Large-Scale Exome Analyses Reveal Novel ASD Genes Impacted By Genetic Variation at All Scales

Background:   The genetic architecture of autism spectrum disorder (ASD) involves the interplay of common and rare variation and their impact on hundreds of genes. During the past two years, large-scale whole-exome sequencing (WES) has proved fruitful in uncovering risk-conferring variation, especially when considering de novo variation, which is sufficiently rare that recurrent de novomutations in a gene provide strong causal evidence.

Objectives:   The Autism Sequencing Consortium (ASC) aims to uncover a large fraction of genetic factors underlying autism and to identify genetic variation at all scales, including single-nucleotide variants (SNVs), indels and larger structural variants (SVs) such as copy number variants (CNVs).

Methods:   Results to date indicate that WES will identify ASD risk genes as a linear function of sample size, until samples reach many thousands of cases. Therefore, the ASC intends to analyze large ASD cohorts for ASD genes. Unlike earlier WES studies, we do not rely solely on counting de novo loss-of-function (LoF) variants, rather, we use novel statistical methods to assess association by integrating de novo, inherited and case-control information for LoF SNVs/indels, missense SNVs predicted to be damaging, and SVs.

Results:  Our first analyses on 3,871 ASD cases and 9,937 ancestry-matched or parental controls revealed 33 autosomal genes at a false discovery rate (FDR) < 0.1, and a broader set of 107 autosomal genes strongly enriched for those likely to affect risk (FDR < 0.30). Our dataset is enriched for genes targeted by two autism-associated RNA-binding proteins (FMRP and RBFOX), genes found with de novo non-synonymous mutations in schizophrenia, and genes encoding synaptic components. Amongst critical synaptic genes found mutated in our study are voltage-gated ion channels, including those involved in propagation of action potentials (e.g., the Na⁺ channel Na_v1.2 encoded by SCN2A), neuronal pacemaking, and excitability-transcription coupling (e.g., the Ca²⁺ channel Ca_v1.3 encoded by CACNA1D). Our dataset is also enriched for chromatin remodeling genes, including enzymes involved in histone post-translational modifications, especially lysine methylation/demethylation, and regulators that recognize such marks and alter chromatin plasticity such as the emergent ASD gene CHD8. More recently, we have looked at SVs in the dataset, especially those hitting an individual gene or portion of a gene (which we term exon dosage variants, or EDVs). We show that we can reliably call SVs and EDVs from WES data and results to date indicate the EDVs are showing a significant contribution to risk. Additional, ongoing analyses examine the ASC data for recessive genes and X-linked genes. Furthermore, the ASC data is being analyzed for evidence of mosaicism.

Conclusions: Our studies have identified a group of 107 high-confidence risk genes that incur de novo LoF mutations in over 5% of ASD subjects and have exposed two tightly intertwined pathways - chromatin remodeling and synaptic development – as major themes in ASD risk. These findings are being further extended by evidence on SVs, either deletions or duplications, disrupting discrete genes. Ongoing analyses of the data are identifying recessive and X-linked loci, and evidence for mosaicism in ca. 5% of the ASD samples.