SU26 - A WHOLE GENOME SEQUENCING STUDY IDENTIFIES A RARE VARIANT IN ANK3 THAT MAY CONTRIBUTE TO BIPOLAR DISORDER

Abstract

Background Over the last decade, a growing number of Genome-Wide Association Studies (GWAS) have been performed to identify common genetic variation that contributes to Bipolar Disorder (BD). These studies have identified a number of genes harboring common risk variants, and also demonstrated that BD has a highly polygenic genetic architecture, with many loci with small effects contributing to the disease. The contribution to risk of BD from rare variants is still not well understood, and whole exome and Whole Genome Sequencing (WGS) studies of BD are only beginning to emerge. Here, to investigate the role of rare variants with potentially greater effect sizes, we used WGS. Methods WGS was performed for 100 BD cases from the Mayo Clinic Bipolar Disorder Biobank and 1000 healthy controls from the Mayo Clinic Biobank. Following bioinformatics processing and genotype calling, Quality Control (QC) was performed. We excluded samples with low coverage or call rate, failed sex check, sample contamination or relatedness. We excluded variants with low call rate, failed HWE test, or failed Variant Quality Score Recalibration (VQSR). After QC, 99 cases and 961 controls with >26 million Single Nucleotide Variants (SNVs) with Minor Allele Frequency (MAF) ≤0.05 remained; sub-setting to annotated non-multiallelic variants within genes resulted in >11million SNVs for analysis. Rare variant gene-level tests were performed using two approaches: (1) burden test of nonsense mutations and missense mutations predicted to be deleterious by SIFT and PolyPhen2; (2) SNP-set kernel association test (SKAT) of all variants with MAF weights. We a priori selected 7 genes implicated in BD by GWAS for prioritized analysis: CACNA1C, ODZ4, ANK3, TRANK1, LMAN2L, NCAN, and SYNE1. Results After restricting analyses to genes in which ≥4 subjects carried variant alleles, 8,242 genes were analyzed by the burden test, and 25,833 genes were analyzed using SKAT. Neither of these gene-level testing approaches identified genome-wide significant associations. Among the previously implicated candidate genes, only ANK3 (p=0.011) and SYNE1 (p=0.006) were associated with BD at a nominally significant level in the burden and SKAT analyses, respectively. In ANK3, the 9 SNVs analyzed by the burden test were observed in 6/99 cases and 18/961 controls; the association result was driven primarily by rs148904927, a SNV that was observed in 5 (5%) cases and 7 (0.7%) controls [OR (95%CI)=7.2 (2.3–23.3)]. Discussion This pilot WGS study suggests that rare variants in ANK3 and SYNE1 may contribute to BD. The ANK3 SNV (rs148904927) that we observed in 5% of the cases (and 0.7% of controls) is a non-synonymous variant (Ser2409Pro) that may function by modifying the isoforms of ANK3 (Ankyrin G), an integral membrane protein that is modulated during developmental stages and is associated with several neuronal dysfunction disorders. According to HaploReg v4.1, Regulome DB, and the UCSC genome browser, this SNV is highly conserved and disrupts multiple binding factor motifs. This SNV is located in the most prominent candidate regulatory element of ANK3 in a DNAse hypersensitivity site enriched with regulatory histone mark (H3K27Ac), both features often associated with regulatory elements. This regulatory element overlaps a large alternative splice coding exon, suggesting that the variant may operate via regulation of transcription levels or isoform. Replication and functional studies are warranted to further investigate the role of this variant.