Genes with Extremely High Penetrance for Autism Are Often Involved in Largescale DNA Remodeling

Background:   Rates of autism comorbidity vary across genetic syndromes, though underlying causes have been poorly addressed. While there has been much interest surrounding neurite and synapse involvement in autism’s etiology, studies suggest there may be earlier embryonic disturbances in genetic syndromes associated with autism that are a reflection of the associated gene dysfunction.

Objectives:   To determine whether genetics underlying monogenic forms of intellectual disability (ID) that are highly-comorbid with autism vary in comparison to IDs with low-to-no rates of autism comorbidity and in what ways that occurs.

Methods:   A group of 492 forms of ID were curated using the Online Mendelian Inheritance in Man (OMIM) database as well as the larger literature. ID’s were grouped according to comorbidity with autism: ID only, ID with highly-comorbid autism (ID/HCA), and ID with variable autism (VarAut). Minimum comorbidity cut-off for inclusion within ID/HCA was 20% for autosomal conditions and 30% in males for X-linked. Gene functions were assessed using a combination of Gene Ontology (GO) and UniProt/Swiss-Prot databases. These databases and The Human Protein Atlas (HPA) were used for confirmation of GO’s Cellular Component results. Analyses for GO, frequency of “regulatory” genes across groups and within ID/HCA subgroup, and frequency of transcription factors across the ID/HCA upper/lower comorbidity ranges were performed using proportions comparisons. Analyses concerning ID/HCA interaction with the core protein-protein interaction (PPI) network (Wnt, Shh, Ncor, SWI/SNF, Notch, Erk1/2, TGF-β/BMP) were analyzed using 2-tailed heteroscedastic T-tests with Bonferroni correction for multiple comparisons (α = 0.0167).

Results:   Compared to ID only and VarAut, ID/HCA genes exhibited an extreme enrichment in protein functions associated with epigenetic, post-transcriptional, and translational regulation (called “regulatory”) (p < 0.001 across all comparisons). Genes from this group were highly enriched in the GO term, “regulation of gene expression (GO:0010468)”. In agreement with this finding, when we looked at “regulatory” gene function, a subgroup that comprised approximately 2/3^rd of ID/HCA, more than half of genes within the upper range of comorbidity (median = 62%) were largescale DNA remodelers (i.e., heterochromatin remodelers, ubiquitin ligases, and methylation regulators); meanwhile, more than 2/3^rd of the lower range were transcription factors/repressors (p = 0.0216; 95% CI of difference of proportions: 0.0588, 0.7412). In addition, approximately 80% of the “regulatory” gene products were located primarily within the nucleus of the cell. Finally, in contrast to other forms of ID, gene products associated with ID/HCA centered around a core PPI network that is foundational to axial patterning of the CNS as well as later processes of neural maturation and plasticity (p < 0.001 across all comparisons). Of particular note to this project, genes associated with 19 of the ID/HCA conditions are not currently included within the SFARI Gene Database.

Conclusions:   These results indicate that high mutation penetrance for the autism phenotype is afforded most strongly by epigenetic, post-transcriptional, and translational regulators. Largescale DNA remodelers predispose towards highest penetrance, followed by specific transcription factors/repressors. In addition, our curated list suggests that the SFARI Gene Database is missing an important subset of highly-penetrant risk genes.