T8. DISENTANGLING CAUSAL AND CORRELATED IMMUNE SIGNALS IN PSYCHIATRIC DISORDERS USING MULTI-TISSUE FUNCTIONAL GENOMICS

Background Psychiatric disorders and immune dysregulation often go hand-in-hand, as shown by epidemiological and genetic evidence. There are several competing hypotheses for these associations: (1) inflammation plays a role in the development of, or nature of, psychiatric disorders; (2) inflammation is a response to psychiatric symptoms or treatment (3) immune-associated signals are overrepresented due to the greater accessibility and characterisation of immune relative to brain tissue, or (4) immune associations are proxy signals of cross-tissue processes which act to cause pathology in the brain, but are detectable in immune cells. To address these competing hypotheses, it is critical to consider functional genomic immune and brain datasets in parallel. Methods We triangulate multiple functional genomic approaches to assess whether the immune system plays a causal role across 29 psychiatric, neurological and auto-immune traits, including psychiatric cross-disorder factors. We integrate GWAS data with immune and brain epigenetic data, as well as gene and protein expression regulatory data (bulk and single cell QTLs) from foetal and adult brains and resting and stimulated immune cells. We use methods including heritability enrichment within epigenetically-defined regions to prioritise relevant cell types, and use an integrated colocalisation-Mendelian randomisation pipeline to estimate which genes in which tissues likely play a causal role. In contrast to conventional high-throughput approaches, we incorporate subthreshold QTL signals in non-implicated tissues to improve inference about tissue-specificity. The therapeutic relevance of prioritised gene-cell type pairs was considered using several drug repurposing frameworks. Results Epigenetic information suggests a role for lymphoid cells in schizophrenia, bipolar disorder and the cross-disorder schizophrenia/bipolar factor, even after correcting for shared elements with the brain, but lymphoid tissues were not prioritized for major depressive disorder. Integration of GWAS summary statistics with molecular data identified multiple gene-tissue pairs causally implicated in psychiatric pathogenesis acting in the brain and immune system, with single-cell datasets improving discovery beyond bulk brain and bulk blood QTL datasets. Strikingly, the majority of prioritized genes were not cell-type specific, with most genes colocalising in both a brain and an immune tissue, although several brain-specific and immune-specific genes were identified, and several genes were prioritized exclusively in stimulated immune cells. Integration with drug target and signature databases identified candidate drugs. Discussion Our findings suggest there is unlikely to be a single shared causal mechanism linking inflammation to psychiatric conditions. The observed immune-specific and stimulation-dependent signals support a contributory role for inflammation to some psychiatric conditions, particularly schizophrenia-bipolar spectrum disorders. However, the extensive sharing of regulatory architecture across tissues cautions against interpreting apparent immune associations without investigation of genetically linked effects acting via the brain. These results emphasise the importance of integrative multi-tissue functional genomic approaches for disentangling causal biology and prioritising therapeutically relevant targets in psychiatric disorders.