Novel cardiovascular gene functions revealed via systematic phenotype prediction in zebrafish

Abstract

Comprehensive functional annotation of vertebrate genomes is fundamental to biological discovery. Reverse genetic screening has been highly useful for determination of gene function, but is untenable as a systematic approach in vertebrate model organisms given the number of surveyable genes and observable phenotypes. Unbiased prediction of gene-phenotype relationships offers a strategy to direct finite experimental resources towards likely phenotypes, thus maximizing de novo discovery of gene functions. Here we prioritized genes for phenotypic assay in zebrafish through machine learning, predicting the effect of loss of function of each of 15,106 zebrafish genes on 338 distinct embryonic anatomical processes. Focusing on cardiovascular phenotypes, the learning procedure predicted known knockdown and mutant phenotypes with high precision. In proof-ofconcept studies we validated 16 high-confidence cardiac predictions using targeted morpholino knockdown and initial blinded phenotyping in embryonic zebrafish, confirming a significant enrichment for cardiac phenotypes as compared with morpholino controls. Subsequent detailed analyses of cardiac function confirmed these results, identifying novel physiological defects for 11 tested genes. Among these we identified tmem88a, a recently described attenuator of Wnt signaling, as a discrete regulator of the patterning of intercellular coupling in the zebrafish cardiac epithelium. Thus, we show that systematic prioritization in zebrafish can accelerate the pace of developmental gene function discovery. KEY WORDS: Systems biology, Zebrafish, Cardiovascular, tmem88a INTRODUCTIONDe novo gene function discovery has been greatly facilitated by systematic gene deletion and observation of resulting phenotypes in scalable model organisms. Indeed, systematic gene disruptions in S. cerevisiae (Costanzo et al., 2010; Giaever et al., 2002), C. elegans (Kamath et al., 2003) and Drosophila (Boutros et al., 2004) have each revealed molecular functions for thousands of genes. However, given the breadth and complexity of observable phenotypes in vertebrates, comprehensive assessment of gene function through serial observation of all possible phenotypes following gene disruption remains infeasible. A more efficient alternative would be to use gene function prediction to prioritize gene candidates for more detailed phenotypic testing based on a variety of known gene and protein properties and relationships. RESEARCH ARTICLE TECHNIQUES AND RESOURCESComputational prediction of molecular function has been effective in assigning physiological roles to genes across eukaryotic model organisms (Deng et al., 2004;Guan et al., 2008; Huttenhower et al., 2009;Karaoz et al., 2004;Lee et al., 2004;Mostafavi et al., 2008;Tasan et al., 2012;Tasan et al., 2008;Troyanskaya et al., 2003). Similar prediction frameworks have been applied to predict associated phenotypes in yeast (King et al., 2003;Saha and Heber, 2006) and worm and to identify putative human disease gene candidates (Linghu et al., 2...

Function-driven discovery of disease genes in zebrafish using an integrated genomics big data resource

“…For example, median fold enrichment by DanioNet for the top 50 or top 100 candidates was over 45, which indicated that DanioNet is highly predictive for the genes involved in wide varieties of anatomical developmental processes. Recently, a classifier based on functional associations between zebrafish genes was used to predict the genes involved in each of the anatomical processes in zebrafish (41). However, this classifier was trained for only a fixed set of ZFA terms, whereas DanioNet-based prediction is not restricted to the predefined ZFA anatomical processes.…”

An extended set of yeast-based functional assays accurately identifies human disease mutations

“…Extrapolating to the set of 1047 human disease-associated genes with a yeast ortholog, complementation assays are within reach for ∼500 disease-associated genes, covering 10% of the ∼5000 disease-associated genes annotated in OMIM or HGMD. This fraction will of course be much higher when all other tractable model organisms are considered, e.g., zebrafish as a model for cardiac developmental disorders (Musso et al 2014).…”

Selecting causal genes from genome-wide association studies via functionally coherent subnetworks

“…Both strategies have been demonstrated to provide high-quality gene-function predictions for (amongst others) H. sapiens 19,24 , M. musculus 17,48,49 , D. rerio 50 , and S. cerevisiae 51 . The union of these two CFNs were then taken as the single CFN used for this study, noting that while gene coverage overlap was high between the two networks, the gene-pair predictions were largely complementary (Supplementary Fig.…”