Gene Expression Regulation in Brains of East Asian, African, and European Descent Explains Schizophrenia GWAS Across Populations

Abstract Psychiatric geneticists have discovered hundreds of common single nucleotide polymorphisms (SNPs) associated with schizophrenia (SCZ) through genome-wide association studies (GWAS). Brain expression quantitative trait loci (eQTL) can successfully explain some of those genetic associations. Differences in genetic association between disparate ancestral populations are often reported, however, it is not known whether such population differences originate from different underlying risk genes or from different allele frequencies and linkage disequilibrium of the same risk genes. Our central hypothesis is that genetic regulation of gene expression within brains, as represented by eQTL, can explain the disease GWAS signals. Population structure influences eQTL as it influences GWAS. The major assumption is that the biological foundation of GWAS and eQTL is the S-E-D relationship, short for SNP-Gene Expression-Disorder. Functional interpretation of GWAS signals relies on the discovery of S-E-D relationships. Due to a lack of brain transcriptome data from populations of non-European descent, interpreting SCZ GWAS results for variants uncommon in other populations presents a significant challenge. To discover the causes of these population differences, we will develop a transcriptome dataset of a new brain collection from East Asians (EA, N = 546) and African Americans (AA, N = 450), combined with samples from the existing PsychENCODE project (EA, N = 18). We will also use data of individuals of African ancestry (AA, N = 411) from the PsychENCODE projects. Along with data from individuals of European descent (EU), which dominates the PsychENCODE (N = 1,321) projects, we will have brain transcription data from three major populations worldwide. Our specific aims include: 1) to relate SNPs to gene expression (the S-E portion of the S-E-D networks), we will develop and compare eQTL and coexpression networks of postmortem brains from three populations, EA, AA and EU; 2) to connect SNP-expression to SCZ GWAS signals (the S-E-D aspect), we will use brain eQTL data to explain SCZ GWAS of EA, EU and AA populations and to identify SCZ risk loci that also serve as regulators of brain gene expression; 3) to develop a novel cross-population predixcan algorithm that can infer genetically regulated gene expression, and identify those differentially expressed in patients. The algorithm will be used to re-analyze existing PGC SCZ data and use Vanderbilt University data to replicate the findings. This study will improve the understanding of the genetic contribution of population variation to SCZ risk.