De novo mutations in bipolar disorder implicate genes involved in neurodevelopment and immunity

Bipolar disorder (BD) is a debilitating disorder affecting ∼2% of the world's population. Genome wide association studies have identified over 100 common risk loci, a few copy number variants, and one gene with increased burden of loss of function (lof) mutations, but there have been few studies of de novo variation in BD and no genes can yet be considered genome-wide significant de novo hits. For the first time, this study explores the contribution of de novo mutations (DNMs) to BD and related conditions in multiplex families ascertained from genetically isolated populations. Exomes of 199 complete trios ascertained as part of the Amish Mennonite Bipolar Genetics (AMBiGen) study were sequenced at Regeneron, quality controlled and analyzed using GATK. DNMs were called using HAIL, followed by stringent sample and variant filters and Sanger sequencing of selected DNMs. Genes with DNMs in cases were combined with genes hit by DNMs among 354 previously-published BD trios for functional enrichment analysis using hypergeometric tests, topic modeling, and tests of cell-type enrichment. A total of 41l of and 239 damaging missense DNMs in 275 genes, including NRXN1, SHANK3, and SPECC1, were detected among individuals with BD and related disorders. Two genes have been reported in previous studies. The combined set of 458 genes was significantly enriched for functions related to learning, post-synaptic organization, nervous system development, and calcium ion transport. These genes significantly overlapped with brain co-expression modules associated with neurogenesis and immunity. These modules were significantly enriched in genes expressed in excitatory neurons, endothelial cells, and microglia. Although no single gene had a significant excess of DNMs, these findings suggest that DNMs in gene sets involved in neurodevelopment and immunity may contribute to BD. Future studies should focus on replicating these findings in additional samples. Genes with significant burdens of DNMs will be good candidates for functional genomic studies.