Molecular cascades and cell-type specific signatures in ASD revealed by single cell genomics

Abstract

Understanding how genetic variation exerts its effects on the human brain in health and disease has been greatly informed by functional genomic characterization. Studies over the last decade have demonstrated robust evidence of convergent transcriptional and epigenetic profiles in post-mortem cerebral cortex from individuals with Autism Spectrum Disorder (ASD). Here, we perform deep single nuclear (sn) RNAseq to elucidate changes in cell composition, cellular transcriptomes and putative candidate drivers associated with ASD, which we corroborate using snATAC-seq and spatial profiling. We find changes in cell state composition representing transitions from homeostatic to reactive profiles in microglia and astrocytes, a pattern extending to oligodendrocytes and blood brain barrier cells. We identify profound changes in differential expression involving thousands of genes across neuronal and glial subtypes, of which a substantial portion can be accounted for by specific transcription factor networks that are significantly enriched in common and rare genetic risk for ASD. These data, which are available as part of the PsychENCODE consortium, provide robust causal anchors and resultant molecular phenotypes for understanding ASD changes in human brain. One-Sentence SummaryWe define the molecular cascades and cells disrupted in post-mortem brain in ASD by performing spatial, single nuclear RNA, and epigenetic profiling, and characterize, at unmatched resolution, the functional regulation of cell-type specific signatures underlying the molecular differences and physiology of ASD. Main TextPsychiatric disorders are defined primarily by behavioral and cognitive characteristics and are classically distinguished from neurological disorders by lacking the associated visible histological or macroscopic pathology observed in neurological conditions. However, a growing body of evidence based on genomic profiling reveals consistent molecular differences in brain tissue from specific neuropsychiatric conditions compared with brain tissue from neurotypical individuals (1-5). In Autism Spectrum Disorder (ASD) robust transcriptomic and epigenetic alterations in the cerebral cortex from patients have been documented over the last decade, delineating a reproducible pattern of molecular pathology (5-13). This robust molecular signature obtained primarily from transcriptomic profiling of bulk cortical tissue has identified convergent biological pathways in ASD brain, which is characterized by an upregulation of immune signaling genes, downregulation of specific neuronal markers, synaptic genes, and an attenuation of the typical patterns of gene expression associated with cortical regional identity (6-8,12-14). These genomic data represent an essential lens through which to understand the cellular and physiological changes occurring in the brains of autistic individuals and to describe potential causal mechanisms via their integration with genetic risk variants (1,4,5). However, small sample sizes in the case of single cell analysis (13), or profiling restricted to bulk tissue have limited biological insights as to the differences in laminar, circuit level, and cell-type specific pathways affected in ASD, as well as their underlying gene regulatory mechanisms. To address these limitations, we leveraged improvement in single cell analyses to profile the largest ASD cohort to date, consisting of 64 cases and controls. This resource, generated as a core component of the PsychENCODE consortium (1,2,4,5,8; http://www.psychencode.org), also enables us to characterize underlying candidate regulatory mechanisms and to connect causal drivers with the observed changes at a cell-type specific level in ASD, providing a deeper and more generalizable understanding of the cell types and biological mechanisms that underlie ASD. These data are available via PsychEncode portals for download (http://psychencode.synapse.org) and on the PsychSCREEN browser (in development, http://psychscreen.beta.wenglab.org).