Genetic and methylomic interrogation of brain cell-type shifts in autism, schizophrenia, and Alzheim
Chloe X Yap1,2,3,4, Daniel D Vo6,17,18, Matthew G. Heffel5, Arjun Bhattacharya7,8, Cindy Wen3,4,5, Yuanhao Yang1,2, Kathryn E Kemper2, Jian Zeng2, Zhili Zheng2, Zhihong Zhu2. Eilis Hannon9, Dorothea Seiler Vellame9, Alice Franklin9, Christa Caggiano10,11, Brie Wamsley3,4,11,12, Daniel H Geschwind3,4,5,11,12, Noah Zaitlen11,13, Alexander Gusev15,16, Bogdan Pasaniuc10,7,5,13,16,
Poster # 13
Most neuropsychiatric disorders lack clearly defined cellular or molecular markers. Previous studies investigating neuropathologic signatures of psychiatric diagnoses have often relied on small cohorts, discrete cell-type markers, and have been unable to disentangle cause from consequence. Furthermore, efforts to investigate brain cell-type proportion (CTP) shifts have been hampered by the expense and sampling bias of single-cell experiments and a focus on deconvolution of bulk RNA-seq, whose biological and statistical properties make it an inferior data type for CTP deconvolution. Here, we leverage advances in brain single-cell methylomics and develop a novel framework to deconvolve 7 brain CTPs from bulk DNA methylation data. We apply this framework to uniformly-processed bulk methylation data from 1,270 postmortem human brain samples, including donors diagnosed with autism (n=31), schizophrenia (n=186), and Alzheimer's disease (n=300). We observe subtle but global diagnosis-associated CTP shifts for Alzheimer's disease (endothelial cell loss), autism (increased microglia) and schizophrenia (decreased oligodendrocytes), with the former two robust to replication. There were also substantial sex- and age-related CTP shifts. We found statistically significant associations between endothelial cell loss and increased common variant risk for Alzheimer's disease, implying that endothelial cell loss may play a causal or pleiotropic role in Alzheimer's disease. In a genome-wide association study, we identify 5 loci significantly associated with cell-type compositional shifts, which in turn mapped to MYT1 (inhibitory neurons), CSF1 (astrocytes), SIPA1L2, GLRX5 and SHPK-TRPV1. These results systematically characterize cell-type vulnerability across neurodevelopmental and neurodegenerative diagnoses and provide a framework for investigation of cellular compositional shifts in the biology of neuropsychiatric traits.