|Home | About | Journals | Submit | Contact Us | Français|
Conceived and designed the experiments: JF SK CC YZ. Performed the experiments: SK KC. Analyzed the data: JF SK EG DS KC JN YZ SW. Contributed reagents/materials/analysis tools: EG DS SW. Wrote the paper: JF SK CC KC JN YZ.
We report results from the first genome-wide application of a rational drug target selection methodology to a metazoan pathogen genome, the completed draft sequence of Brugia malayi, a parasitic nematode responsible for human lymphatic filariasis. More than 1.5 billion people worldwide are at risk of contracting lymphatic filariasis and onchocerciasis, a related filarial disease. Drug treatments for filariasis have not changed significantly in over 20 years, and with the risk of resistance rising, there is an urgent need for the development of new anti-filarial drug therapies. The recent publication of the draft genomic sequence for B. malayi enables a genome-wide search for new drug targets. However, there is no functional genomics data in B. malayi to guide the selection of potential drug targets. To circumvent this problem, we have utilized the free-living model nematode Caenorhabditis elegans as a surrogate for B. malayi. Sequence comparisons between the two genomes allow us to map C. elegans orthologs to B. malayi genes. Using these orthology mappings and by incorporating the extensive genomic and functional genomic data, including genome-wide RNAi screens, that already exist for C. elegans, we identify potentially essential genes in B. malayi. Further incorporation of human host genome sequence data and a custom algorithm for prioritization enables us to collect and rank nearly 600 drug target candidates. Previously identified potential drug targets cluster near the top of our prioritized list, lending credibility to our methodology. Over-represented Gene Ontology terms, predicted InterPro domains, and RNAi phenotypes of C. elegans orthologs associated with the potential target pool are identified. By virtue of the selection procedure, the potential B. malayi drug targets highlight components of key processes in nematode biology such as central metabolism, molting and regulation of gene expression.
The arrival of the post-genomic era has brought with it the possibility of in silico selection of drug targets in major human pathogens using rational target-based approaches. Soon after the first microbial genomes were sequenced, comparative and subtractive genomic strategies were proposed to isolate potential drug targets from an organism's complete catalog of gene products. Probable essentiality could be inferred from inter-genomic sequence conservation , and possible lead compound toxicity could be disfavored by focusing on targets that lack close homologs in mammals , . For many bacterial genomes, functional data is now available enabling direct identification of essential genes and has been incorporated into the approach . Unfortunately, for metazoan pathogens, including human helminth parasites, there is a dearth of complete genomic sequences. To complicate matters further, many parasites are genetically intractable, making gene functions difficult to establish experimentally. However, by using a related model organism as a proxy for missing functional genomic data and applying multiple layers of subtractive filters based on comparative sequence analysis, we can pre-validate a pool of targets to facilitate their entry into drug discovery programs. This methodology was tested successfully in parasitic nematodes, albeit incompletely as only fragmentary EST sequence data was available , , and has been endorsed by the World Health Organization as a promising approach to identify new helminth drug targets .
Worldwide, helminth parasites result in a combined conservative disease burden of 8 million DALYs (Disability Adjusted Life Years) . Lymphatic filariasis and onchocerciasis are tropical diseases caused by filarial parasites that are transmitted to humans by insects. Collectively, they afflict approximately 150 million people in over 80 countries with more than 1.5 billion at risk of infection . The mainstay of filarial disease control for several decades has been a limited number of drugs, predominantly diethylcarbamazine, benzimidazoles (e.g. albendazole) and avermectins (e.g. ivermectin) . Ivermectin exerts its anthelmintic effect by modulating the activity of glutamate-gated chloride channel while albendazole binds to tubulin so as to inhibit its polymerization and the subsequent formation of microtubules. The mode of action of DEC is still not understood . These compounds suffer various drawbacks such as not being effective against all stages of the parasite, the requirement for annual or semi-annual administration, possible side effects and contra-indications for certain individuals. Furthermore, signs of emerging drug resistance are becoming increasingly apparent , . Therefore novel chemotherapeutics and vaccines are urgently needed.
In this report, we describe the results from the first application of the in silico filtering methodology to a metazoan parasite genome, the completed draft sequence of Brugia malayi . We have expanded our previous analysis, which was limited to nematode ESTs , and applied this methodology to the complete gene complement predicted for this organism. By incorporating a custom ranking algorithm, we were able to identify and prioritize a pool of 589 potential targets for further study. We also discuss the significance of those candidate targets in terms of nematode biology.
Filarial parasites are related to the free-living nematode Caenorhabditis elegans, a model organism with a fully sequenced and extensively annotated genome. Multiple independent genome-wide analyses of gene function for nearly all ~20000 C. elegans genes have been undertaken using high-throughput RNA interference (RNAi). This data, comprising ~61000 entries, is publicly accessible via Wormbase . The set of genes with non-wild type phenotypes in RNAi screens constitutes a pool of phenotypically significant and potentially essential C. elegans genes. We reasoned that homologs of these genes in B. malayi are also likely to be essential. C. elegans is generally believed to be a valid model for less genetically tractable parasitic nematodes –. Indeed, there is good concordance between the phenotypes resulting from the few cases where genes from filarial nematodes have been targeted by RNAi and similar experiments targeting their C. elegans orthologs –.
Using release 150 of Wormbase (http://www.wormbase.org), we recovered 4827 C. elegans genes with non-wild type RNAi phenotypes (RNAi positive set). From the 11771 predicted gene products in the data snapshot of the B. malayi genome used in our studies, we identified 7435 as having an ortholog in C. elegans (Materials and Methods). Of these, 3059 were mapped to the RNAi positive set, constituting a predicted “essential” B. malayi genome. The majority of these essential genes have close human homologs and were removed. The remainder is a set of 589 first-pass candidate drug targets (Fig. 1, Table S1).
Analysis of protein domains in the target set shows the presence of several over-represented domains as compared to the whole genome (Table S2), suggestive of an important role in nematode biology. The C2H2 type zinc-finger domain and basic helix-loop-helix dimerization domain are over-represented 3- and 4-fold respectively in the target list, as compared to the whole genome, indicative of proteins that bind to nucleic acids and are presumably involved in essential gene regulation and developmental pathways in the parasite. The collagen triple helix repeat, over-represented by 5-fold, reflects unique components of the cuticle and extracellular matrix. Twenty-four potential targets contain InterPro domains that can be mapped to 14 distinct Enzyme Commission (E.C.) numbers (Table S3). Functional classification of the target set using gene ontology (GO) annotations (Table S4) and statistical analysis of the GO term content (Table 1) revealed several over-represented terms including cuticle structure and ion transport.
While the pool of 589 candidates reflects a 20-fold reduction in the search space, it is still too large to enter drug-screening pipelines. To rank the output and identify the most promising potential targets, we developed a computational algorithm for integrating and weighting the biological data from C. elegans and B. malayi (Table 2). The aim of the prioritization algorithm was to predict the efficacy, selectivity and tractability of each candidate target. Hasan et al. recently used a similar approach for prioritizing potential drug targets in Mycobacterium tuberculosis .
Potential targets were rewarded for high sequence similarity with C. elegans orthologs, but penalized heavily for the presence of a close homolog in humans. Based on the protein length ratios of the orthologs, we identified and penalized B. malayi gene models that were incomplete or fragmented. Examples of such gene models include two previously proposed drug targets, 2,3-bisphosphoglycerate-independent phosphoglycerate mutase  (model 13047.m00009) and chitin synthase 2  (models 12621.m00166 and 14328.m00023) respectively; despite being penalized, these gene models appear in the top half of the ranked list based on their high scores in other positive ranking criteria. In some instances, manual prediction of the complete coding region revealed strong similarity to human proteins which was not detected using the incomplete or fragmented models. RNAi phenotype data for C. elegans (obtained from Wormbase) was used to prioritize B. malayi orthologs with respect to their potential efficacy. All reported C. elegans RNAi phenotypes were binned into nine categories and assigned weights based on the severity of the observed phenotype (see Methods and Table S5). Adult/larval lethality/arrest was assigned the highest weight. Replicating the adult lethality phenotype would be an important first step towards developing an effective and much-needed macrofilaricide (compound targeting adult worms). To overcome the complications arising from false positives we used ‘phenotype redundancy’  as a measure of confidence, in which independent experiments using different reagents targeting a single gene produce the same phenotype. The product of severity and redundancy for each phenotype category was summed up and normalized by the total number of RNAi experiments for each gene to provide an aggregate confidence score. Interestingly, when the frequency distribution of the binned RNAi categories for C. elegans sequences orthologous to the target pool was compared with that expected from the whole genome, we observed that reproductive and embryonic phenotypes (sterility and embryonic arrest/lethality) associated with genes involved in highly conserved metazoan processes were under-represented, whereas post-embryonic phenotypes were slightly over-represented (Fig 2). The latter bodes well for our attempts to prioritize drug targets for larvicidal and macrofilaricidal discovery.
Targets were also prioritized based on data for stage specific expression from approximately 24000 ESTs derived from various stage and gender specific B. malayi libraries . Of 589 targets, 252 had corresponding EST sequences. We compiled expression data from microfilariae (L1), L2, L3, L4 and adult stages of the parasite and assigned highest weight to targets which have evidence of expression in all five stages. Next were targets that are expressed in the adults, L4, L1, L3 and L2 stage, in decreasing order of priority.
Other important prioritization criteria included predicted ‘druggability’ and expressability. Druggability can be described as the presence of protein folds that favor interactions with drug-like chemical compounds. Hopkins et al identified 130 InterPro protein domains that are targeted by established and experimental small molecule drugs that follow the Lipinsky rule of 5 (LR5) . Similarly, a list of 70 EC numbers of known enzyme targets and respective marketed drugs was compiled . Proteins with LR5 druggable domains or druggable EC numbers were given a high priority. An important factor for selection of targets for rational drug design is their potential to be expressed in heterologous systems for protein production, purification and crystallization. A genome wide survey for high throughput expression of C. elegans proteins in Escherichia coli found that protein expression and solubility are inversely correlated with hydrophobicity. Proteins having GRAVY (grand average of hydropathicity) scores below an empirically derived cutoff of −0.4 were more likely to be soluble . To prioritize drug targets in B. malayi, we penalized proteins with a GRAVY score higher than −0.4. A complete set of data values used for prioritizing the potential targets are available in Supplementary Data Set S1.
The ranked output (Tables 3 and S1), sorted by the sum of the individual scores for each predicted target, was then manually curated to improve functional annotations where possible. Twelve known or previously proposed targets were identified; nine of these are among the top 40 targets shown in Table 3, endorsing the validity of our approach. Two potential targets, triacylglycerol lipase and adenosine deaminase, having domains associated with druggable enzymes and ten targets with LR5 domains, including the rhodopsin-like GPCR superfamily and integrins (alpha-chain), were found concentrated in the top-half of the list. Many of the candidates were predicted to participate in a variety of essential processes which have no counterpart in mammals, such as molting and synthesis of chitin. Perhaps surprisingly, we also found potential targets that participate in important processes shared across Metazoa. These potential targets are functionally analogous to proteins present in mammals yet they bear no sequence similarity. These include the glycolytic/gluconeogenic enzyme 2,3-bisphosphoglycerate-independent phosphoglycerate mutase (iPGM) characterized previously  and the innexin family of gap junction protein . The functions of some of our potential targets are described below in more detail.
Several potential B. malayi targets identified by our bioinformatics approach may mediate molting. Nematode molting, which takes place 4 times from hatching to adulthood, is a highly regulated and complex process involving the synthesis and secretion of a new exoskeleton, followed by the separation and shedding of the old cuticle . Steroid hormones have been implicated in triggering molting in nematodes, as found in arthropods , . A recent genome-wide RNAi screen in C. elegans has identified 159 genes that are required for molting . These genes may mediate distinct aspects of the process, from intracellular signaling (such as hypodermal-specific transcription factors) to extracellular execution (such as cuticle-digesting proteases). The sequencing of the B. malayi genome has revealed that almost all these genes have a B. malayi counterpart , pointing to phylum-wide conservation in the molting machinery, validating C. elegans as a good model for this process. There is wide agreement that molting represents an excellent process for chemotherapeutic intervention, given that it is an ancestral feature of the phylum Nematoda and does not occur in vertebrates , . Consistent with this, we recovered more than a dozen B. malayi orthologs of proteins necessary for molting in C. elegans which could be considered potential drug targets. These include the B. malayi orthologs of C. elegans NOAH-1 and NOAH-2, which contain zona pellucida (ZP) domains and several plasminogen N-terminal (PAN) modules. These proteins share similarity with Drosophila melanogaster NompA, a component of the extracellular matrix . Other high-ranking targets include the orthologs of C. elegans bli-5 and mlt-11, which encode predicted serine-peptidase inhibitors containing multiple Kunitz/Bovine trypsin inhibitor domains. These protease inhibitors may play a role in regulating the activity of hypodermally-expressed subtilisin-like peptidases, such as BLI-4, which could be required for processing cuticular collagens and activation of further collagen processing/degrading enzymes, such as astacin metallopeptidases . Significantly, Kunitz-type serine protease inhibitors have been implicated in molting in the related filarial nematode Onchocerca volvulus , further supporting the hypothesis that the molecular machinery involved in the molting process is conserved between filarial and rhabditid nematodes.
We also identified B. malayi orthologs of C. elegans mlt-8 and mlt-9. mlt-8 encodes a novel protein that has been proposed to act as an amplifier of endocrine cues during synthesis of the new cuticle, while MLT-9 may be involved in hypodermal signaling . In addition, we identified orthologs of the C. elegans Patched signaling family member ptr-23 and Hedgehog signaling family members qua-1 and wrt-4. These genes have been demonstrated to play a role in molting, even though their functions in the process remain unclear , –. In particular, qua-1, which has been implicated in hypodermal signaling, encodes a nematode-specific cysteine peptidase capable of autocatalytic activation. qua-1 is essential for ecdysis and viability: deletion mutants arrest at the first molt (L1 to L2) exhibiting severe morphological abnormalities. qua-1 orthologs are both well conserved and ubiquitous throughout the phylum Nematoda , making QUA-1 a particularly attractive target for the development of specific inhibitors .
C. elegans has become one of the preferred models to investigate the assembly and molecular interactions of cell junctions because cell-cell and cell-matrix attachment components are generally well conserved between nematodes and vertebrates (reviewed in ). However, a few nematode-specific components do exist, some of which were identified in our screen, including the B. malayi homologs of C.elegans ajm-1 and pat-12/gei-16. The C. elegans coiled-coil protein AJM-1 localizes to apical junctions and is required for embryonic elongation and maintenance of epithelial integrity , . C. elegans pat-12/gei-16 has been implicated in the formation of Fibrous Organelles (FOs), which are found exclusively in nematodes and mediate attachment between body wall muscle and the cuticle across the hypodermis. FOs are essential for viability, ensure maintenance of body rigidity and allow for locomotion . Phenotypic inspection of pat-12/gei-16 mutants, together with the molecular characterization of the gene product function, suggest that the protein acts as an adaptor providing linkages between the various structural components of FOs (Benjamin D. Williams and Caroline A. Behm, personal communication; , ). It is noteworthy that in the human filarial nematode O. volvulus, the homolog of gei-16 encodes the well-characterized OvB20 larval antigen , . Immunogold electron microscopy of O. lienalis with a OvB20-specific serum revealed localization to discrete foci in the hypodermis and cuticle , suggesting that the essential function of pat-12/gei-16 homologs in formation of FOs is likely to be evolutionarily conserved in filiarial nematodes.
Eight B. malayi innexin homologs were identified as potential targets (see Tables S1 and S2). Innexins are invertebrate structural proteins that form intercellular channels, or gap junctions, allowing electrical coupling between adjacent cells (reviewed in ). Distantly related connexins in vertebrates perform analogous functions. In C. elegans, the innexin family comprises 25 paralogs, showing different spatio-temporal expression patterns . Detailed studies on seven C. elegans inx genes have revealed that particular inx genes are required for distinct processes including locomotion, egg laying, synchronized contraction of the pharyngeal musculature and inhibition of oocyte maturation , . Notably, the innexin genes unc-7 and unc-9, which are required for locomotion, also modulate response to the anthelmintic drug ivermectin –.
Chitin is a structural component of the eggshell  and pharynx  of nematodes and it is absent in mammals. As expected, our analyses revealed the two chitin synthase genes previously proposed as drug targets in B. malayi ,  and O. volvulus . These genes are orthologs of the two chitin synthase genes present in the C. elegans genome that are responsible for chitin deposition in the eggshell (chs-1) and pharynx (chs-2) and essential for development . Functional conservation of nematode chitin synthases is highly likely since the B. malayi chs-1 transcript is predominantly found in the oocytes and early embryos . Orthologs of two other C. elegans genes (H02I12.1 and W03F11.1) encoding proteins containing putative chitin binding domains, were also identified. Interestingly, RNAi against H02I12.1, which contains a peritrophin A chitin-binding module, compromises the egg osmotic integrity during early embryogenesis , suggesting that this gene plays a role in eggshell chitin deposition. Thus, aspects of chitin metabolism are clearly essential in nematodes and involve a number of components worthy of further evaluation as drug targets.
The sugar galactofuranose (Galf) is an important component of cell surface glycoconjugates of several prokaryotic and eukaryotic pathogens and has been shown to be essential for viability and virulence –. From the B. malayi genome, we annotated two putative orthologs of UDP-galactopyranose mutase (GLF), the enzyme that is required for biosynthesis of Galf. Both the sugar and the enzyme are absent from mammals making GLF an attractive drug target .
In nematodes, the glucose disaccharide trehalose is proposed to serve as an energy reserve and a protectant against various environmental stresses such as heat, cold and freezing, oxidative and osmotic stress, anoxia, even dessication and anhydrobiosis , . It is an abundant storage sugar in the filarial nematodes Brugia pahangi and Acanthocheilonema viteae  and is also found in bacteria, fungi and insects but not in mammals. We identified trehalose-6-phosphate phosphatase as an ortholog of the essential C. elegans gene gob-1 (gut obstructed). Removal of this gene activity in C. elegans gives rise to larval lethality, partly due to intestinal blockage and subsequent starvation . This gob-1 lethality is completely suppressed when the upstream trehalose-6-phosphate synthase genes are deleted, indicating that the lethality is due to toxic build-up of the intermediate trehalose-6-phosphate .
Mammals take up various unsaturated fatty acids from food as essential nutrients whereas C. elegans has fatty acid desaturases that catalyze the production of polyunsaturated fatty acids . Among the highly ranked targets was the B. malayi ortholog of the essential C. elegans fat-2 gene encoding a Δ-12 fatty acid desaturase that converts oleic acid (18:1) to linoleic acid (18:2) implying that B. malayi also synthesizes polyunsaturated fatty acids rather than acquiring them from the host environment.
The glycolytic/gluconeogenic pathway is present in most cellular organisms, however, the enzymes in the pathway may not be conserved. We identified a 2,3-bisphosphoglycerate-independent phosphoglycerate mutase (iPGM) as such an example. This enzyme has a distinct sequence and structure from the 2,3-bisphosphoglycerate-dependent phosphoglycerate mutase (dPGM) found in mammals. Both enzymes are responsible for the interconversion of 2-phosphoglycerate and 3-phosphoglycerate, however different catalytic mechanisms are involved. The biochemical activities of both B. malayi and C. elegans iPGM enzymes have been demonstrated as well as the essentiality of the gene for nematode development. Down regulation of C. elegans iPGM using RNAi, results in embryonic and larval lethality .
Other potentially interesting targets revealed by our analysis include orthologs of C. elegans transcription factors lin-14, die-1 and pry-1 known to be involved in key developmental and morphogenetic processes. C. elegans lin-14 is a nematode-specific transcription factor required for larval stage-specific gene expression . Mutations in lin-14 cause cell lineage defects in several cell types. The C. elegans gene die-1 belongs to the zinc finger family of transcription factors. Loss of die-1 affects epithelial cell rearrangements during embryonic epidermal morphogenesis, leading ultimately to embryonic arrest . We also recovered the B. malayi homolog of C. elegans pry-1  encoding a protein with limited homology to vertebrate Axins, which act as scaffold proteins in the Wnt/beta-catenin signaling pathway . Despite its sequence divergence, PRY-1, like Axin, serves as a negative regulator in the Wnt signaling pathway in C. elegans and can functionally complement for the Danio rerio (zebrafish) axin1 knockout masterblind . This example illustrates how specific components of signaling pathways, which are conserved between vertebrates and nematodes but have diverged at the primary sequence level, may differ sufficiently to allow for the development of nematode-specific inhibitors.
We also identified genes involved in RNA processing. Trans-splicing, which involves the addition of a short leader sequence to the 5′-end of mRNA, is an essential step in the maturation of most mRNAs in nematodes and several other invertebrates and protozoa (reviewed in ). Our analysis identified the B. malayi orthologs of two known components (SL30p and SL95p) required for in vitro RNA trans-splicing in embryonic lysates from the human nematode Ascaris lumbricoides . Recently, orthologs of these two genes in C. elegans (sut-1 and sna-2 respectively) have also been implicated in RNA trans-splicing . Additionally, we identified an ortholog of C. elegans ego-1, which belongs to a family of RNA-directed RNA polymerases. ego-1 is essential for viability and fertility and in particular plays a crucial role in germline development, where it promotes cell proliferation, meiosis, and gametogenesis. It is thought that EGO-1 influences all these distinct processes by inducing and reinforcing germline RNAi of specific genes –. While many components of the RNAi pathway appear to be missing from the B. malayi genome, most notably the spreading machinery , presence of ego-1 suggests conservation of the role of this class of RNA-directed RNA polymerases in germline silencing across Nematoda.
In addition to drug target discovery, our method highlights proteins participating in biological processes that are necessarily conserved across parasitic and free-living worms; in the case of B. malayi and the sequenced Caenorhabditids these processes span an evolutionary distance of 350 million years since their last common ancestor . This substantially extends our confidence in identifying nematode-centric processes over those conserved only between the Caenorhabditid genomes. Significantly, 50% of the targets were annotated as hypothetical proteins. These may participate in completely novel nematode processes and are worthy of further study.
The recently completed draft genomic sequence of B. malayi has enabled us to predict potentially essential genes and apply a method for rational drug target discovery. In contrast to empirical methods, the bioinformatics approach described herein yields a larger pool of candidates and is not biased, thereby providing a wider range of potential targets. Given the threat of emerging drug resistance resulting from continued reliance on a limited repertoire of available drugs, a wider array of choices for drug targets will be invaluable. The method is also tunable and quickly provides a manageable set of targets for closer analysis. By adjusting the parameters of the comparative sequence analysis, the initial target pool size can be increased or decreased by an order of magnitude. Varying the weights for the factors used in the prioritization scheme can tailor the ranking to the needs of the end-user.
The basic subtractive filtering methodology is applicable to a wide variety of sequenced pathogens, ranging from microbial species to the metazoan parasite analyzed here. Although it is currently limited by the availability of complete genome sequence and functional genomics data, the rapid pace of technological advancements in these areas will soon overcome those limitations, and we expect this methodology to gain widespread applicability.
DNA sequences and protein translations for the B. malayi genome are as described . The sequence set used in this study differs slightly from the final released genome, though efforts were made to maintain synchrony with the final release. Stage specific ESTs and tentative consensus sequences for B. malayi were obtained from the TIGR gene indices project (now housed at Dana Farber Cancer Institute, http://compbio.dfci.harvard.edu/tgi). Complete DNA coding sequence, protein sequence, and RNAi phenotype data from release 150 of the C. elegans genome was obtained from Wormbase (http://www.wormbase.org). Human genome protein sequences were obtained from Ensembl release 41 (http://www.ensembl.org) corresponding to the NCBI build 36 of the human genome.
Ortholog assignments were based on WashU BLASTP all-vs-all analysis, Jaccard clustering, and bidirectional best hit clustering, as described . These assignments were supplemented with one-way best hits using NCBI BLASTP  with an e-value cutoff of 1×10−20, BLOSUM 62 as the scoring matrix and B. malayi sequences as the query. Similarity between B. malayi sequences and human sequences was established using one-way best hits with NCBI BLASTP with an e-value cutoff of 1×10−13 and B. malayi sequences as the query. E-value cutoffs were empirically adjusted to maintain a reasonable target pool size for subsequent literature scanning and retain known potential drug targets, chitin synthase 1 and 2, within the target pool.
Gene ontology (GO) term assignments were obtained as described  where essentially the following procedure was used. Interpro domain assignments were first applied to B. malayi proteins using InterproScan. GO terms attached to the InterPro domains were transferred to the B. malayi gene products using Interpro2GO (http://www.ebi.ac.uk/interpro). A custom GO slim subset of ontology terms generated by TIGR for the B. malayi sequencing project was used to provide a broad overview of the ontology content. Over-representation of GO terms was analyzed using the program Ontologizer  with a p-value cutoff of 0.05, Bonferroni correction, and term-for-term methodology.
Average hydropathy scores (GRAVY) were calculated as the average of the individual hydropathy scores for each amino acid using the data of Kyte and Doolittle . EC numbers were mapped to B. malayi proteins using pre-compiled mapping of EC numbers to GO terms, revision 1.54, available from http://www.geneontology.org/external2go/ec2go.
C. elegans RNAi phenotypes associated with orthologs of the B. malayi candidate drug target sequences were binned into 9 categories with corresponding weights as shown:
To establish a background distribution, all C. elegans RNAi phenotypes were binned into the same categories. Background frequencies were used to estimate expected frequencies for a sample size equal to the size of the RNAi phenotype set associated with the orthologs of the B. malayi
Ranked list of candidate targets. Previously identified targets are shown with bold identifiers. Manually added annotations are shown in italics. B. malayi pub locus and descriptions are from Ghedin et al. . C. elegans gene names and RNAi phenotypes are from Wormbase.
(0.30 MB PDF)
Frequency of Interpro domains in the target sequences.
(0.09 MB PDF)
EC numbers mapped to targets using ec2go.
(0.05 MB PDF)
GO terms associated with target pool sequences. The GO terms are a subset of the GO hierarchy (GO slim). All children of the GO slim nodes are mapped up to the nearest parent in the slim hierarchy. Counts total the occurrences of the exact GO term listed and all its children.
(0.07 MB PDF)
RNAi phenotype components of each binning category.
(0.12 MB PDF)
Data set for target prioritization. Data values used in assigning scores for prioritization of targets. Maximum bit scores for alignments of putative B. malayi, C. elegans and Human orthologs were obtained from BLASTP results (see materials and methods). Bitscores of 0.0 are recorded when no similarity was identified with an E-value below the threshold used in the BLAST comparison. The total number of RNAi experiments reported for each target gene are based on wormbase release 150. Pheno Bins record the number of instances that a phenotype was reported in these experiments that belongs to each of 9 phenotype bins (see text). In this table, the “other/unclassified” bin was split into “other” and “unclassified” bins. Stage expression count refers to the number of distinct life cycle stages (L2, L3, L4, adult and microfilariae) having EST evidence for a particular target gene. L2, L3, L4, adult and microfilariae record the number of ESTs for that stage. Total Score was calculated as described in Table 2. Known targets are indicated in bold. Na indicates “no value”.
(0.13 MB PDF)
We thank Dr. Donald Comb and New England Biolabs for long-standing generous and unwavering support of research aimed at alleviating filariasis, and our colleagues Drs. Richard Roberts, Larry McReynolds and Barton Slatko for comments on the manuscript. We also thank members of the Filarial Genome Consortium and the filarial research community for their continued support and encouragement. The Filarial Genome Consortium was initiated by grants from WHO/TDR/UNDP/WorldBank.
Competing Interests: The authors have declared that no competing interests exist.
Funding: This work was funded by New England Biolabs. Funding for portions of this work was also provided by a grant from the National Institute for Allergy and Infectious Diseases (NIAID/NIH) awarded to E.G.