Genomic Landscape of Autism Spectrum Disorders

Background: Identifying genetic factors underlying ASD susceptibility is challenging due to the clinical heterogeneity in symptoms and disease severity exhibited by the affected individuals. Recent genome wide association and copy number variant studies have not only identified over 240 genes and genomic regions carrying variants that lead to ASD susceptibility, but have also shown that a significant portion of these variants reside in the intergenic regions suggesting the importance of regulatory polymorphisms in ASD. Some variants behave in a Mendelian fashion showing high penetrance whereas others exhibit low penetrance where the phenotypic effect of a mutation within members of the same family is different. Current findings support a model that common variants with a moderate effect together with rare and de novo mutations in genes involved in neuronal development and signaling underlie ASD susceptibility.

Objectives: (i) To identify rare functional coding polymorphisms (ii) To identify rare functional noncoding polymorphisms

Methods: We sequenced 18 autistic individuals using two platforms (i) Whole Exome (ii) Custom Capture: 3.5 MB of prioritized 100 ASD-candidate genes and their conserved elements.

Results: We identified ~24,000 SNPs in each individual on the exome platform, 95% of which were catalogued in dbSNP.  We performed systematic annotation using ANNOVAR to identify synonymous, missense and nonsense polymorphisms. We then used POLYPHEN2 and SIFT to study the effect of missense mutations on protein function. Finally, in order to identify rare SNPs we used data from 1000 genomes to obtain allele frequencies of potentially pathogenic variants (identified as ‘damaging’ by POLYPHEN2 or SIFT). Of all the SNPs, ~15,500 fall within exons and ~7,000 resulted in the change of an amino acid. Subsequent predictions using POLYPHEN2 and SIFT revealed ~2,000 SNPs that were potentially pathogenic, of which ~250 were ‘novel’. Similar analysis of coding SNPs in our prioritized 100 ASD-candidate genes identified ~30 missense mutations and revealed remarkable heterogeneity as each individual carried on average 8-10 rare (<2% allele frequency) potentiallly ‘damaging’ SNPs in genes such as DISC1, MAP2, CACNA1C, CACNA1H, MADCAM1, BZRAP1, HTR2A, CDH22 and OPRM1. To analyze the noncoding variants we utilized histone modification marks, specifically H3K9ac, and DNAse1 hypersensitive site (DHS) information from the human brain to score and prioritize important regulatory regions and SNPs within them. We identified ~2,700 noncoding SNPs of which ~1600 resided in potential regulatory regions marked by DHS and H3K9ac marks. We used a score threshold of 0.3 to identify  ~70 high scoring SNPs in genes such as NBEA, BZRAP1, PACRG, ABAT, NLGN1, DACH1, NRXN1, CDH22, CDH9, CADPS2, CADM1, CTNNA3.

Conclusions: Analysis of SNPs identified functionally important variants in key Autism genes, but most importantly, it highlights extreme genetic heterogeneity intrinsic to ASD. In addition to the observed coding variation we have also identified rare and novel SNPs in potential regulatory elements, including predicted p300 binding sites, experimentally validated neuronal enhancers. Understanding the predicted effects of these alleles on gene function and evaluation of these alleles in functional assays, individually and in combinations, will provide key insights into the mechanisms underlying neurodevelopmental anomalies associated with autism.