Large-scale genome-wide association studies (GWAS) have successfully identified hundreds of common variants associated with psychiatric disorders. Meanwhile, large-scale whole exome sequencing (WES) studies that test rare variants in aggregate have identified hundreds of genes for which loss-of-function is associated with increased risk of neurodevelopmental disorders. To understand the exact biological mechanisms in the human brain through which these genetic risk factors impart risk for disease, the first step in post-GWAS analyses is to identify genes that are regulated by disease-associated variants. Thereafter, the functional roles of disease genes from either GWAS or WES studies can be studied in an appropriate experimental system adequately recapitulating the disease-relevant context. In this dissertation, we primarily focus on this two-stage process of identifying genes and understanding their biological functions (herein referred to as the “variant-to-gene-to-function” problem). We approach this problem by integrating existing, large-scale genotype array and postmortem human brain RNA-seq data from PsychENCODE. The first chapter focuses on proof-of-concept isoform-resolution expression analyses, where we fine-map the most significant GWAS locus for autism spectrum disorder and implicate reduced expression of a specific isoform of the XRN2 gene as the underlying driver of GWAS signal. In this chapter, we also jointly model constituent isoforms of a given gene with multivariate variance components linear mixed models, enabling systematic dissection of the genetic architecture of isoform-level expression in the human brain, for the first time. The second chapter shifts its focus to characterizing the functional role of the high-confidence schizophrenia risk gene C4A in the human brain by investigating the effect of complex structural variation of C4 genes on gene expression and co-expression. By annotating changes in C4A co-expression, we find putative molecular correlates of synaptic pruning and convergence of schizophrenia polygenic effects in synaptic processes, indicating that neuronal and synaptic pathways are the driving force conferring schizophrenia risk. Altogether, this dissertation aims to refine and advance our understanding of mental illnesses by characterizing the neurobiological mechanisms through which known genetic risk factors contribute to psychiatric disorders.

Neuropsychiatric genome-wide association (GWAS) studies, including for autism spectrum disorder (ASD) and schizophrenia (SCZ), show strong enrichment for regulatory elements in the developing brain. However, prioritizing risk genes and mechanisms is challenging without a unified regulatory atlas. Here, I have characterized the functional genetic regulation in the developing human brain across gene, isoform, and local splicing features by systematically harmonizing 672 diverse samples. I identified 15,752 genes harboring gene/isoform-expression and/or splicing quantitative trait loci (xQTLs), mapping 3,739 to cellular contexts. I discovered that gene expression heritability drops during development, especially within the developmental window of 10 to 18 weeks post-conception. Through cell type deconvolution and heritability analysis in homogenously cultured progenitor and neuronal cells, I demonstrated that the drop of heritability likely reflects both the increasing cellular heterogeneity and the intrinsic properties of neuronal maturation. I also discovered that isoform-level regulation, particularly in the second trimester, mediated the largest proportion of neuropsychiatric GWAS heritability. Via colocalization, I robustly prioritized mechanisms for ~60% of GWAS loci across 5 neuropsychiatric disorders, exceeding adult brain findings. Finally, the genetic regulation in the developing human brain was contextualized within gene/isoform co-expression networks, revealing the comprehensive landscape of genetic regulation in development and disease.

Psychiatric disorders are not well understood. Their diagnosis is based purely on behavioral symptoms and they lack a clearly defined pathology in brain, which challenges our ability to understand their biological roots. However, it is well established that psychiatric disorders are heritable, and large-scale genetic studies have begun to identify now thousands of psychiatric genetic risk variants.1 Discovering how these genetic variants converge within discrete neurobiological pathways is a critical next step for understanding psychiatric disorder mechanisms and identifying new targets for therapeutic development. In search for these convergent pathways, transcriptomic studies have started to identify gene expression changes within human postmortem brain samples from psychiatric patients compared to neurotypical controls. The transcriptome – the set of expressed RNA transcripts present in a given tissue or cellular samples – represents a snapshot of the cell-types and subcellular, molecular processes present and active in sequenced samples. As such, transcriptomic profiling of brain samples from psychiatric cases versus controls may provide increased resolution to identify a molecular pathology of disease not observed via traditional approaches. For example, in ASD, upregulation of microglial, astrocyte, and immune signaling genes, downregulation of specific synaptic genes, and attenuation of regional gene expression differences have been observed with transcriptomic analyses.2,3 While transcriptomic studies have substantially improved our understanding of psychiatric neuropathology, they are limited in scope to single psychiatric disorders and few brain regions. Considering the growing evidence for genetic overlap between distinct psychiatric disorders,4 it is a reasonable next step to determine if these disorders also share biological signatures in the brain. Comparing and contrasting gene expression changes across distinct psychiatric disorders – as well as across the entire cerebral cortex - will provide a fuller picture of the spatial landscape and specificity of molecular dysregulation in the psychiatric disease brain, pinpointing potential regions of particular vulnerability and biological pathways involved in psychiatric disease mechanisms.

To obtain this cross-disorder and multi-regional understanding of psychiatric gene expression changes, here I present a comprehensive set of transcriptomic investigations conducted by myself and others, spanning multiple psychiatric disorders and brain regions. In Chapter 2, I share our published mega-analysis of gene expression microarray datasets containing frontal cortex samples from subjects diagnosed with schizophrenia, bipolar disorder, ASD, and major depressive disorder subjects, compared with non-psychiatric controls.5 We find that polygenic overlap parallels transcriptomic overlap, and that psychiatric genetic risk variants are associated with downregulated neuronal genes found in ASD, schizophrenia, and bipolar disorder. In Chapter 3, I present my contributions to our published collaborative work with the PsychENCODE Consortium,6 in which we compiled and uniformly processed genotype and RNA-sequencing data from more than 2,000 postmortem human brain samples to gain an understanding of how the entire transcriptome is impacted in frontal cortex samples from subjects diagnosed with schizophrenia, bipolar disorder, and ASD. Here, I detail my work integrating polygenic risk scores --measures of common genetic burden for psychiatric disease -- with transcriptomic changes to obtain a deeper understanding of how genetic variants directly regulate psychiatric gene expression changes. In Chapter 4, I present our work characterizing ASD transcriptomic across 11 distinct regions spanning the ASD cerebral cortex. We find widespread dysregulation across the cerebral cortex, with this dysregulation exhibiting the greatest magnitude of effect in the occipital region. ASD genetic risk variants are associated with genes downregulated cortex-wide that contribute to neuronal synaptic plasticity pathways, heavily implicating neuronal synaptic plasticity in ASD neuropathology. Together, these transcriptomic analyses expand our understanding of the molecular pathology of psychiatric disorders across distinct disorders and the cerebral cortex, implicating specific genes, cell-types, and biological pathways in psychiatric neuropathology.

Abstract Bibliography1. Sullivan, P. F. & Geschwind, D. H. Defining the Genetic, Genomic, Cellular, and Diagnostic Architectures of Psychiatric Disorders. Cell 177, 162–183 (2019). 2. I. Voineagu et al., Transcriptomic analysis of autistic brain reveals convergent molecular pathology. Nature. 474, 380–384 (2011). 3. Parikshak, N. N. et al. Genome-wide changes in lncRNA, splicing, and regional gene expression patterns in autism. Nature 540, 423–427 (2016). 4. Bulik-Sullivan, B., Finucane, H., Anttila, V. et al. An atlas of genetic correlations across human diseases and traits. Nat Genet 47, 1236–1241 (2015). 5. Gandal, M. J. et al. Shared molecular neuropathology across major psychiatric disorders parallels polygenic overlap. Science 359, 693–697 (2018). 6. Gandal, M. J. et al. Transcriptome-wide isoform-level dysregulation in ASD, schizophrenia, and bipolar disorder. Science 362, (2018).

Background

Methods

Results

Conclusions