Combined Analysis of Exome Sequencing Points Toward a Major Role for Transcription Regulation During Brain Development in Autism

Background: Four recent studies of the coding regions of the human genome (the ‘exome’), suggest that new (de novo) mutations in hundreds of genes may contribute to the risk of autism spectrum disorder (ASD). While the experimental strategy in the different efforts is almost identical, the four studies were published independently, and no integrative analysis has yet been reported. Notably, limited conclusions regarding the specific systems of genes disrupted by de novo mutations can be drawn based on each study alone. This stems from the relatively small fraction of mutations identified in each study in which there is a clear functional phenotype at the protein level.  

Objectives:  We performed an integrative meta-analysis of the four studies, to uncover systems of genes affected by de-novo exonic mutations in ASD.

Methods:  We first focused on genes containing nonsense, frameshift, or splice site de-novo mutations. To characterize the genes, we analyzed the enrichment of cellular processes and gene ontology (GO) using the Database for Annotation, Visualization and Integrated Discovery (DAVID). We then used a published dataset of brain gene expression throughout different life stages, to cluster the genes based on their expression during the developmental stages of the human brain. To cluster the genes we performed a Weighted Gene Co-Expression Analysis (WGCNA). We then broadened our scope by including non-synonymous substitutions, and performed a protein-protein interaction analysis using the Disease Association Protein-Protein Link Evaluator (DAPPLE).

Results:  Among the genes with disruptive mutations, we found a significant enrichment for “chromatin regulator”. This enrichment was significant compared to a large control exome sequencing cohort, as well as compared with the silent mutations in the same individuals, strongly supporting its specificity to ASD. When clustering the genes based on their expression during brain development, the chromatin regulator genes were mostly clustered in a large module of genes which is strongly expressed prenatally, with a sharp decrease in expression after birth. In contrast, silent mutations were significantly less likely to be in genes highly expressed prenatally. Finally, to test the interactions at the protein level, we included the non-synonymous mutations, and performed a protein-protein link analysis using DAPPLE. This analysis found a significant connectivity between the genes, and identified a list of strongly interacting genes, which was also highly enriched for genes involved in chromatin regulation.

Conclusions:  While it has been proposed that the origins of ASD are at the synapse, our meta-analysis of de novo mutations shows that the many of the recently identified mutations are in genes that are involved in transcriptional regulation, specifically chromatin related proteins, which are active during brain development. These findings, together with the association of other genes in this category with autism and intellectual disability, highlight the need to further study this type of genes as risk factors for ASD.