1 . 27 kb and range from 1 . 17–1 . 60 kb ., In addition , AVE002 is the first RNA bacteriophage described to contain two nonoverlapping ORFs between the RdRp and maturation genes; neither of the two ORFs has discernable similarity to known proteins ., While one of these ORFs likely represents the coat protein , the other ORF might represent a novel lysin or have homologous function to the Qβ read-through protein ., From the 119 ssRNA partial genomes , there were 100 ORFs predicted exclusive of the RdRp and maturation genes ., Aside from eight ORFs that had predicted leviviral coat domains and one that had a MS2 lysin domain , none of the other 91 ORFs had primary sequence alignment to any known bacteriophage coat or lysin protein ., These ORFs may encode proteins that are coat or lysin orthologs that are unrecognizable because of the greater evolutionary divergence of those genes as compared to the RdRp or maturation protein , or they could have completely novel functionalities ., Even in the former case , the extreme evolutionary divergence may result in novel host tropisms or novel mechanisms of bacterial lysis ., Further elaboration of these bacteriophage genomes will likely identify additional novel genome organizations and additional novel ORFs of unknown function ., The cystovirus protein sequence queries of public sequencing datasets identified partial genomes that provided evidence of an RNA bacteriophage with a novel genomic organization ., All five reference cystoviruses are trisegmented , with a ~6 . 4 kb L segment encoding both the RdRp and packaging NTPase and a ~2 . 9 kb S segment separately encoding a peptidoglycan degradation enzyme ., We identified EMS013 , a single 11 . 2 kb assembled sequence from a metatranscriptomic sample , originally isolated from the Zodletone sulfur spring , containing three individual ORFs that were annotated with these three functions ( Fig 2B ) ., Two additional samples from this sulfur spring in the same study also contained sequences that aligned to this partial genome ., An assembled sequence that contains both a cystoviral L and S genes is notable as there is ongoing debate as to the evolutionary origin of cystoviruses 42 ., One model suggests that cystoviruses share a common eukaryote-infecting ancestor with segmented eukaryotic RNA viruses ., The competing model suggests that cystoviruses originate from an unsegmented bacteria-infecting dsRNA bacteriophage ., While this partial genome is based solely on in silico assembly , a provocative hypothesis is that this bacteriophage could represent evidence of an unsegmented cystovirus ancestor ., As the majority of the novel RNA bacteriophage partial genomes were detected in metagenomic datasets derived from complex microbial communities , the host bacterium of each bacteriophage could not be explicitly determined by our analyses in most cases ., One notable exception was the detection of three dsRNA motif-containing assembled sequences in a publically available bacterial transcriptomic study derived from pure culture of Streptomyces avermitilis 43 ., This bacteriophage had the traditional genome organization of cystoviruses , although many ORFs could not be definitively annotated by either sequence or structural alignment ( Fig 2C ) ., This was named Streptomyces bacteriophage ϕ0 , keeping with nomenclature conventions of other cystoviruses ., The study was composed of two conditions with three replicates each , and five out of six samples contained sequences from the RNA bacteriophage ., The presence of sequences in each of the specimens in this experiment , combined with the annotation of the study as being derived from bacterial monoculture , strongly suggests that S . avermitlis represents the true host for this bacteriophage ., As the known RNA bacteriophages are only believed to infect proteobacteria , Streptomyces bacteriophage ϕ0 , if experimentally confirmed to infect S . avermitilis , would represent the first RNA bacteriophage known to infect bacteria in a phylum other than the proteobacteria ., Moreover , it would be the first RNA bacteriophage known to infect a gram-positive bacteria , thereby dramatically broadening the known bacterial host range of RNA bacteriophages ., While many bacteriophages were found from ecologies known to harbor RNA bacteriophages , namely mammalian stool and sewage , we identified numerous bacteriophages from novel ecological niches ( S1 Table ) ., Interestingly , we identified numerous bacteriophages that originated from microbial communities of invertebrate hosts , including pools of insects and aquatic invertebrates such as crabs , sponges , and barnacles ., Additionally , bacteriophages were identified from microbial sediments associated with extreme aquatic environments , such as sulfur springs and benthic cold seeps ., In order to evaluate spatial and temporal trends associated with novel RNA bacteriophages , we defined the prevalence of RNA bacteriophages in stool from previously described cohorts 32 of rhesus macaques collected at the Tulane National Primate Research Center ( TNPRC ) and the New England Primate Research Center ( NEPRC ) using both metagenomic and RT-PCR-based approaches ., Specimens from NEPRC were available at two separate time points , 24 and 64 wk post SIV infection ., Based on metagenomic sequencing data , 34 out of 120 specimens had at least one sequence from one novel RNA bacteriophage , with ten specimens containing at least one sequence from two or more novel RNA bacteriophages ( Fig 3 ) ., To independently assess the prevalence of a subset of these RNA bacteriophages , we screened this cohort using two sets of PCR primers: one set for AVE000 and another for AVE001 ., We detected AVE000 in five out of 120 rhesus macaques and detected AVE001 in 13 specimens ., The AVE000-positive amplicon sequences shared 95%–99% nucleotide identity with each other , while AVE001-positive amplicons shared 84%–100% nucleotide identity with each other ., Phylogenetic analysis demonstrated that there was apparent geographic segregation of both AVE000 and AVE001 , as sequences from each primate research center formed distinct clusters ( Fig 4 , Tables 2 and 3 ) ., All specimens positive by metagenomic sequencing for AVE000 or AVE001 were confirmed to be RT-PCR positive ., In addition , four specimens that were negative by metagenomic sequencing were RT-PCR positive , most likely because of the increased sensitivity of the RT-PCR assay compared to metagenomic sequencing ., The geographic clustering of the amplicon sequences combined with their observed diversity strongly argues against the possibility of laboratory contamination , as all of these specimens were prepared using the same protocol and reagents ., From the RT-PCR analysis , we found only two macaques were positive for the same RNA bacteriophage at two separate time points , suggesting that AVE000 and AVE001 generally do not persist ( Fig 4 ) ., Similarly , from the metagenomic analysis , the vast majority of the RNA bacteriophages were only present at a single time point ., This acute presence of these RNA bacteriophages is in stark contrast to the persistent nature in the primate gut of lytic DNA bacteriophages , specifically the Microviridae bacteriophages 23 ., In this study , we have vastly increased the number of known RNA bacteriophage phylotypes and demonstrated their presence in a wide range of habitats worldwide ., However , this is clearly an underestimate of the total number of RNA bacteriophage species , as there are undoubtedly many more novel RNA bacteriophages that remain undiscovered ., While our work has clearly identified a much greater diversity of RNA bacteriophages , there are still obvious taxonomic groups that are missing ., For example , RNA bacteriophages that contain negative-sense genomes or helical capsids have still not been identified , both of which underscore some of the many remaining gaps left in our understanding of RNA bacteriophage diversity ., Some of these novel RNA bacteriophages may be present in existing metagenomic datasets that simply cannot be currently recognized because of a lack of primary sequence alignment ., Furthermore , the vast majority of metagenomic studies are still heavily DNA-centric ., With the increased recognition of the importance of RNA bacteriophages and RNA eukaryotic viruses , the number of RNA-inclusive metagenomic datasets will surely grow , leading to additional discoveries of novel RNA bacteriophages ., Critically , the multitude of RNA bacteriophages already identified in this study provide a unique opportunity to define their natural contributions to ecology , explore novel aspects of their life cycle , and potentially exploit them as novel tools for bacteriophage therapy ., Mouse experiments yielding metagenomic data were performed with the approval of the Washington University IACUC , under protocol #20140244 .

Introduction, Results and Discussion, Methods

Bacteriophage modulation of microbial populations impacts critical processes in ocean , soil , and animal ecosystems ., However , the role of bacteriophages with RNA genomes ( RNA bacteriophages ) in these processes is poorly understood , in part because of the limited number of known RNA bacteriophage species ., Here , we identify partial genome sequences of 122 RNA bacteriophage phylotypes that are highly divergent from each other and from previously described RNA bacteriophages ., These novel RNA bacteriophage sequences were present in samples collected from a range of ecological niches worldwide , including invertebrates and extreme microbial sediment , demonstrating that they are more widely distributed than previously recognized ., Genomic analyses of these novel bacteriophages yielded multiple novel genome organizations ., Furthermore , one RNA bacteriophage was detected in the transcriptome of a pure culture of Streptomyces avermitilis , suggesting for the first time that the known tropism of RNA bacteriophages may include gram-positive bacteria ., Finally , reverse transcription PCR ( RT-PCR ) -based screening for two specific RNA bacteriophages in stool samples from a longitudinal cohort of macaques suggested that they are generally acutely present rather than persistent .

Bacteriophages ( viruses that infect bacteria ) can alter biological processes in numerous ecosystems ., While there are numerous studies describing the role of bacteriophages with DNA genomes in these processes , the role of bacteriophages with RNA genomes ( RNA bacteriophages ) is poorly understood ., This gap in knowledge is in part because of the limited diversity of known RNA bacteriophages ., Here , we begin to address the question by identifying 122 novel RNA bacteriophage partial genome sequences present in metagenomic datasets that are highly divergent from each other and previously described RNA bacteriophages ., Additionally , many of these sequences contained novel properties , including novel genes , segmentation , and host range , expanding the frontiers of RNA bacteriophage genomics , evolution , and tropism ., These novel RNA bacteriophage sequences were globally distributed from numerous ecological niches , including animal-associated and environmental habitats ., These findings will facilitate our understanding of the role of the RNA bacteriophage in microbial communities ., Furthermore , there are likely many more unrecognized RNA bacteriophages that remain to be discovered .

sequencing techniques, microbiology, metagenomics, sequence motif analysis, molecular biology techniques, rna sequencing, rna alignment, bacteria, microbial genomics, research and analysis methods, sequence analysis, marine bacteria, viral genomics, rna structure, sequence alignment, molecular biology, biochemistry, rna, nucleic acids, virology, genetics, biology and life sciences, genomics, organisms, macromolecular structure analysis

This study uses computational metagenomics and molecular experimentation to massively expand the known genomic and ecological diversity of RNA bacteriophages, identifying novel tropisms and genes.

journal.pgen.1001303

Genome-Wide Transcript Profiling of Endosperm without Paternal Contribution Identifies Parent-of-Origin–Dependent Regulation of AGAMOUS-LIKE36

Seed development is a tightly regulated process that is controlled , both before and after fertilization and requires tight coordination of parental gene expression 1 ., A paradigm for the importance of balanced parental contribution is the observation that certain genes in the developing offspring of flowering plants are exclusively or preferentially expressed from only one of the two parental genomes , a phenomenon called genomic imprinting that has also been observed in mammals 2 , 3 ., The relevance of parent-of-origin effects was first found in interploidy crosses 4 ., Typically , an increase in the paternal genome results in larger seeds , while the opposite is observed if the maternal gene dosage is higher than normal 5 ., This is in agreement with the parental conflict theory , which implies that fathers direct maximal amount of maternal resources to their own offspring and thereby promote growth ., Mothers on the other hand would seek to distribute the resources equally among all their offspring , and balance their resource between themselves and their offspring ., Thus , maternal factors are thought to dampen growth 6 ., In mammals , imprinted genes are often involved in growth control 7–10 ., In Arabidopsis , the endosperm is the major tissue regulating the flow of nutrients to the embryo , and is therefore a likely site for parent-of-origin dependent gene expression ., Imprinting results from differences in epigenetic marks , involving DNA methylation and post-translational modifications of histones on the parental alleles 11 , 12 ., Trimethylation of lysine 27 on histone H3 ( H3K27me3 ) leading to repression of gene expression , has been found to be a particularly important imprinting mechanism in plants ., In Arabidopsis seeds , H3K27me3 mark is set by the FIS Polycomb Repressive Complex 2 ( PRC2 ) , which consists of at least four components; the histone methyltransferase MEDEA ( MEA ) , FERTILIZATION INDEPENDENT SEED 2 ( FIS2 ) , FERTILIZATION INDEPENDENT ENDOSPERM ( FIE ) , and MULTICOPY SUPPRESSOR OF IRA 1 ( MSI1 ) ., The corresponding genes were identified in screens for autonomous endosperm development , indicating that the FIS complex acts as a repressor of endosperm development prior to fertilization 13–17 ., An equally important regulatory mechanism in imprinting is DNA methylation resulting from the activity of several different methyltransferase enzymes , where each has specificity for cytosine ( C ) in certain sequence contexts ., So far , imprinting has been shown to be under the influence of MET1 , the major Arabidopsis maintenance DNA methyltransferase involved in CG-methylation 11 , 18–20 ., DNA demethylation can be achieved either by a passive process i . e . the repression of MET1 expression 21 , 22 , or by an active mechanism involving DNA glycosylase enzymes such as DME 23 ., Several lines of evidence show that DME , which is expressed in the central cell of the female gametophyte , is necessary for maternal-specific gene expression in the endosperm 11 , 18 , 19 , 24 ., So far , only about a dozen genes in Arabidopsis have been identified to have parental-specific gene expression , and they illustrate different modes of imprinting 3 ., MEA , ARABIDOPSIS FORMIN HOMOLOGUE 5 ( AtFH5 ) and PHERES 1 ( PHE1 ) are imprinted by the action of FIS PRC2 , where only the latter is paternally expressed 13 , 25–31 ., FIS2 , FLOWERING WAGENINGEN ( FWA ) and MATERNALLY EXPRESSED PAB C-TERMINAL ( MPC ) are all maternally expressed and regulated by the dual action of MET1 and DME 11 , 19 , 24 , 32–34 ., Recently , five novel imprinted genes , HOMEODOMAIN GLABROUS 3 ( HDG3 ) , HDG8 , HDG9 , At5g62110 and ATMYB3R2 were identified by differential DNA methylation in embryo and endosperm 35 ., In comparison to Arabidopsis , more than 100 genes have been shown to have a uniparental or preferential parental expression pattern in mammals 36–39 ., This suggests that additional genes in Arabidopsis are imprinted ., Furthermore , the low number of known imprinted genes in plants precludes the identification of general principles in this kind of gene expression control and thus , the identification of further imprinted genes is pivotal ., Moreover , the targets of imprinted genes , as well as genomic pathways and regulatory modules influenced by imprinted genes are largely unknown ., Here , we have designed a microarray strategy for the identification of seed regulators by exploiting the cdka;1 mutation ., Using this approach , we have identified a cluster of previously uncharacterized AGAMOUS-LIKE ( AGL ) Type-I MADS-box transcription factors that are downregulated in endosperm with no paternal contribution ., Here , we report that AGL36 is imprinted by the dual action of MET1 and DME ., In addition , AGL36 is regulated throughout endosperm development in its maternal expression cycle by the Polycomb FIS-complex , thereby identifying a novel mode of regulation for imprinted genes ., Here we have used cdka;1 as a tool to identify factors sensitive to the vital parental gene balance in the endosperm ., In heterozygous cdka;1 mutants , the second pollen mitosis is either missing or is severely delayed ., However , mutant pollen can successfully fertilize the egg cell while leaving the central cell unfertilized 40 , 41 ., A detailed analysis by Aw and colleagues has revealed that a second sperm cell is delivered to the central cell , but that karyogamy does not take place 42 ., Although not properly fertilized , the majority of the central cells in cdka;1 fertilized ovules ( 70–90% ) are triggered to initiate endosperm proliferation 40 , 42 , 43 ., Thus , fertilization by cdka;1 sperm cells creates a unique situation where endosperm initially develops without any paternal contribution ( in the following also referred to as cdka;1P ) ., The endosperm , however remains under-developed , and ultimately the seed aborts , further demonstrating the importance of the paternal contribution to the endosperm for proper seed development ., Since activation of maternal alleles by loss of maternal FIS PRC2 could rescue seed lethality 43 , we hypothesized that the disturbance of parental gene balance in the endosperm is the main cause leading to developmental arrest of cdka;1P at 3–4 days after pollination ( DAP ) ., To identify factors and mechanisms sensitive to such an imbalance in gene dosage in the endosperm and with that likely key regulators of seed development , we performed microarray transcript profiling of cdka;1 fertilized seeds at 3 DAP ( Figure S1A ) ., Due to the heterozygous nature of the cdka;1 mutant line used , a transcript that is absent in cdka;1p seeds will lead to a reduction of maximal 50% in the genome profiling experiment ., For example , genes that are only expressed from the paternal genome would show such reduced expression levels ( Figure S1B ) ., Likewise , maternally expressed genes that require activation by a paternally expressed gene ( s ) would be downregulated ( Figure S1C ) , whereas genes that are acted upon by paternally expressed repressors were expected to be upregulated in the microarray screen ( Figure S1D ) ., When we compared the transcriptional profiles of Ler x cdka;1 versus Ler x Col seeds 3 DAP , we detected 17223 nuclear genes that were expressed in all biological replicates of both mutant ( cdka;1 set ) and wild-type ( WT set ) seed profiles ., Our result is in good agreement with a set of genes identified by Goldberg & Harada laboratories ( GH ) in globular stage seeds of Arabidopsis Ws-0 plants as 68% of our genes were also identified by GH , and our gene set included >90% of the GH globular seed gene set ( Figure 1A; http://seedgenenetwork . net , 44 ) ., To further validate the quality of our dataset , we examined the expression pattern of genes known to be preferentially expressed from the paternal allele ., To date , only three genes have been identified that show a predominant paternal expression pattern; PHE1 , HDG3 and At5g62110 , where all three genes were found to be downregulated in our arrays ( Figure S1E ) , supporting our working hypothesis that paternally expressed genes can be detected amongst downregulated genes ., In addition , out of seven imprinted maternally expressed genes present in our microarray sets , four were also detected as downregulated ( Figure S1E ) ., This could reflect required activation by paternal factors ( Figure S1C ) , or be a result of more complex deregulation in response to change in gene dosage ., To exclude array artifacts we tested all down-regulated genes by means of real-time PCR and could confirm their deregulation ( Figure 1B ) ., Due to the background noise in the microarray experiment , modest but reproducible downregulation of arithmetic ratios ( ar ) ranging from 0 . 5 to 1 . 0 will produce False Discovery Rates ( FDR , see materials and methods ) with insignificant q values ., Since the absence of paternally expressed genes was the simplest hypothesis to account for downregulation , we defined a functional limit for screening purposes that allowed us to detect two out of three known paternally expressed genes in the array ., Both PHE1 and HDG3 are detected at q values of 0 . 35 and a downregulation cutoff of 0 . 8 ( ar ) ., Consequently these values were chosen and used to filter the microarray data ., Using these criteria , a set of 602 genes was extracted ( q≤0 . 35 and ar ≤0 . 8 ) , subsequently called Down 0 . 8 ., For upregulation , we worked with two gene sets ., For the first set , Up 1 . 2 , we used parameters equivalent to the downregulated set ( q≤0 . 35 and ar ≥1 . 2 ) , which resulted in a set of 1030 genes ., For the second set , Up 1 . 5 , resulting in 323 genes , we chose ar ≥1 . 5 , a threshold for deregulation commonly used in genome-wide expression studies ( Table S3 ) ., To test whether the deregulated genes could preferentially be attributed to a certain seed structure , we compared our data to gene sets expressed in different seed regions and compartments of globular stage seeds using data generated by Goldberg & Harada ( GH ) laboratories available at http://seedgenenetwork . net 44 ., The overlap between the upregulated gene sets and the GH embryo , seed coat and endosperm was significantly lower than expected for independent sets of genes , indicating that among the upregulated genes we preferentially find those that are below the detection limit of the GH analyses ., However looking at the downregulated genes , the picture was different ., While we found slightly less overlap than expected by chance for the GH embryo set , the overlap was clearly larger than expected by chance for GH seed-coat ( 1 . 2<2 . 7e−07 ) and even more significant for the GH endosperm ( rf =\u200a1 . 3 , p<2 . 0e−13 , Figure S2A , S2B ) ., In order to functionally classify the deregulated gene sets according to their molecular functions we used the GO Slim classification system ( Figure 1C ) ., Only for the GO Slim term “Transcription factor activity” we find a higher percentage and significant over-representation of both up- and down-regulated groups when compared to all genes on the array/all genes expressed ., Since key regulators of seed development are likely to be transcription factors ( TF ) , we analyzed this class in detail ., When comparing the fraction of deregulated genes among the different TF families , the Mγ MADS-box transcription factors clearly stood out with more than 60% of the seed expressed members being downregulated in Ler x cdka;1 arrays ( Figure S3A , S3B ) ., We therefore focused on this MADS Type-I class for further analysis ., Searches in publically available expression databases ( www . genevestigator . com , Figure S4 ) revealed that all identified genes were exclusively expressed in the seed and predominantly in the endosperm ., From the identified Type-I Mγ MADS-box genes , we selected AGL36 for further in depth analysis ( Figure S4 ) ., AGL36 was the previously undescribed Mγ candidate that interacted with the highest number of described AGLs in a Y2H screen performed by de Folter et al 45 ., Both AGL36 and PHE1 have been shown to interact with AGL62 , which plays a major role in endosperm development 45 , 46 ., Within the Mγ class , AGL36 clusters together with AGL34 and AGL90 47 , which are both also detected as downregulated in our microarray experiment ( Figure S4 ) ., AGL36 shares 85 . 7% and 84% nucleotide identity with AGL34 and AGL90 , respectively ( Figure S8 ) ., On the amino acid level this results in of 80 . 2% similarity of AGL36 with AGL34 and 83 . 9% similarity with AGL90 ., Real-time PCR measurement of AGL36 relative expression level three days after pollination ( 3 DAP ) in Ler ovules fertilized with either Col or cdka;1 pollen confirmed that AGL36 expression was reduced in cdka;1 fertilized seeds , ( 27% when normalized towards ACT11 , and 36% when normalized towards GAPA ) compared to wild-type seeds ( Figure 2A ) ., To determine whether AGL36 has parental-specific expression , we took advantage of an AGL36 Single Nucleotide Polymorphism ( SNP ) existing between the Col and Ler ecotypes ., This SNP allows the PCR product of Col cDNA to be digested by AlwNI , leaving the Ler cDNA PCR product intact ( Figure 2B ) ., We performed reciprocal crosses between Col and Ler ecotypes , and analyzed the digested RT-PCR fragments on an Agilent Bioanalyzer Lab-on-a-Chip , allowing accurate measurement of fragment sizes and their concentrations ., When Colmaternal is crossed with Lerpaternal , we only detected the Col bands ( 165 bp+234 bp ) after AlwNI digestion , indicating only maternal expression ( Figure 2C ) ., Similarly , in the reciprocal cross when Lermaternal is fertilized with Colpaternal pollen , the cDNA PCR digest resulted only in an undigested band ( 399 bp ) originating from Ler , indicative of maternal expression ( Figure 2C ) ., This testified that AGL36 was only expressed from the maternal genome after fertilization and thus identified as a novel imprinted gene ., AGL36 expression level in wild-type seeds ( Ler x Col ) at different stages of seed development was monitored over a period of 12 days after pollination ., Initially , a low expression level was detected ( 1 DAP ) , followed by a rapid increase and subsequent peak in AGL36 expression at 4 DAP , when the embryo is at the late globular stage of development , before declining ( Figure 3A ) ., At the embryo heart stage , corresponding to 6 DAP , AGL36 expression had decreased to similar levels as 1 DAP ., To address whether AGL36 imprinting is maintained throughout its expression cycle , we performed a SNP analysis of the RT-PCR product obtained from Ler x Col crosses harvested during 1 to 12 DAP ( Figure 3B ) ., We found that AGL36 expression is originating from the maternal genome ( Ler ) throughout the experiment ., By plotting the molarities of the maternal band obtained by Agilent Bioanalyzer , an expression profile closely identical to the pattern obtained in the real-time PCR analysis was found ( Figure 3C ) ., To rule out that the observed maternal expression is due to expression of AGL36 in the ovule integument , which is a maternal tissue , we generated a reporter construct consisting of 1752 bp of the AGL36 promoter region fused to a GUS reporter ( pAGL36::GUS ) ( Figure 4A ) ., Single-copy lines carrying this construct were used in reciprocal crosses with wild-type Ler and Col plants to examine GUS expression at 3 and 6 DAP ., When inherited maternally , pAGL36::GUS expression in the seed was indeed found to be restricted only to the fertilization product ( Figure 4B , Figure S7D ) ., In the reciprocal cross , when pAGL36::GUS was inherited from the paternal genome , no GUS expression was detected , ( Figure 4C , Figure S7E ) ., Consistent with the SNP analysis , this demonstrated that AGL36 was imprinted and only maternally active throughout its expression cycle ., Furthermore , the 1 . 7 Kb promoter fragment used in this analysis appears to be sufficient to confer parent-of-origin specific expression of the reporter ., To further investigate the biological function of AGL36 , we screened the Koncz T-DNA collection for insertions 48 ., We identified a mutant line , agl36-1 , harboring a single T-DNA insertion 16 bp upstream of the AGL36 ATG start codon ( Figure S5A ) ., The agl36-1 line showed Mendelian segregation of the T-DNA insertion , as 75% of the plants were resistant to Hygromycin ( N\u200a=\u200a1025 , χ2\u200a=\u200a0 , 83 , Table S1 ) ., To test the transmission through the male and female gametes directly , reciprocal crosses of both hemizygous and homozygous agl36-1 mutant plants with wild-type plants were performed ( Table S1 ) ., In a reciprocal cross , a hemizygous mutant will segregate 50% of the T-DNA resistance marker if the disrupted gene is not vital for gametophyte transmission or function ., Thus , gametophyte requirement can be scored directly as reduced frequency of resistant plants 49 ., In reciprocal crosses with agl36-1 , no transmission distortion through female or male gametophytes could be observed ( N\u200a=\u200a661 , χ2\u200a=\u200a0 , 13 and N\u200a=\u200a1015 , χ2\u200a=\u200a0 , 00 respectively , Table S1 ) ., The position of the T-DNA insertion in agl36-1 predicts AGL36 expression failure , and indeed real-time PCR analyses of 3 DAP seeds of homozygous agl36-1−/− plants compared to Col wild-type indicate a 1000-fold AGL36 downregulation in the mutant seeds ( Figure S5B ) ., In line with an imprinted and maternal-only expression of AGL36 , close to 50% reduction of the transcript level was observed in 3 DAP hemizygous agl36-1+/− seeds ( Figure S5B ) ., We thereby concluded that agl36-1 represents a loss-of-function allele of AGL36 ., Although depletion of AGL36 did not interfere with the fitness of the mutant allele in our experimental system , we have shown that AGL36 is specifically expressed from the maternal allele in the fertilization product , in a time frame between 2 and 6 DAP ., To investigate whether this was reflected morphologically or developmentally in the developing seed , we compared embryo and endosperm development in wild-type and homozygous agl36-1−/− seeds within the AGL36 expression time frame ., After fertilization of the egg and the central cell , the endosperm in Arabidopsis undergoes three syncytial rounds of nuclear divisions before the first asymmetric division of the zygote that creates the apical embryo proper and the basal suspensor that connects the embryo proper and the maternal tissue ( Figure S5C ) ., At the 2 DAP stage , no obvious difference could be observed between wild-type and agl36-1−/− seeds , both typically harboring a 1–2 cell embryo proper and a 16–32 nucleated endosperm ( Figure S5C , left section ) ., The embryo continues to divide through radial , longitudinal and transverse divisions to produce the so-called globular stage at 4 DAP ( Figure S5C , middle section ) ., The endosperm also undergoes 3–4 syncytial nuclear divisions and remains uncellularized as cell proliferation at the upper half of the embryo forms the cotyledon primordia at the so-called heart stage at 6 DAP ( Figure S5C , right section ) ., Although the main AGL36 expression peak occurs during this time frame , no obvious deviation between wild-type and agl36-1−/− could be observed at these stages ., Similarly , using an endosperm specific pFIS2::GUS reporter 33 , a wild-type endosperm division pattern was observed in agl36-1+/− seeds ( Figure S5D ) ., The majority of imprinted , maternally expressed genes identified in Arabidopsis so far have been shown to be paternally silenced by mechanisms involving symmetric CG methylation , maintained by MET1 11 , 18 , 19 ., Although not directly linked to imprinting , methylation can also be directed by CHROMOMETHYLASE 3 ( CMT3 ) that has specificity for CNG , and members of the DOMAINS REARRANGED METHYLTRANSFERASE ( DRM ) family; DRM1 and DRM2 , that are mainly responsible for asymmetric CHH methylation 50 ., In order to address the involvement of DNA methylation in the regulation of paternal AGL36 expression , we performed SNP analyses of 3 DAP ovules from reciprocal crosses with mutants that have been shown to be involved in DNA methylation ., In the SNP RT-PCR analysis of mutant pollen crossed to wild-type , paternal AGL36 expression is expected if the tested mutants are involved in AGL36 imprinting ., CMT3 DNA methylation has been reported to be guided to specific sites by KRYPTONITE ( KYP ) H3K9 methylation 51 ., When mutant cmt3-7 and kyp-2 pollen were crossed to Col wild-type plants , no difference in AGL36 expression was observed ( Figure 5A ) ., In the reciprocal cross with cmt3-7 also no difference could be detected compared to wild-type expression ( Figure S6 ) ., DRM1 and DRM2 are mainly responsible for asymmetric DNA CHH methylation 50 and rely on small interfering RNAs , processed by ARGONAUTE4 ( AGO4 ) , for target template guidance 52 ., In our assays , fertilization by pollen lacking DRM1;DRM2 and pollen lacking AGO4 had no effect on the AGL36 expression pattern ( Figure 5A ) ., Likewise , AGL36 expression in the reciprocal cross was identical to wild-type ( Figure S6 ) ., DECREASE IN DNA METHYLATION1 ( DDM1 ) is involved in maintenance of DNA methylation 53 ., In our SNP RT-PCR analyses where mutant ddm1-2 pollen was used to fertilize wild-type ovules , paternal AGL36 expression was not activated ( Figure 5A ) ., In summary , CMT3 , KYP , DRM1;DRM2 , AGO4 and DDM1 appear not to be involved in the establishment nor maintenance of AGL36 imprinting ( Figure 5A , Figure S6 ) ., However , paternal AGL36 expression was detected when plants hemizygous for the met1-4 mutation were used as pollen donor in crosses with wild-type Ler ( Figure 5B ) ., In the reciprocal cross , using met1+/− as the maternal partner , no AGL36 expression from the paternal genome could be observed ( Figure 5B ) ., Furthermore , we performed crosses using pollen from homozygous met1-4 parents ., When first generation homozygous met1 plants were used as pollen donor on wild-type plants , prominent AGL36 expression from the paternal Col genome could be observed ( Figure 5B ) ., This strongly suggests that the repression of the paternal copy of AGL36 is lifted due to the met1-4 mutation , and that MET1 is required for maintaining paternal inactivation of AGL36 ., In the reciprocal crosses , only expression from the maternal genome could be detected , both in the heterozygous and the homozygous met1-4 situation , further substantiating the requirement of MET1 in the male germline in order to maintain AGL36 imprinting ( Figure 5B ) ., Maternal AGL36 expression levels using homozygous met1-4 as the maternal cross partner appeared to be equal to maternal levels in the reciprocal crosses ( Figure 5B ) ., This opens for the interpretation that DNA methylation is not required for the regulation of maternal AGL36 expression ., In public expression databases , AGL36 is reported to be expressed in the seed and more precisely in the endosperm 54 ( Figure S4 ) ., In order to monitor AGL36 expression in vegetative tissues and its dependence on DNA methylation , we performed a real-time PCR experiment on vegetative tissues from reciprocal Ler x Col crosses and homozygous met1-4 tissues ., In biological replicates of progenies from both reciprocal crosses , weak AGL36 expression ranging from 1–6% of the seed expression level could be detected in seedlings , leaves and flowers ( Figure 6A ) ., This showed that AGL36 was expressed throughout the plant life cycle , although at very low levels ., In the same experiment , we monitored expression in met1-4 tissues ., AGL36 expression levels were 50–90-fold higher in met1-4 leaves compared to seed expression levels ( Figure 6A ) ., In a direct comparison , expression levels were elevated 2000-fold in homozygous met1-4 leaves compared to wild-type Col x Ler leaves ( Figure 6B ) ., In flowers , the upregulation was more than 20-fold in met1-4 compared to wild-type Col x Ler flowers ( Figure 6C ) ., In conclusion , these data showed that silencing of AGL36 in vegetative tissues involves MET1 , suggesting that the absence of maintenance DNA methylation elevates vegetative AGL36 expression beyond the maternal expression levels found in seeds ., In order to investigate the parental expression pattern of AGL36 in vegetative tissues , we performed SNP analyses of flowers from F1 hybrids of Ler and Col reciprocal crosses ., In both reciprocal crosses , AGL36 appeared to be expressed equally from the parental Ler and Col genomes , indicating biparental expression in flowers ( Figure 6D ) ., This indicates that parental-specific expression , i . e . imprinting of AGL36 , as expected , only takes place in the seed and that a low basal biparental expression is present throughout the plant life cycle ., Interestingly , biallelic expression in flowers suggests that further silencing of AGL36 takes place in the male germline before uniparental expression in the seed ( Figure 6D ) ., According to our data , the action of MET1 suppresses AGL36 expression throughout the vegetative phase and this suppression is maintained in the fertilization product through the male germline ., AGL36 imprinting thus requires specific activation of the maternal allele ., DNA demethylation by DME has previously been shown to mediate maternal-specific gene expression in the endosperm 11 , 18 , 19 , 24 , and we therefore investigated AGL36 expression in dme-6 mutant plants ., Since dme cannot be maintained in a homozygous state , we harvested siliques of dme-6+/− heterozygous plants pollinated with Col pollen at 3 and 6 DAP ., We monitored the relative expression by means of real-time PCR using FWA and FIS2 as controls ., At 3 DAP , both controls were downregulated by 69±0 . 09% and 53±0 . 30% respectively ( Figure 6E ) , in line with a lack of functional DME in 50% of the seeds in heterozygous dme-6+/− plants ., AGL36 was downregulated in a similar manner as FIS2 ( 41±0 . 20% ) , suggesting that DME is indeed involved in early activation of the maternal AGL36 allele ., We also tested the expression of FWA and FIS2 in 6 DAP samples and found that their downregulation were sustained as predicted ( Figure 6E ) ., However , to our surprise AGL36 expression in dme-6+/− seeds was elevated more than 50-fold ( Figure 6E ) ., This result was unexpected , and implicated a more intricate regulation of AGL36 ., DME is required for the activation of MEA , the core histone H3K27 methyltransferase ( HMTase ) of the PRC2 FIS-complex 46 , 55 , 56 ., To determine whether PRC2 FIS is involved in the regulation of AGL36 , we analyzed the relative expression of AGL36 over time ( 1 to 12 DAP ) in mea mutant seeds compared to wild-type ( Figure 7A ) ., While AGL36 expression in wild-type seeds was at its maximum at 4 DAP , we observed that AGL36 expression in mea seeds surpassed the maximum levels of wild-type at 4 DAP , and reached its highest levels at around 6 DAP ., At this point , the AGL36 relative expression in mea mutant seeds was approximately 40-fold higher than wild-type expression at the same stage , and 7-fold higher than the maximum AGL36 level found in wild-type seeds at 4 DAP ( Figure 7A ) ., Our data thus indicate that the FIS-complex is indeed a repressor of AGL36 expression , and could also explain the elevated AGL36 expression level in 3 DAP dme-6+/− seeds ( Figure 6E ) ., In line with these findings , we found highly elevated AGL36 relative expression levels in mutant seeds from three different mutant alleles of mea ( Figure 7C ) ., Similar results were also obtained with mutants of other components of the FIS PRC2 complex ( FIS2 , FIE and MSI1 , data not shown ) ., To investigate whether FIS activity was exerted on the maternal and/or paternal allele of AGL36 , we performed SNP analyses on the RT-PCR product of AGL36 obtained from mea mutant plants ( in Ler background ) pollinated with Col wild-type pollen ., We found that AGL36 is expressed only from its maternal allele in the mea background throughout the duration of our experiment ( Figure 7B ) ., In comparison to the expression pattern in wild-type ( Figure 3B ) , strong ectopic maternal expression was also observed at 9 and 12 DAP stages ., No paternal expression could be observed in these stages ., By plotting the molarities of the maternal band detected by the Agilent Bioanalyzer , an expression profile for the maternal allele could be generated ( Figure 7B , lower panel ) ., This demonstrated that in the absence of MEA , AGL36 expression continues to increase after 4 DAP , and although the intensity decreases from 6 DAP , high level of AGL36 is maintained at 12 DAP ., Hence , the FIS-complex represses the maternal allele of AGL36 after the 3 DAP stage ., To further substantiate that maternal AGL36 expression is regulated by the maternal action of MEA , we crossed mea mutant plants with pollen expressing the pAGL36::GUS reporter line ., Here , no obvious activation of the paternal transgene could be observed at 3 DAP ( Figure S7A ) ., Surprisingly , at 6 DAP , corresponding to embryo heart stage , weak expression of the paternal copy in the embryo could be found ( Figure S7A ) ., In addition , we performed reciprocal crosses with the pAGL36::GUS reporter line in mutant mea background ., When the transgene was contributed from the female side in mea background , a GUS signal was found in 3 DAP stages that increased drastically up to 6 DAP ( Figure S7B ) ., In the reciprocal cross however , no expression could be observed ( Figure S7C ) ., The E ( z ) class of H3K27 histone methyltransferases ( HMTases ) in Arabidopsis consists of MEA , SWINGER ( SWN ) and CURLY LEAF ( CLF ) that participate in different PRC2 complexes ., To test whether AGL36 repression is a specific function of FISMEA PRC2 , we analyzed AGL36 expression in homozygous swn-4 and clf-2 seeds ., For mutants of both HMTases values similar to the wild-type situation were found , and in conclusion AGL36 appear to be specifically regulated by FISMEA PRC2 ( Figure 7C ) ., In summary , maternal AGL36 expression appears to be repressed specifically by the maternal action of FIS PRC2 ., For all genes known to be imprinted by PRC2 , the FIS-complex is involved in the repression of the silenced allele 25-27 , 30 , 56 ., Our data suggest that silencing of the paternal AGL36 allele requires MET1 whereas the maternal allele is activated by DME ., Modulation of female AGL36 expression by PRC2 thus represents a novel mechanism in this type of gene expression system , and adds an additional level of parent-of-origin specific gene expression to the scheme ., In order to investigate if this regulation applies to other genes imprinted by the dual action of MET1/DME 11 , 18 , 19 , we analyzed the relative expression levels of FWA , FIS2 , AGL36 and MPC in a mea mutant ., At 3 DAP expression levels were unchanged or slightly downregulated ( 0 . 40–0 . 99 ) for all genes tested ( Figure 7D ) ., However , while the expression of FWA and FIS2 remained stable at 6 DAP , AGL36 and MPC levels were elevated up to 80-fold ( Figure 7D ) ., Thus , genes imprinted by means of MET1/DME can be divided in two classes based on their dependence of FIS PRC2 for additional regulation of the expressed allele ., Whereas one class appears not to be regulated by FIS PRC2 , the other class depends on the action of the FIS-complex for developmental regulation of its expression ., Here , we report that AGL36 is a novel imprinted gene that is only expressed from its maternal allele in the endosperm ., Silencing of the paternal allele requires the action of MET1 , as paternal expression is restored in met1 mutants ., In public high-density DNA methylation maps prepared from wild-type seedlings ( http://signal . salk . edu ) , both the AGL36 transcribed region and the 5′and 3′regulatory regions are decorated by CG methylation ., In line with this , AGL36 was expressed at very low levels in vegetative tissues ., Transcript levels however , were highly elevat

Introduction, Results, Discussion, Materials and Methods

Seed development in angiosperms is dependent on the interplay among different transcriptional programs operating in the embryo , the endosperm , and the maternally-derived seed coat ., In angiosperms , the embryo and the endosperm are products of double fertilization during which the two pollen sperm cells fuse with the egg cell and the central cell of the female gametophyte ., In Arabidopsis , analyses of mutants in the cell-cycle regulator CYCLIN DEPENDENT KINASE A;1 ( CKDA;1 ) have revealed the importance of a paternal genome for the effective development of the endosperm and ultimately the seed ., Here we have exploited cdka;1 fertilization as a novel tool for the identification of seed regulators and factors involved in parent-of-origin–specific regulation during seed development ., We have generated genome-wide transcription profiles of cdka;1 fertilized seeds and identified approximately 600 genes that are downregulated in the absence of a paternal genome ., Among those , AGAMOUS-LIKE ( AGL ) genes encoding Type-I MADS-box transcription factors were significantly overrepresented ., Here , AGL36 was chosen for an in-depth study and shown to be imprinted ., We demonstrate that AGL36 parent-of-origin–dependent expression is controlled by the activity of METHYLTRANSFERASE1 ( MET1 ) maintenance DNA methyltransferase and DEMETER ( DME ) DNA glycosylase ., Interestingly , our data also show that the active maternal allele of AGL36 is regulated throughout endosperm development by components of the FIS Polycomb Repressive Complex 2 ( PRC2 ) , revealing a new type of dual epigenetic regulation in seeds .

Seeds of flowering plants consist of three different organisms that develop in parallel ., In contrast to animals , a double fertilization event takes place in plants , producing two fertilization products , the embryo and the endosperm ., Imprinting , the parent-of-origin–specific expression of genes , typically takes place in the mammalian placenta and in the plant endosperm ., A prevailing hypothesis predicts that a parental tug-of-war on the allocation of available recourses to the developing progeny has led to the evolution of imprinting systems where genes expressed from the mother dampen growth whereas genes expressed from the father are growth enhancers ., The number of imprinted genes identified in plants is low compared to mammals , and this precludes the elucidation of the epigenetic mechanisms responsible for this specialized expression system ., Here , we have used genome-wide transcript profiling of endosperm without paternal contribution to identify seed regulators and , among these , imprinted genes ., We identified a cluster of downregulated MADS-box transcription factors , including AGL36 , that was subsequently shown to be imprinted by an epigenetic mechanism involving the DNA methylase MET1 and the glycosylase DME ., In addition , the expression of the active AGL36 allele was dampened by the FIS Polycomb Repressive Complex , identifying a novel mode of regulation of imprinted genes .

genetics and genomics/gene expression, plant biology/plant growth and development, genetics and genomics/epigenetics, plant biology/plant genetics and gene expression, genetics and genomics/plant genetics and gene expression

journal.pbio.1001025

Local Ca2+ Entry Via Orai1 Regulates Plasma Membrane Recruitment of TRPC1 and Controls Cytosolic Ca2+ Signals Required for Specific Cell Functions

Store-operated Ca2+ entry ( SOCE ) is activated in response to a reduction of Ca2+ in the ER ., SOCE generates local and global Ca2+i signals that regulate a wide variety of cellular functions 1 , 2 ., The first store-operated Ca2+ channel to be characterized , the Ca2+ release-activated Ca2+ ( CRAC ) channel , has a high selectivity for Ca2+ versus Na+ and displays a typical inwardly rectifying current-voltage relationship ., CRAC channel accounts for the SOCE in lymphocytes and mast cells 3–6 and has recently been detected in some other cell types 7–9 ., Key molecular components of the channel are STIM1 and Orai1 ., STIM1 is an ER Ca2+ binding protein that has been established as the primary regulator of SOCE 10–12 ., In response to store depletion STIM1 oligomerizes and translocates to ER/PM junctional domains where it aggregates into puncta ., The site of these aggregates is the location where STIM1 interacts with and activates channels involved in SOCE 13–15 ., Orai1 is the pore-forming subunit of the CRAC channel 16–18 ., Following store depletion , Orai1 , which is localized diffusely in the plasma membrane in resting cells , is recruited by STIM1 into the puncta and gated by interaction with a C-terminal region of STIM1 19 , 20 ., While expression of this STIM1-domain induces spontaneous CRAC channel activation in extra ER/PM junctional domains , the site of the STIM1 puncta represents the cellular location where endogenous SOCE is activated by store depletion 21 ., Store depletion also leads to activation of relatively non-selective Ca2+-permeable cation channels , usually referred to as SOC channels , that have been associated with SOCE in several other cell types 2 , 22–25 ., Despite more than a decade of studies , the molecular components of these channels have not yet been established and their function and regulation remain somewhat controversial ., TRPC channels have been proposed as molecular components of SOC channels ., Data in this regard are strongest for TRPC1 2 , 26–34 although TRPC3 and TRPC4 also appear to contribute to SOCE in some cell types 23 , 25 , 35–38 ., Numerous studies show that disruption of TRPC1 attenuates SOCE and SOCE-dependent cell function 23 , 26–34 ., We have previously provided extensive data to demonstrate that TRPC1 is a critical component of SOC channels and SOCE in the human salivary gland cell line , HSG 30 , 39–42 ., Further , salivary gland acinar cells from TRPC1−/− mice display reduced SOCE and SOC channel activity , which account for loss of sustained KCa activation and , consequently , salivary fluid secretion 29 ., However , the role of TRPC1 in SOCE has been questioned based on the lack of function of heterogously expressed channels 43 ., Further , some tissues from TRPC1−/− mice do not display any changes in SOCE 44 , 45 ., The strongest evidence for the regulation of TRPC1 following store depletion has been provided by data demonstrating that STIM1 interacts with and activates TRPC1-SOC channels in response to Ca2+ store depletion 39 , 42 , 46 ., SOC channels are attenuated by knockdown of endogenous STIM1 and spontaneously activated by expression of the STIM1 mutant , D76ASTIM1 42 , 46 ., An important study showed that TRPC1 is gated by electrostatic interaction between STIM1 ( 684KK685 ) and TRPC1 ( 639DD640 ) 47 ., An intriguing finding is that STIM1 alone is not sufficient for activation of TRPC1-SOC channels following Ca2+ store depletion ., Functional Orai1 is also required since knockdown of Orai1 or expression of functionally deficit Orai1 mutants prevents TRPC1 activation 39 , 48 ., We have shown earlier that store depletion leads to the recruitment of a TRPC1/STIM1/Orai1 complex that is associated with the activation of SOCE 39 , 42 ., Thus , while STIM1 is the primary protein involved in SOC channel gating , both TRPC1 and Orai1 appear to contribute to SOC channel activity ., There has been much debate about the essential role of Orai1 in TRPC1-SOC channel function and more specifically regarding whether TRPC1 and Orai1 contribute to a single SOC channel pore or whether Orai1 is a regulatory subunit of SOC channels ., In this study we have assessed the critical role of Orai1 in regulation of TRPC1 function following intracellular Ca2+ store depletion and determined the contributions of TRPC1 and Orai1 to SOCE ., We report that TRPC1 and Orai1 constitute two distinct channels that contribute to SOCE in HSG cells ., Suppression of TRPC1 function unmasks the underlying CRAC channel function ., Further , in response to store depletion , STIM1 mediates association of Orai1 and TRPC1 within ER/PM junctional domains ., Ca2+ entry via Orai1/STIM1-CRAC channel triggers plasma membrane insertion of TRPC1 and gating is achieved by interaction with STIM1 ( 684KK685 ) residues ., Remarkably , while both Orai1 and TRPC1 contribute to Ca2+i increase following store depletion , they impact different cellular functions ., Ca2+ entry mediated by TRPC1 is the primary regulator of KCa channel and partially contributes to NFκB activation while Orai1-mediated Ca2+ entry alone is sufficient for maximal NFAT activation and partial NFκB activation ., Together these findings reveal the molecular events that determine activation of TRPC1 channels following store depletion ., We suggest that local Ca2+ entry mediated by Orai1 determines plasma membrane insertion of TRPC1 while gating by STIM1 controls its activation ., Thus , Orai1 and STIM1 not only determine Ca2+ signals generated by CRAC channels but by regulating TRPC1 channel activity rapidly modulate Ca2+i and thus significantly impact various cell functions ., Compared to SOCE in control HSG cells ( transfected with vector or scrambled siRNA; black traces in Figure 1 ) , knockdown of endogenous Orai1 , STIM1 , or TRPC1 attenuated thapsigargin ( Tg ) -stimulated Ca2+ influx by >90% , >80% , or >60% , respectively ( Figure 1A ) ., These conditions did not significantly affect internal Ca2+ release ., Western blots ( Figure S1A ) demonstrate the effectiveness of TRPC1 knockdown in these cells ., Ca2+ entry induced by Tg treatment of HSG cells was blocked by 1 µM Gd3+ and 20 µM 2APB ( Figure S1B ) ., Further , expression of TRPC1 , TRPC1+STIM1 , Orai1+STIM1 , or TRPC1+STIM1+Orai1 increased Tg-stimulated Ca2+ entry ( Figure S1G ) , which was also blocked by 1 µM Gd3+ and 20 µM 2APB ( Figure S1C–F ) ., Together , these data are consistent with our previous studies 42 that Orai1 , STIM1 , and TRPC1 contribute to endogenous SOCE in HSG cells ., Additionally , the contributions of TRPC1 , STIM1 , and Orai1 to SOCE were not dependent on the level of stimulation ( Figure S2 ) ., The relative decrease in SOCE induced by individual knockdown of the three proteins was similar in cells stimulated with 100 µM carbachol ( CCh , a maximal stimulatory concentration ) or 1 µM CCh ( submaximal stimulatory concentration ) ., The contribution of TRPC1 and Orai1 to SOCE in HSG cells was further examined by using whole cell patch clamp technique 2 , 16 , 17 , 40 to record the current generated by intracellular Ca2+ store depletion ( Figure 1B ) ., Consistent with our previous findings , Tg stimulation of cells resulted in activation of ISOC in HSG cells that is distinct from the typical ICRAC currents measured in RBL cells and T lymphocytes 40 ., We have previously reported 40 that ISOC is a relatively Ca2+-selective cation current with Erev around +20 mV and pCa2+/pNa+\u200a=\u200a40 ( ICRAC displays Erev>+60 mV and Ca2+/Na+ selectivity ≥400 ) ., Silencing of Orai1 expression blocked generation of ISOC while knockdown of TRPC1 by shRNA significantly reduced the amplitude of the inward current but induced more pronounced loss of the outward current ., Thus the residual current detected in 6/10 shTRPC1 treated cells was more inwardly rectifying , i . e . more like ICRAC ( Figure 1B , blue trace ) ., These findings indicate the possibility that ICRAC in HSG cells can be masked by the larger relatively non-selective TRPC1-mediated current that is activated under the same conditions ., The extent of TRPC1 knockdown would then determine the detection ICRAC ., In the present set of experiments , 40% of the cells displayed ISOC or reduced ISOC ., Our present data are somewhat contradictory to our previous finding that the residual current in Tg-stimulated submandibular gland acinar cells from TRPC1−/− mice was a much reduced transient current that was linear and did not display ICRAC-like properties ( i . e . activation by low 2APB or increase in DVF medium ) 29 ., We suggest that other TRPC channels or volume-regulated channels could account for the linear current ., While further studies are required to determine the channel ( s ) involved in this residual current , our previous findings strongly demonstrate that TRPC1 contributes to SOCE and is critically required for salivary gland fluid secretion ., The two C-terminal residues of STIM1 ( 684KK685 ) mediate gating of TRPC1 via electrostatic interaction with TRPC1 ( 639DD640 ) residues 47 ., Consistent with this , expression of a STIM1 mutant that lacks ability to gate TRPC1 , STIM1 ( 684EE685 ) , induced suppression of SOCE in HSG cells while expression of WT-STIM1 resulted in a small increase in function ( Figure 2A ) ., Expression of the TRPC1 mutant that cannot be gated by STIM1 , TRPC1 ( 639KK640 ) , induced a similar suppression of endogenous SOCE ( Figure 2A , blue trace ) ., Further , TRPC1 was not activated by store depletion when expressed with STIM1 ( 684EE685 ) in HEK293 cells ( Figure S3A ) , but when STIM1 and TRPC1 mutants were expressed together ( i . e . “charge swap” between the proteins ) there was recovery of SOCE ( Figure S3A ) ., Importantly , STIM1 ( 684EE685 ) stimulated Orai1 similar to WT-STIM1 ( Figure S3B ) ., A key finding of this study , shown in Figure 2B , is that expression of STIM1 ( 684EE685 ) resulted in generation of ICRAC in response to Tg-induced Ca2+ store depletion in >70% of HSG cells displaying currents ., Together the data in Figures 1B and 2B suggest that ISOC in HSG cells is composed of a small Orai1-mediated ICRAC and a larger TRPC1-mediated non-selective current ( note that we have not yet measured an isolated TRPC1+STIM1 current ) ., To conclusively demonstrate that endogenous Orai1 mediates ICRAC in HSG cells we expressed the STIM1-Orai1-activating region ( SOAR ) 20 ., A large increase in basal Ca2+ entry ( Figure 2C ) and spontaneous ICRAC was seen in these cells ( Figure 2D ) ., SOAR-induced spontaneous SOCE was abolished by knockdown of endogenous Orai1 but was not affected by knockdown of endogenous TRPC1 ( Figure 2C ) ., In contrast , Tg-stimulated Ca2+ entry in SOAR-expressing cells was significantly reduced by knockdown of TRPC1 ( Figure 2E , the residual Ca2+ entry reflects spontaneous Orai1-dependent Ca2+ influx ) ., In aggregate , these data provide strong evidence that endogenous Orai1 mediates ICRAC without any contribution from TRPC1 while SOCE and ISOC display significant contribution from TRPC1 ., Importantly , the function of TRPC1 requires Orai1 ., To identify the mechanism involved in regulation of TRPC1-SOC channels we examined the effect of intracellular Ca2+ store depletion on the surface expression of TRPC1 ., In resting cells the surface expression of TRPC1 ( i . e . in the biotinylated fraction ) was relatively low ., Tg treatment of cells ( Figure 3A , left panel , total TRPC1 and GAPDH are shown in input ) significantly enhanced ( about 3-fold , Figure 3C ) the insertion of TRPC1 into the plasma membrane ., An important finding of this study ( Figure 3A ) is that Tg-stimulated increase in plasma membrane insertion of TRPC1 was dependent on Orai1 ., Decreasing Orai1 expression or compromising Orai1 function by expression of the dominant negative mutant Orai1-E106Q ( Figure 3A , middle and right panels , respectively , see Figure 3C for average data ) severely reduced Tg-stimulated surface expression of TRPC1 without significantly affecting the resting level of TRPC1 ., To examine whether Ca2+ entry was involved in TRPC1 trafficking , biotinylation of TRPC1 was assessed in cells stimulated with Tg in nominally Ca2+-free medium or in normal Ca2+-containing medium with 1 µM Gd3+ ., Both conditions blocked the increase in the surface expression of TRPC1 induced by Tg ( Figure 3B and C ) ., These effects on TRPC1 trafficking were not due to loss of TRPC1/STIM1/Orai1 clustering , which was not affected in cells expressing Orai1-E106Q 39 or in the absence of external Ca2+ ( unpublished data ) ., The role of Orai1-mediated Ca2+ entry was more directly assessed by using Orai1-E106D , an Orai1 mutant that is permeable to Ca2+ in Ca2+-containing medium , but unlike the wild type channel , it is permeable to Na+ in nominally Ca2+-free medium ., Tg treatment of cells expressing this mutant induced surface expression of endogenous TRPC1 in Ca2+-containing medium but not in Ca2+-free medium ( Figure 3D ) ., Finally , trafficking of TRPC1 was examined in HSG cells expressing STIM1 ( 684EE685 ) , which display ICRAC in response to Ca2+ store depletion ( see Figure 2B ) ., Although TRPC1 activation was suppressed in these cells , trafficking of the channel was not altered ( Figure 3E ) ., In aggregate these novel data suggest that Orai1-mediated Ca2+ influx is sufficient for plasma membrane insertion of TRPC1 but not activation; the latter depends on STIM1 ., The mechanism involved in the clustering of TRPC1 with STIM1 and Orai1 was assessed by TIRFM ., Ca2+ store depletion resulted in co-localization of YFP-TRPC1 and Orai1-CFP into puncta in the sub-plasma membrane region ( Figure 4A , HA-STIM1 was co-expressed in these cells ) ., Further , STIM1 co-clustered with both the channels following Tg stimulation of the cells ( Figure 4B ) ., As has been reported for Orai1 , Orai1-TRPC1 clustering also required co-expression of STIM1 ( unpublished data ) and was not detected in cells when endogenous STIM1 expression was knocked down ( Figure 4C ) ., More significantly , co-IP of endogenous TRPC1 and Orai1 was abolished in cells treated with siSTIM1 ( Figure 4D ) but not in cells expressing STIM1 ( 685EE685 ) ( Figure 4E , F ) ., TRPC1 clustering was not dependent on Orai1 since co-clustering of TRPC1 with STIM1 was unaffected by knockdown of Orai1 ( Figure S4 , compare data in A and B ) ., Thus , STIM1 determines TRPC1 clustering in the sub-plasma membrane region following Ca2+ store depletion , and Orai1-mediated Ca2+ entry regulates its surface expression ., Based on these findings we hypothesize that TRPC1 is present in recycling vesicles that traffic in and out of the plasma membrane region ., Following store depletion when STIM1 clusters in ER/PM junctional domains , it interacts with TRPC1 possibly via the ERM domain 46 and increases the retention of TRPC1-containing vesicles ., Concurrently , STIM1 also recruits Orai1 into the same regions , thus bringing the two channels in close proximity to each other ., Ca2+ entry via Orai1 induces fusion of TRPC1-containing vesicles to the plasma membrane followed by gating of the channel by STIM1 ., Further studies are required to elucidate the mechanisms involved in trafficking and plasma membrane insertion of TRPC1 ., We next examined whether relatively global or local Ca2+i increase regulates plasma membrane insertion of TRPC1 ., Figure 5A shows that loading HSG cells with 200 µM BAPTA-AM prior to Tg stimulation ( details given in Methods ) did not suppress trafficking of TRPC1 induced by Tg , although Tg-stimulated global Ca2+i increase was completely suppressed ( Figure 5B , compare red trace with black trace , which shows Ca2+i increase in cells loaded with low BAPTA-AM ) ., In addition , Tg-stimulated ISOC was not altered by replacing EGTA in the pipette solution with 10 mM BAPTA ( Figure S5B , C ) , although the latter condition completely suppressed KCa activation in Tg-stimulated cells ( Figure S5C , right panel ) ., TRPC4 and TRPC5 are directly activated by elevation of intracellular Ca2+i 49 , and a recent study demonstrated that Ca2+ entry mediated via Orai1 or other Ca2+ entry channels , including voltage-dependent channels , can directly enhance TRPC5 activity 50 ., To determine whether Ca2+i increase directly activates TRPC1 , whole cell current measurement was done with Ca2+ in the pipette solution clamped to 0 . 1 µM or 1 µM ., No current was detected with 0 . 1 µM Ca2+ ( unless Tg was included in the external medium , Figure S5A , black and blue traces ) , 1 µM Ca2+ ( Figure S5A , red trace ) , or up to 5 µM Ca2+ ( unpublished data ) ., Note that 1 µM Ca2+ induces >90% activation of TRPC4 and TRPC5 49 , 50 ., These data also rule out possible contribution of other Ca2+-dependent cation channels to SOCE 51 ., In aggregate , these data suggest that local Ca2+ entry via Orai1 determines plasma membrane insertion of TRPC1 and that Ca2+i elevation due to intracellular Ca2+ release is insufficient for triggering TRPC1 insertion ., Further when cells were treated with Tg in a Ca2+ free medium for 5 min , there was no increase in TRPC1 expression in the plasma membrane until Ca2+ was added to the external solution ( Figure 5C , right panel ) ., As shown above , when cells were stimulated with Tg in a Ca2+-containing medium ( Figure 5C , left panel ) , TRPC1 insertion in the plasma membrane was enhanced ., Surprisingly , subsequent removal of Ca2+ from the external solution ( for 10 min ) did not change the level of TRPC1 in the surface membrane ., Functional consequences of these treatments are shown in Figure 5E–F ., In this experiment , HSG cells were treated with Tg in Ca2+-free medium prior to whole cell current measurements in DVF medium ., Typical inwardly rectifying ICRAC with rapid inactivation was detected in these cells ( Figure 5E ) , consistent with the lack of TRPC1 insertion in the plasma membrane under these conditions ., However , when pre-treatment was done in Ca2+-containing medium , ISOC was detected in the DVF medium ( Figure 5F ) ., Note that the ISOC in DVF was relatively sustained , again consistent with the stable biotinylation of TRPC1 ., In aggregate , the findings presented above suggest that Orai1-mediated Ca2+ entry triggers insertion of TRPC1 in the plasma membrane , followed by activation of the channel by STIM1 ., Thus while channel insertion into the plasma membrane appears to depend on local increases in Ca2+i , TRPC1 internalization does not strictly depend on a decrease in Ca2+i ., Further studies will be required to determine the exact molecular mechanisms involved in internalization of TRPC1 ., The data presented above demonstrate that two STIM1-gated channels , Orai1 and TRPC1 , are activated in response to internal Ca2+ store depletion in HSG cells ., To establish the relative contributions of these channels in SOCE-mediated Ca2+ signaling , we examined three SOCE-activated mechanisms: KCa channel , NFκB , and NFAT ., Figure 6A demonstrates that expression of STIM1 ( 684EE685 ) in HSG cells induced a slow , much diminished ( >80% reduction ) , and transiently activated KCa current compared to that in control cells ., As shown above ( Figure 2B ) , only CRAC channel activation was seen in cells expressing this STIM1 mutant ., Thus , Orai1-mediated Ca2+ entry does not appear to be sufficient for activation of KCa activity following Tg stimulation ., Further , NFκB activation ( Figure 6B ) was significantly decreased by the knockdown of TRPC1 expression , and predictably knockdown of Orai1 induced an even greater loss of activity ., Significantly , expression of SOAR did not lead to much activation of NFκB ., Remarkably , TRPC1 had minimal contribution to the regulation of NFAT since knockdown of Orai1 but not TRPC1 suppressed NFAT activation ( Figure 6C ) ., Thus , Orai1-mediated Ca2+ entry is sufficient for regulation of NFAT and for partial stimulation of NFκB , but not for KCa activation ., In contrast , TRPC1-mediated Ca2+ entry regulates KCa channel activity and contributes to NFκB signaling , but not NFAT activation ., Similar to the findings in HSG cells , KCa activity was severely reduced in acinar cells from submandibular glands of TRPC1−/− mice , which could account for loss of salivary fluid secretion in these animals 29 ., While our current findings suggest that Orai1+STIM1 dependent regulation of TRPC1 would be very critical for regulating salivary gland function , functional interaction between these proteins will depend on their precise localization within acinar cells , as is required in HSG cells ( Figure 4B ) ., We have previously reported that TRPC1 is localized in the basal and lateral regions of submandibular gland acinar cells 29 , 52 and that TRPC1 and STIM1 co-IP following stimulation of acini by either Tg or CCh 53 ., To determine possible physiological relevance of the present findings , we examined the localization of TRPC1 , Orai1 , and STIM1 in submandibular glands excised from resting and pilocarpine-stimulated mice ( tissue was fixed in vivo in mice following pilocarpine injection and after an increase in saliva secretion was detected ) ., In the samples from unstimulated mice , endogenous Orai1 was prominantly detected in the apical and lateral regions of submandibular gland acini ( Figure S6A , upper panels , green signal , Orai1 signal shown by white arrows ) , co-localization of Orai1 with the luminal membrane protein AQP5 is also shown ( red signal , right panel ) ., STIM1 showed diffused localization within the acinar cells from unstimulated mice ( Figure S6B , red signal , upper panel ) ., Consistent with our previous findings , diffuse localization of TRPC1 was detected in the basal and lateral regions ( green signal , upper panels , the same sections were labeled for STIM1 and TRPC1 ) ., In samples obtained from stimulated mice , Orai1 and AQP5 localization did not change ( Figure S6A , lower panels ) ., However , a dramatic translocation of TRPC1 and STIM1 to the basal and lateral membrane regions was seen with relative decrease in intracellular staining ( Figure S6B , lower panels , see white arrows ) ., Thus stimulation induces co-localization of STIM1 , Orai1 , and TRPC1 in the lateral membrane region of cells ., While further studies are required to determine whether sufficient Orai1 is present in the basolateral membrane to regulate TRPC1 , our data strongly suggest that regulation of TRPC1 by STIM1 and Orai1 is feasible within the lateral membrane region of salivary gland acinar cells ., Our findings are generally consistent with the strong co-localization of Orai1 and STIM1 in the lateral membrane region of stimulated pancreatic acinar cells 54 ., STIM1 was also localized in the basal membrane and co-localized with heterologously expressed , but not endogenous , Orai1 , in these cells ., This study suggested that localization of Orai1 and STIM1 in the lateral membrane was consistent with the proposed site of Ca2+ entry in exocrine acinar cells 55–57 ., The findings described herein address several important and as-yet unresolved questions regarding the molecular components of TRPC1-SOC channel , the mechanism involved in regulation of the channel in response to store depletion , and its contribution to SOCE ., We report that the previously described ISOC 39 , 40 , 58 , which is stimulated by store depletion and dependent on TRPC1 , STIM1 , and Orai1 , is a sum of Orai1/STIM1-mediated ICRAC and TRPC1/STIM1-mediated non-selective cation current ., Our findings suggest that the latter relatively larger current masks the underlying ICRAC since suppression of TRPC1 function either by shTRPC1 or by expression of the STIM1 ( 684EE685 ) mutant , which does not gate TRPC1 , facilitates detection of ICRAC ., Further , SOAR-activated ICRAC required Orai1 but not TRPC1 ., Thus Orai1 and TRPC1 are components of two distinct channels ., These findings provide strong argument against the possibility that TRPC1 and Orai1 contribute to the same channel pore or that Orai1 is a regulatory , non-conducting , subunit of TRPC channels 59 ., We also report that Orai1-mediated Ca2+ entry triggers plasma membrane recruitment of TRPC1 ., These data reveal a novel function for Orai1 that can explain its critical requirement in the activation of TRPC1 channels following Ca2+ store depletion ., We show that Ca2+ store depletion leads to enhanced surface expression of TRPC1 , which is blocked when Ca2+ is removed from the external medium or SOCE is inhibited by addition of Gd3+ ., Knockdown of endogenous Orai1 expression or expression of non-functional Orai1 mutants ( Orai1-E106Q ) also lead to loss of TRPC1 in the plasma membrane ., Notably , in cells expressing Orai1-E106D , TRPC1 trafficking is supported in Ca2+-containing medium but not Ca2+-free medium ., Together , these findings provide strong evidence that surface expression of TRPC1 is determined by the Ca2+ permeability of Orai1 and that TRPC1 is gated by STIM1 and not directly by Ca2+i increase ., Presently we cannot conclusively rule out the involvement of possible downstream signaling pathway ( s ) activated by Orai1-mediated Ca2+ entry ., The data presented above also reveal important aspects of TRPC1 , Orai1 , and STIM1 clustering that are critical in the regulation of TRPC1 within the same ER/PM junctional domains where Orai1 is regulated by STIM1 ., We show that in response to store depletion TRPC1 co-clusters with STIM1 and Orai1 ., More importantly while Orai1 is not required for clustering and association of TRPC1-STIM1 , localization of STIM1 in the ER/PM junctional domains is critical for recruitment and association of Orai1 and TRPC1 ., Thus far there are no data to show that TRPC1 and Orai1 directly interact with each other , although both channels interact with STIM1 ., STIM1 interacts with Orai1 via the SOAR domain , which also leads to gating of the channel ., In the case of TRPC1 while the C-terminal 684KK685 residues of STIM1 are involved in gating the channel , the ERM domain 46 could interact with the channel and serve as a scaffold to retain TRPC1 within the ER/PM junctional regions ., We suggest that interaction with STIM1 allows the channels to be localized in close proximity to each other , facilitating Orai1-mediated Ca2+ entry to locally regulate plasma membrane insertion of TRPC1 ., However , our data show that internalization of TRPC1 is apparently not dependent on Ca2+i ( Figure 5C ) ., Thus , TRPC1 can remain active provided the Ca2+ stores are depleted and STIM1 is localized in the peripheral domains ., Based on our data , we suggest the following sequence of events in the activation of TRPC1:, ( i ) Ca2+ store depletion leads to translocation of STIM1 to ER/PM junctional domains and recruitment of Orai1 ( localized within the plasma membrane ) and TRPC1 ( likely localized in intracellular trafficking vesicles ) ,, ( ii ) Orai1 is activated by STIM1 and Ca2+ entry via Orai1 triggers exocytosis of TRPC1 , and finally, ( iii ) STIM1 gates plasma membrane TRPC1 ( depicted in the model shown in Figure 7 ) ., We also demonstrate the unique contributions of TRPC1 and Orai1 to SOCE ., Remarkably , different cellular functions are regulated when Orai1 alone is activated compared to conditions when both channels are activated ., Our data suggest that TRPC1 augments the Ca2+i increase resulting from Orai1-mediated Ca2+ entry ., Consistent with this , TRPC1-mediated Ca2+ entry is required for KCa function and contributes to NFκB activation , both of which require relatively higher Ca2+i , but not for NFAT activation , which can be activated at lower Ca2+i ( see Figure 7 ) 29 , 60 ., Interestingly , the requirement of TRPC1 for KCa activity is similar to our previous finding that submandibular gland acinar cells from TRPC1−/− mice display loss of sustained KCa activity , which accounts for the decrease in fluid secretion in these glands ., We have previously shown that TRPC1 is localized in the basal and lateral regions of acinar cells 29 , 52 and that TRPC1 and STIM1 associate following stimulation of acini 53 ., Since Orai1 is critical for TRPC1 function , localization of these proteins in the salivary gland acinar cells is a key determinant for the functional interaction between them ., Feasibility for the interaction of the three proteins and regulation of TRPC1 in the gland is demonstrated by our data ( Figure S6 ) , showing that following agonist stimulation Orai1 , TRPC1 , and STIM1 are strongly co-localized in the lateral membrane region of acinar cells while TRPC1 and STIM1 also appear to be colocalized in the basal region ., In salivary gland acinar cells agonist stimulation leads to Ca2+i elevation , which is first detected in the apical region of the cells and then spreads to basal and lateral regions , irrespective of the level of stimulation 55 , 61 ., Although further studies will be required to confirm the presence of Orai1 in the basal membrane region of acini , co-localization of TRPC1 , Orai1 , and STIM1 in the lateral membrane region of stimulated cells supports our suggestion that Orai1 can regulate TRPC1 function in this region and thus modulate SOCE ., In conclusion , the data described above reveal novel insight into the molecular components and regulation of TRPC1-SOC channels ., Our findings provide strong evidence that TRPC1 and Orai1 constitute distinct SOC and CRAC channels , respectively , both of which are gated by STIM1 in response to store depletion and contribute to SOCE in the same cell ., The critical step in the activation of TRPC1 is its insertion into the plasma membrane , which is governed by Orai1-mediated local Ca2+ entry ., In addition to gating TRPC1 and Orai1 , STIM1 also mediates the association of the two channels within discrete ER/PM junctional domains , which is the site for SOCE 19 , 21 ., The three proteins are also co-localized in the membrane region predicted to be the site of SOCE in acinar cells 56 , 57 , thus highlighting the potential physiological relevance of our findings ., Importantly , TRPC1 augments Ca2+ entry mediated by Orai1-CRAC channels and is required for activation of KCa channels and NFκB , but not NFAT , signaling ., As has been suggested , the amplitude , frequency of oscillations , or spatial patterning of Ca2+i changes determines the regulation of different cell functions 1 , 4 , 51 , 55 , 60 , 62 ., Although further studies are required to elucidate exactly how TRPC1 alters the primary Ca2+i signals generated by Orai1 , the present data suggest that regulation of TRPC1 trafficking can provide a mechanism for rapidly modulating Ca2+i ., STIM1 is emerging as a versatile ER Ca2+ sensor that regulates multiple target proteins in response to Ca2+ store depletion ., In addition to activation of Orai1 and TRPC channels , STIM1 has been reported to inhibit Cav1 . 2 channels 63 , 64 and activate adenylyl cyclase 65 , both of which depend on Ca2+ store depletion ., While regulation of TRPC1 and Cav1 . 2 require association of the channels with Orai1 within ER/PM junctional domains , Orai1 function does not appear to be involved in STIM1-dependent inhibition of Cav1 . 2 ., Thus Orai1 and STIM1 by coordinating the regulation of other ion channels and signaling components can modulate Ca2+i and critically impact SOCE-mediated Ca2+ signaling and a variety of cellular functions ., HSG cells were cultured in MEM medium , supplemented with 10% heat-inactivated fetal bovine serum , and 1% penicillin/streptomycin ., Sequences for the siOrai1 , siSTIM1 , and shTRPC1 targeting to human Orai1 , STIM1 , and TRPC1 , respectively , were similar to previously described sequences 42 ., All siRNA duplexes were obtained from Dharmacon ., Lipofectamine RNAiMAX ( Invitrogen ) was used for siRNA transfection while Lipofectamine 2000 was used for other plasmids ., Cells were typically transfected 24 h after plating and experiments were performed 48 h post-transfection ., All other reagents used were of m

Introduction, Results, Discussion, Methods

Store-operated Ca2+ entry ( SOCE ) has been associated with two types of channels: CRAC channels that require Orai1 and STIM1 and SOC channels that involve TRPC1 , Orai1 , and STIM1 ., While TRPC1 significantly contributes to SOCE and SOC channel activity , abrogation of Orai1 function eliminates SOCE and activation of TRPC1 ., The critical role of Orai1 in activation of TRPC1-SOC channels following Ca2+ store depletion has not yet been established ., Herein we report that TRPC1 and Orai1 are components of distinct channels ., We show that TRPC1/Orai1/STIM1-dependent ISOC , activated in response to Ca2+ store depletion , is composed of TRPC1/STIM1-mediated non-selective cation current and Orai1/STIM1-mediated ICRAC; the latter is detected when TRPC1 function is suppressed by expression of shTRPC1 or a STIM1 mutant that lacks TRPC1 gating , STIM1 ( 684EE685 ) ., In addition to gating TRPC1 and Orai1 , STIM1 mediates the recruitment and association of the channels within ER/PM junctional domains , a critical step in TRPC1 activation ., Importantly , we show that Ca2+ entry via Orai1 triggers plasma membrane insertion of TRPC1 , which is prevented by blocking SOCE with 1 µM Gd3+ , removal of extracellular Ca2+ , knockdown of Orai1 , or expression of dominant negative mutant Orai1 lacking a functional pore , Orai1-E106Q ., In cells expressing another pore mutant of Orai1 , Orai1-E106D , TRPC1 trafficking is supported in Ca2+-containing , but not Ca2+-free , medium ., Consistent with this , ICRAC is activated in cells pretreated with thapsigargin in Ca2+-free medium while ISOC is activated in cells pretreated in Ca2+-containing medium ., Significantly , TRPC1 function is required for sustained KCa activity and contributes to NFκB activation while Orai1 is sufficient for NFAT activation ., Together , these findings reveal an as-yet unidentified function for Orai1 that explains the critical requirement of the channel in the activation of TRPC1 following Ca2+ store depletion ., We suggest that coordinated regulation of the surface expression of TRPC1 by Orai1 and gating by STIM1 provides a mechanism for rapidly modulating and maintaining SOCE-generated Ca2+ signals ., By recruiting ion channels and other signaling pathways , Orai1 and STIM1 concertedly impact a variety of critical cell functions that are initiated by SOCE .

Store-operated Ca2+ entry is present in all cell types and determines sustained cytosolic Ca2+ increases that are critical for regulating a wide variety of physiological functions ., This Ca2+ entry mechanism is activated in response to depletion of Ca2+ in the endoplasmic reticulum ( ER ) ., When ER Ca2+ is decreased , the Ca2+-sensor protein STIM1 aggregates in the ER membrane and moves to regions in the periphery of the cells where it interacts with and activates two major types of channels that contribute to store-operated Ca2+ entry: CRAC and SOC ., While gating of Orai1 by STIM1 is sufficient for CRAC channel activity , both Orai1 and transient receptor potential channel 1 ( TRPC1 ) contribute to SOC channel function ., The molecular composition of SOC channels and the critical role of Orai1 in activation of TRPC1 have not yet been established ., In this study , we demonstrate that TRPC1 and Orai1 are components of distinct channels , both of which are regulated by STIM1 ., Importantly , we show that Orai1-mediated Ca2+ entry triggers plasma membrane insertion of TRPC1 which is then gated by STIM1 ., Ca2+ entry via functional TRPC1-STIM1 channels provides additional increase in cytosolic Ca2+ that is required for regulation of specific cell functions such as KCa activation ., Together , our findings elucidate the critical role of Orai1 in TRPC1 channel function ., We suggest that the regulation of TRPC1 trafficking provides a mechanism for rapidly modulating cytosolic Ca2+ following Ca2+ store depletion .

cell physiology, biochemistry, ion channels, proteins, physiology, biology, anatomy and physiology