COMBINED WHOLE EXOME SEQUENCING AND LINKAGE ANALYSIS REVEALS LINKAGE TO 10Q11-10Q21 LOCUS WHICH IS NOT EXPLAINED BY GWAS-ASSOCIATED SNP OR RARE VARIANTS IN ANK3

Bipolar Disorder (BD) is a common, complex psychiatric condition characterized by recurrent fluctuations of mood. Historically, linkage approaches implicated a number of genetic loci, but signals typically failed to be replicated due to heterogeneity of family samples in individual studies and lack of power. Recent Genome-Wide Association Studies (GWAS) have implicated common variants in CACNA1C, ODZ4 and ANK3 as increasing risk of BD, although those individual variants have small effect sizes. Sequencing studies are underway to characterise the impact of common and rare variants in these and other genes. Here we perform a combined whole exome sequencing and linkage study, with the aim of identifying linkage signals and dissecting the contribution of common and rare variants. Whole Exome Sequencing (WES) was performed in 117 subjects from 15 Australian multiplex BD families of European origin (72 affected, of whom 62 had diagnoses of BD type-1 or schizoaffective disorder-manic type). Genotypes were called using samtools pileup and filtered to include haplotype-informative markers (HapMap CEU) using LINKDATAGEN. Familial relationships were confirmed by pair-wise identity-by-descent in PLINK. Genome-wide linkage analysis was performed (using ~6,000 WES-derived SNPs) under non-parametric Kong & Cox linear and exponential models in Merlin. To reduce clinical heterogeneity we considered patients with severe forms of illness as affected (n=62), and subjects with BD-II or depression considered unknown. Regions with suggestive linkage (LOD>2) were fine mapped to reduce inter-marker intervals to <1 cM. Empirical significance of linkage was calculated using 10,000 simulated data sets and permuting phenotypes. Linkage analysis conditioning on GWAS-associated SNPs was performed using LAMP. To assess contribution of potentially damaging functional variation, an allele frequency weighted gene-based rareFBAT test was used. Linkage analysis across 15 extended BD families found a peak on chromosome 10q11-21 (maxSNP=rs10761725; LODlinear=2.52, empirical p=0.045; LODexp=3.03, empirical p=0.046). The 95% confidence interval spanned 15.9 cM (chr10:51,588-67,727 Mb; hg19) and included top GWAS-associated gene ANK3 and 38 other protein-coding genes captured by WES. Conditional linkage analysis with ANK3 GWAS-associated SNP (rs10994397, MAF=0.076 in CEU) demonstrated that this risk variant did not explain the linkage signal, with only a small contribution (LOD=0.63, p=0.089). Examination of haplotype breakpoints did not establish a clear localisation of the gene/s driving the signal. We identified 56 missense or truncating variants (regardless of frequency) informative for association analysis across 23 protein-coding genes, including 5 in ANK3. Preliminary analysis using rare FBAT gene-based tests find no significant evidence for association with any of the 23 genes after multiple testing correction. We found no evidence that variants increasing risk to BD are constrained within any single gene in the 10q11-21 region, including ANK3. Our findings are consistent with the 10q11-21 region containing multiple functional variants that contribute to BD susceptibility with modest effects. Our study suggests heterogeneity within 10q11-21, and highlights the need for examination of genetic variants beyond the ANK3 common risk variants previously identified.