ASD-Associated Genomic Variants in 16p11.2 and CHD8 Exhibit Clinically and Biologically Functional DNA Methylation Signatures

Background: One of the greatest challenges in studying autism spectrum disorder (ASD) is the degree of etiologic heterogeneity. Over 200 ASD-risk genes have been identified, but each gene accounts for <1% of all ASD cases and genetic causes have been identified in only ~25% of cases. A role for epigenetic dysregulation in ASD etiology is supported by the many ASD-risk genes that function as epigenetic regulators. We and others have proposed that aberrant epigenetic mechanisms resulting from genetic and/or environmental influences may alter biological pathways important for normal brain development. Previous studies have identified alterations in DNA methylation (DNAm), the most commonly assessed epigenetic mark, in ASD patients. However, the impact of these findings is limited by small sample sizes and inconsistent results across studies.

Objectives:  1) To investigate the role of epigenetic dysregulation in ASD. 2) Improve the discovery of DNAm differences by examining more homogeneous groups of individuals substratified based on known ASD-associated genomic variants.

Methods:  Genome-wide DNAm was measured using the Illumina Infinium HumanMethylation450 BeadChip array in DNA extracted from whole blood for all groups. First, we compared DNAm in a heterogeneous ASD group (n=52) with age- and sex-matched neurotypical controls (n=30). Second, DNAm was also assessed from patients with genomic variants conferring an increased risk for ASD: 16p11.2 deletions at the 600kb risk locus (16p11.2del; n=9) or heterozygous loss-of-function mutations in a chromatin modifier, chromodomain helicase DNA binding protein 8 (CHD8^+/-; n=7). These groups were compared with age- and sex-matched controls (n=23 and n=21, respectively). We used a modified version of our laboratory’s bioinformatics pipeline employing hierarchical clustering, principal components analysis and non-parametric statistical comparisons to identify significantly differentially methylated CpG sites.

Results:  Although no significantly differentially methylated sites were identified to fully distinguish heterogeneous ASD cases from controls, unique DNAm patterns were identified for the 16p11.2del and CHD8^+/- groups when compared with controls. These DNAm patterns consist of specific sets of differentially methylated sites (adjusted p<0.05, absolute difference ≥5%) comprising novel DNAm signatures. These signatures separated 16p11.2del and CHD8^+/- individuals from both heterogeneous ASD and control groups with high sensitivity and specificity. They also accurately classified 16p11.2 CNV and CHD8 sequence variants of unknown significance, distinguishing between pathogenic and benign mutations. Examining the genes in each signature revealed biological pathways that could be important to the pathophysiology of ASD, overlapping with known ASD-risk genes. Furthermore, some of the genes identified in our CHD8^+/- blood DNAm signature overlapped differentially expressed genes in CHD8^+/- (CRISPR/Cas9) human induced pluripotent stem cell-derived neuronal precursor and neuronal cells, demonstrating the cross-tissue functional significance of our CHD8^+/- epigenetic signature.

Conclusions:  Our approach constitutes a novel and clinically applicable method of molecular classification for ASD, identifying DNAm biomarkers in an easily accessible tissue. These findings elucidate the etiology of subgroups of ASD in the context of the crucial cross-talk between genetics and epigenetics. Combined, these data will enhance our understanding of the underlying biological mechanisms of ASD and facilitate the identification of novel therapeutic targets to facilitate precision medicine-based treatment options.