Combining Autism and Intellectual Disability Exome Data Yields Insight into Both Disorders

Background:   Autism spectrum disorder (ASD) is known to have a complex genetic architecture. Recently, exome sequencing studies have enjoyed great success using de novo single nucleotide variants (SNVs) and short indels to identify risk genes. However, calculations indicate that there are hundreds of additional genes that confer risk yet to be discovered.

Objectives:  ASD and intellectual disability (ID) are known to co-occur and consequently their risk genes show considerable overlap. These findings motivate an analysis that contrasts and combines discoveries in ASD and ID.

Methods:   The Autism Sequencing Consortium (ASC), in collaboration with the Deciphering Developmental Disabilities (DDD) consortium, assimilated published exome sequencing data from 6564 probands and controls. We used TADA (Transmission And De novo Association) to identify likely risk genes in both disorders individually and we combined the trios to find additional genes that affect risk for both disorders. We applied the algorithm DAWN (Detecting Association With Networks) to networks estimated from brain gene expression data to discover subnetworks of interacting risk genes. We then determined which gene networks communities showed enrichment of ASD and ID risk genes.

Results:   Our results revealed 31 and 64 risk genes (FDR < .05) associated with risk for ASD and ID, respectively (with 12 genes identified in both), and 16 additional genes when combining the data. Based on the pattern of de novo loss-of-function (dnLoF) variants, we estimated that the total number of autosomal dominant genes in which a dnLoF imparts substantial risk for ASD and ID, respectively. We estimated a 95% confidence interval for ASD of 500-950 and for ID of 185-225. Relative to the mean NVIQ in the sample, a proband with LoF mutation in one of the genes with the strongest signal for both disorders (ASD.ID) had an average drop in NVIQ of 24 points and a LoF mutation in an ID-related gene produced a similar reduction in NVIQ in the ASD proband. A LoF mutation in an ASD-related gene, which was not also implicated in ID, was associated with a weaker effect on NVIQ, reducing it relative to the sample mean by almost 12 points. We identified two functional ASD clusters enriched for chromatin modification. We used those components to identify which targets are enriched for ID-related risk genes. Genes strongly affecting cognitive function were primarily represented in clusters involving chromatin modification, implicating gene regulation as a mechanism shared by ID and ASD. Moreover, the results highlight disruption of corticofugal projections neurons as a source of risk.

Conclusions:   We estimated that there are significantly fewer ID-related genes and that a large fraction of the ID-related risk genes are also ASD-related. Genes found in ID samples had a marked impact on IQ even in ASD, compared to genes identified only in ASD. Genes strongly affecting cognitive function mapped to ASD gene networks involving chromatin modification. The results implicate disruption of neocortical development as a mechanism shared by ID and ASD.