Evaluation of Copy Number Variations In Autism Spectrum Disorders

Background:   The genetic landscape of autism risk is far from established. However, recent studies have demonstrated a consistent correlation of rare copy number variations (CNVs) with ASDs.

Objectives:   Given the extremely heterogeneous nature of ASDs, independent studies are required to identify novel candidate genes and refine previously identified genes and regions for further clarification of their potential pathogenic effect. 

Methods:   A genome-wide SNP-array was utilized for CNV detection in an ASD case-control dataset with samples of European ancestry. A total of 813 unrelated ASD cases and 592 unrelated healthy pediatric controls survived stringent sample and intensity quality control criteria. High confidence CNVs were identified through two distinct CNV-calling algorithms. 

Results:   Our results demonstrate a significantly heavier burden of deletions in cases as compared to controls, indicating that cases carry more and larger CNVs on average and that these CNVs are more likely to disrupt genes and have frequencies of less than 5%. A size-based examination reveals deletions larger than 1Mb were exclusively detected in cases, implicating novel CNV regions at 10p15.2, 13q12.12, 13q33.1, 14q23.2-3, and 17p12. CNV region- and gene-based association analyses consistently indicated the involvement of the regions containing ASH1L, NCOA6, OMG, and SETDB1 in ASD risk.  Gene set enrichment analysis suggested a possible link of REACT_383 [DNA replication], KEGG: hsa04612 [Antigen processing and presentation], KEGG: hsa04514 [cell adhesion molecules] and GO:0003713 [transcription coactivator activity] in ASDs. Examination of 472 rare case-specific de novo CNVs revealed several novel candidate genes within several previously implicated pathways, including the ubiquitin signaling pathway (ITCH, UBAP2, UBE2B, UBR2,and USP3), neural development (ANK2, NFASC, NMU, NRG4 and OMG), synaptic transmission (RIMS2), and members of the methyl-CpG-binding domain family (SETDB1 and SETDB2).  

Conclusions:   Cases tend to have a heavier mutational burden, mainly from genic CNVs with rare to low frequencies. Another five potential pathogenic novel large deletions were identified in ASDs.  The study further supports the potential involvement of previously implicated signaling pathways and reveals several new genes in these pathways.