Epigenomics of ASD

Background: Current data suggest the etiology of autism spectrum disorder (ASD) is multifactorial, likely including genomic and epigenomic alterations.  The contribution of epigenetic variants to the molecular etiology of ASD is still unclear. We proposed that the limited success of ASD epigenetic studies to date is due to the limited numbers of epigenetically diverse ASD cases such studies analyzed.

Objectives: To optimize identification of DNA methylation variants (DMVs) relevant to ASD etiology, we planned to stratify ASD cases into subgroups according to clinical and genomic features.

Methods: Our ASD cases were stratified into subgroups using the genotype and/or clinical phenotypes such as somatic growth as secondary ascertainment features. We chose to study two syndromes, Sotos and CHARGE, because such individuals have high comorbid rates of ASD and harbor mutations in epigenes, ie genes known to regulate epigenetic modifications. Specifically, Sotos and CHARGE syndromes are caused by mutations in NSD1 and CHD7 respectively. They present with ID and contrasting growth phenotypes, ie overgrowth and undergrowth respectively. We searched for DMVs in DNA from multiple tissues (blood, fibroblast, or buccal).

We also studied nonsyndromic ASD patients secondarily ascertained by somatic overgrowth (n=50), as well as 50 age-, sex- and ethnicity-matched controls. 

DNAm was assessed using sodium bisulfite-converted DNA on the Illumina Infinium HumanMethylation450 BeadChip array, which interrogates >485,000 CpG sites. Hyper- and hypo-methylated regions in ASD were verified by sodium bisulfite pyrosequencing.

Results: We first studied patients with Sotos (n=38) and CHARGE (n=29) syndromes, finding highly specific and sensitive DNA methylation (DNAm) patterns or signatures in blood. Similar signatures were found in other tissues such as fibroblast or buccal cells for each syndrome. Notably, a number of the DNAm variants (DMVs) within these signatures are located in genes previously implicated in ASD etiology, eg PCDH in Sotos and FOXP2 and HOXA1 in CHARGE.

Our second analysis focused on nonsyndromic ASD cases with overgrowth. We identified a subgroup (n=18) with an epigenetic signature clearly distinct from Sotos. These cases were not clinically consistent with a known syndromic diagnosis. They were negative by targeted mutation analysis for several disorders (eg Fragile X, PTEN). Notably, ASD cases positive for PTEN mutations did not share this specific epigenetic signature. We are currently exome sequencing these 18 ASD patients to identify mutations, presumably in a novel ASD gene. 

Conclusions: We found that clinical and/or genomic stratification of syndromic or nonsyndromic ASD cases constitute successful strategies to optimize identification of epigenetic variants in ASD. We expect such epigenetic variants will prove to be functionally relevant as many overlap genes already implicated in ASD etiology by genomic studies.

The epigenetic signatures we have discovered have the potential to be translated into diagnostic tools, allowing for earlier detection of ASD susceptibility in some patients. Ultimately we expect that clinical phenotyping, and genomic/epigenomic variant profiling of ASD individuals will optimize our ability to provide early diagnosis, accurate prognosis and personalized therapeutic options.