ASD Gene Discovery with Whole-Exome Sequencing

Background: ASD is a highly heritable disorder, however ASD gene discovery has been hindered by the extensive locus heterogeneity. De novo copy number variation (CNVs) and de novo loss of function (LoF, i.e. nonsense, canonical splice site and frameshift) variants have been associated with ASD through the observation of a higher rate of such variants in cases than unaffected controls.

Objectives: To identify ASD-associated genes through whole-exome sequencing; to use these genes to inform on the underlying etiology of ASD.

Methods: We present the analysis of 300 new ASD families from the Simons Simplex Collection analyzed with whole-exome sequencing as well as a combined re-calling and re-analysis of previously published ASD exome data from 965 families, yielding a total of 1,265 families. The analysis focuses on multiple de novo mutations clustering in the same gene in unrelated probands. Statistical thresholds for genome wide significance were established using previously described methods, taking into account the type of mutation (LOF vs missense etc), the sequence context including GC content, and gene size.

Results: This study is ongoing: at the time of submission a total of 147 genes were observed to have at least one de novo LoF variant in an affected individual. Based on the distribution of de novo LoF variants in the unaffected siblings, 60% of these genes are expected to be associated with ASD. Five of these 147 genes had two de novo LOF mutations (p= 0.04 and q=0.006) and four showed three independent LoF variants (p=0.0001, q=0.0002). Furthermore, this observation of nine ASD-associated genes with at least two hits, and the time course with which they were identified, is consistent with a model of 1,000 genes contributing to ASD causation. A further four genes were identified as being highly likely to contribute to ASD causation based on the additional information derived from inherited LoF and de novo ‘probably damaging’ missense variants. Analysis of the 147 de novo LoF variants with the DAVID functional annotation tool yielded a single biological process that survived correction for multiple comparisons: chromatin organization (p=0.04).

Conclusions: Whole-exome sequencing provides a means to reliably identify genes that carry large risks for ASD.  Analysis of the 147 de novo LoF variants with the DAVID functional annotation tool yields a single, statistically significant biological process, chromatin organization, implying that direct analysis of gene regulatory and expression networks could provide important new insights into the pathophysiology of ASD. While this study supports prior estimates of extensive locus heterogeneity, functional enrichment results comport with prior data suggesting that these genes will likely converge on a much smaller number of relevant molecular mechanisms.