THE SHARED GENETICS BETWEEN BRAIN STRUCTURE AND PSYCHIATRIC DISORDERS: FROM GWAS TO DEVELOPMENTAL MECHANISMS

Abstract

Structural brain abnormalities at both macro- and micro-scale are highly heritable and have been commonly observed in neuropsychiatric conditions, and theorised to be endophenotypes. However, it is unclear to what extent they are attributable to medication, disease progression, or other sequalae that do not reflect disease aetiology. By concurrently studying the genetics of magnetic resonance imaging (MRI) phenotypes and neuropsychiatric conditions, and mapping to single-cell multiomics data, it is possible to prioritise overlapping cellular and developmental mechanisms between these phenotypes. We conducted GWAS on 11 T1- and diffusion-weighted MRI phenotypes in the UK Biobank (N=53,710) and ABCD (N = 4863) cohorts at both global and regional levels. Genetic correlation and polygenic score analyses identified a negative association between global and regional measures of brain size (surface area, volume, intrinsic curvature and folding index) and ADHD, depression and anxiety, which was robust across datasets and sensitivity analyses. Mendelian randomisation analyses confirmed causality from brain size to neuropsychiatric phenotypes and not vice-versa. Genomic structural equation modelling indicated that the overlap between brain size and depression/anxiety was no longer significant adjusting for ADHD, whereas ADHD was still correlated adjusting for depression or anxiety. On this basis, we hypothesised that brain size and ADHD overlap either in specific cell types, or in genetic processes that span multiple cell types. Multiple mapping methods between GWAS and single-cell RNA-seq in adult and developing brains identified enrichment of cortical size in neural progenitors, and ADHD in deep-layer excitatory neurons, but no overlapping cell-type enrichment. Given the enrichment in successive developmental stages, we identified putative gene expression programmes using consensus non-negative matrix factorisation (cNMF) and replicated using independent single-cell RNA-seq datasets. Overlapping enrichment was found in cNMF3, a programme active from intermediate progenitors (IPCs) to newborn excitatory neurons. Gene ontology analysis showed that both phenotypes and the cNMF3 factor were associated with axonogenesis-related processes. We further analysed the transcriptomic diversity of IPCs in an independent dataset, where cNMF3 factor is enriched for genes up-regulated in neuronal-like IPCs compared to early-stage IPCs, suggesting that it is critical for the transition from IPCs to excitatory neurons. Finally, we sought to quantify pathway-level overlaps between brain size and ADHD, and prioritise shared molecular targets. Leveraging transcription factor-target gene relationships derived from single-cell ATAC-seq in developing brains, we estimated the genetic correlation stratified by transcription factor using GNOVA, and prioritised several high-level regulator genes for cell fate determination, including TCF12, NFIB, and MEIS2. Notably, these transcription factors are highly expressed in the cNMF3 factor and their target genes are enriched for both ADHD and brain size. These findings provide a window into one possible cellular and developmental mechanism linking ADHD to overall brain size. Ongoing work will address whether this represents a subgroup of ADHD. The analytic methods used in this study will be translatable to further imaging-genetic studies identifying mechanistic overlap with psychiatric disorders.