NEW YORK (GenomeWeb) – Researchers have linked paternally inherited cis-regulatory structural variants to autism spectrum disorder.
While de novo protein-altering variants contribute to about a quarter of autism spectrum disorder (ASD) cases, researchers from University of California, San Diego, suspected that variants affecting regulatory elements could also contribute to risk of the condition.
UCSD's Jonathan Sebat and his colleagues searched for structural variants in cis-regulatory elements (CRE-SVs) within the whole genomes of more than 9,000 people from 2,600 families affected by ASD. As they reported today in Science, the researchers found that paternal-origin CRE-SVs were more likely to be inherited by their children with ASD rather than by their unaffected children.
"Our results suggest that rare inherited noncoding variants predispose children to ASD, with differing contributions from each parent," the researchers wrote in their paper.
The researchers' discovery set included 829 families, spanning 880 affected and 630 unaffected individuals, who had undergone whole-genome sequencing. These families had previously been screened for de novo loss-of-function mutations and large copy number variants using exome and microarray approaches, but those analyses came up empty. This, the researchers said, meant their cohort was more likely to contain novel and non-coding variants.
Using a pipeline they developed that included a support-vector machine-based software tool, the researchers uncovered an average 3,746 structural variants per person. For three individuals, the researchers, who included employees of Oxford Nanopore Technologies, performed nanopore sequencing to validate the structural variants.
Sebat and his colleagues also identified genomic regions that are generally intolerant to structural variations, reasoning that when they do occur in those spots, they would be more likely to be pathogenic. Within the discovery cohort, as compared to simulations, they found that structural variants were depleted within promoters and untranslated regions.
They then conducted family-based association testing to find that paternal-origin SVs affecting intolerant genes were more likely to be inherited by their children with autism than by those without it. Maternal CRE-SVs, meanwhile, were not associated with ASD.
Similar results were found, the researchers reported, in a cohort of 1,771 families from the 1000 Genomes Project. Further, a combined analysis of the 2,600 families also noted the association between ASD and paternal CRE-SVs.
Sebat and his colleagues found that some CRE-SVs recurrently affected intolerant genes like CNTN4, LEOP1, RAF1, and MEST. For instance, they uncovered two de novo loss-of-function variants that disrupted LEO1 — a gene previously implicated in autism and developmental delay — and both those LEO1 deletions led to the elimination of an upstream regulatory element.
They also found de novo mutations within their discovery sample, including 104 deletions, 19 duplications, eight complex structural variants, and 23 mobile element insertions.
The researchers said that the paternal origin effect they uncovered was surprising, especially as previous autism studies have noted a maternal bias in the transmission of truncating variants.
According to Sebat and his colleagues, there are three possible scenarios that could explain the effect they observed. First, they said it could be due to a "bilineal two-hit model" in which the proband inherits a variant of large effect from the mother and a CRE of moderate effect from the father to contribute to disease. Or, the paternal origin effect could be due to epigenetic mechanisms, though the researchers said that was unlikely. Lastly, they said it could be due to a meiotic drive, in which allele-specific selection occurs differently in paternal and maternal germ cells, a scenario they also deemed unlikely.