There is no one gene that, when mutated, causes autism. But over the past decade, researchers have identified hundreds of gene variations that seem to affect brain development in ways that increase the risk of autism. However, these scientists mainly searched for variants in the DNA that directly encodes the building blocks of proteins. Now, a new study probing so-called noncoding DNA has found that alterations in regions that regulate gene activity may also contribute to autism. And surprisingly, these variations tended to be inherited from fathers who aren’t autistic.
“This is a really good article—it’s somewhat provocative and it makes us think about [autism genetics in a] different way,” says Lucia Peixoto, a neuroscientist and computational biologist at Washington State University in Spokane, who was not involved in the research. “I think it’s a great contribution to the field.”
Research into the genetic risk for autism has mainly focused on how mutations that arise spontaneously in an individual’s genome—rather than being inherited from a parent—disrupt protein-coding regions and lead to the condition. That’s because these sporadic mutations have relatively large effects and studies have shown that such mutations, although individually rare, together contribute to about 25% to 30% of cases, says Jonathan Sebat, a geneticist at the University of California, San Diego. But only about 2% of the genome consists of protein-coding areas. Sebat says the large noncoding portion of our DNA—often previously referred to as “junk DNA”—has so far been ignored in autism research.
Sebat’s team was especially interested in the parts of noncoding DNA that regulate gene expression. They looked at whole-genome sequences from 829 families that included autistic individuals, their nonautistic siblings, and their parents. Assessing the influence of individual DNA base changes is particularly difficult in noncoding regions, so they instead identified bigger alterations, so-called structural variants, in which large sequences of DNA are inverted, duplicated, or deleted.
Each individual has thousands of structural variants in their genome, so the researchers narrowed down their analysis to examine just a handful of regulatory regions where genetic variation seemed most likely to cause disruption. They chose these by finding regions where the general population has less variation than expected, suggesting that genetic changes there could be detrimental. These included sites involved in regulating gene activity during brain development and initiating the transcription of genes.
The scientists then examined whether structural variants in these regions were associated with autism by examining the pattern of transmission from parents to their autistic and nonautistic children. Researchers have assumed that mothers are more likely to pass on autism-promoting gene variants. That’s because the rate of autism in women is much lower than that in men, and it is thought that women can carry the same genetic risk factors without having any signs of autism. But when a mother passes these genes to her sons, they are not protected in the same way and thus will be affected.
The team found that mothers passed only half of their structural variants on to their autistic children—a frequency that would be expected by chance alone—suggesting that variants inherited from mothers were not associated with autism. But surprisingly, fathers did pass on substantially more than 50% of their variants. This suggests that autistic children might have inherited risk variants in regulatory regions from their fathers but not their mothers, the researchers report today in Science.
To check that this result held up, Sebat’s team then tested a second, larger sample of 1771 families. Once again, autistic children received more structural variants from their fathers but not mothers—though the size of the effect wasn’t quite as large in this second sample.
“This is completely opposite to … what we had previously assumed,” Sebat says. Peixoto finds the paternal bias surprising as well, although she already suspected that the inherited component of autism would be more apparent in noncoding regions. Compared with mutations in protein-coding regions, variants in regulatory regions usually have “smaller but additive effects. And when you have a smaller effect, you are much more likely to pass [it] along from generation to generation.”
Based on these results, Sebat proposes a more complex model of how autism arises, in which mothers pass on mutations affecting coding regions, which have large effects that women are protected from, while fathers pass on variants affecting noncoding regions; their effects are much more moderate and may only cause symptoms when combined with risk variants from mothers.
Dalila Pinto, a molecular geneticist at the Icahn School of Medicine at Mount Sinai in New York City, says the study provides “very insightful preliminary findings.” She said she will be interested to see whether the results are replicated in even larger genome databases—and whether additional variants will be identified. Peixoto agrees: Although the research is still at an early stage, she says, it “open[s] a door in a different direction.”