In the hope of one day treating schizophrenia proactively as a developmental disease, researchers continue to seek a clearer profile of genetic risk. The field made a leap forward with the identification of 108 genomic loci associated with schizophrenia in a landmark genome-wide association study (GWAS) by the Psychiatric Genomics Consortium. However, identifying GWAS loci alone has not led to new therapies, in part because their relationship to broader genetic networks remains unclear.
“While knowing GWAS loci associated with schizophrenia is tremendously important, those loci may not be where high-risk genes are located. The loci could be controlling other high-risk genes in entirely different regions,” said Bingshan Li, Ph.D., an associate professor in the Vanderbilt Genetics Institute at Vanderbilt University Medical Center. “The genuine risk genes may not even be in closest proximity to index SNPs (single nucleotide polymorphisms).”
Li and colleagues at Vanderbilt and several other centers have taken the next step in pinpointing high-risk genes that may drive schizophrenia. Using a novel computational framework they call “Integrative Risk Gene Selector” (iRIGS), the team has synthesized multiple lines of genomic and epigenomic evidence related to the known GWAS loci for schizophrenia to infer 104 high-confidence risk genes. Results of this analysis were published in Nature Neuroscience.
Bridging GWAS and Biology
Genetic evidence could guide drug therapy for schizophrenia, given the known role of heritability in the disease. However, current antipsychotics uniformly function by blocking the same target, the dopamine receptor D2, based on a mechanism discovered over 60 years ago. The recent identification of the 108 GWAS loci for schizophrenia has not yet revealed novel drug-targetable genes.
“Poor understanding of the etiology of schizophrenia is a major barrier to finding more effective treatment,” Li said. “Our hope is that by identifying specific high-risk genes, we can help to bridge the gap between our insights from GWAS and the known biology of schizophrenia.”
One way crossing the GWAS-biology divide may help is in repurposing existing drugs for schizophrenia. Li says risk genes shared across different diseases represent a “natural lever.” His team’s analysis pointed to one high-risk gene, GRM3, for example, which has been indicated previously as genetically associated with schizophrenia and is a frequent drug target, including for nervous system drugs.
Computing Risk Genes
Li and colleagues define their iRIGS as a Bayesian framework for probabilistically inferring risk genes based on the integration of: 1) multiple lines of multi-omics data, and 2) biological gene networks. Devising this unique framework allowed them to integrate many existing data sources (e.g., chromatin interaction such as Hi-C, de novo mutations, Genotype-Tissue Expression project) as supporting evidence.
To identify a set of 104 high-confidence risk genes from the massive pool of candidates around the known schizophrenia-associated loci, Li and colleagues devised a unique joint modeling process that both simplified the computation and increased its probabilistic value.
“iRIGS jointly integrates genomic features of a set of risk genes rather than individual genes such that the weak evidence for individual risk genes is amplified by joining forces with the other ones, boosting inference accuracy,” wrote Li and co-authors in Nature Neuroscience.
The team confirmed strong connections between the 104 genes with existing biological models of schizophrenia. The genes accounted for a significantly enriched heritability, were highly consistent with leading pathophysiological disease features (are predominantly expressed in brain tissues, especially prenatally), and were significantly enriched in targets of currently approved drugs.
“This framework opens the door for multiple research directions.”
Li is optimistic about future therapeutic applications of the research to schizophrenia. “I think we’ll have a better understanding of how these genes predispose risk prenatally, and that may potentially inform development of interventions for schizophrenia. It’s an ambitious goal, but by understanding the mechanism we could better focus drug development.”
More broadly, the development of the iRIGS framework could bridge the gap between GWAS modeling and gene biology for other diseases. “This framework opens the door for multiple research directions,” Li said. “It is our hope that it will catalyze the translation of GWAS and biology for a variety of complex diseases.”