THIS ARTICLE IS MORE THAN FIVE YEARS OLD
This article is more than five years old. Autism research — and science in general — is constantly evolving, so older articles may contain information or theories that have been reevaluated since their original publication date.
Analyzing thousands of sequences, researchers have homed in on miniscule portions of the genome that they say may be most crucial in determining autism risk. They presented the unpublished results Wednesday at the 2015 American Society of Human Genetics Annual Meeting in Baltimore.
The analysis will allow researchers to sift through thousands of missense mutations — those that alter a single amino acid in a protein — in people with autism.
Missense mutations are often harmless or have subtle effects. As a group, the missense mutations found so far are only marginally more common in people with autism than in controls. To find autism risk factors, geneticists typically focus instead on ‘loss-of-function’ mutations, which destroy a protein.
In the new work, scientists have identified areas adding up to 14 percent of the exome — the 1 percent of the genome that codes for proteins. They found that people with autism are three times as likely as controls to have a missense mutation in these regions. Outside of these refined regions, this difference goes away.
“For the first time, we can zero in on a small subset of missense variants that have a high probability of being relevant,” says lead researcher Mark Daly, associate professor of medicine at Massachusetts General Hospital.
Researchers may be too quick to assume a mutation is linked to autism because it’s de novo — that is, present in an individual with autism but not in his or her parents. But not all of these mutations will be risk factors, says Thomas Bourgeron, professor of genetics at the Institut Pasteur in Paris.
“Anything that can help to decipher whether a gene is pathogenic or not will be very helpful,” he says. “We have to be very ‘haute couture’ about the genes we look at, and try to see where are their conserved parts.”
The new work follows from a study last year that used data from roughly 6,000 sequences from individuals without autism to estimate the rate at which any one gene would be expected to be mutated by chance.
The calculation takes into account the baseline rate of mutation of a particular nucleotide trio — for example, AGA to ATA — in noncoding regions of the genome. It also accounts for gene length, as long genes randomly accumulate more mutations than short ones.
Daly and his colleagues then looked for loss-of-function de novo mutations in more than 1,000 families that have one child with autism but unaffected parents and siblings. They found 1,003 genes that carry more of these mutations than expected in people with autism and fewer than expected in the 6,000 controls1.
The new work takes this a step further by breaking up the genes into their protein-coding fragments, or exons. The exact location of a missense mutation within a gene can significantly affect whether the mutation is benign or harmful.
The researchers could dive deeper into this analysis than ever before because of the release last year of nearly 61,000 exomes from the general population, the largest-ever set of such sequences, by the Exome Aggregation Consortium (EXAC).
“We expect so many missense mutations in EXAC that we can begin to look at finer-grained patterns,” says Kaitlin Samocha, a graduate student in Daly’s lab who presented the findings.
The researchers applied their calculation to EXAC sequences and found nearly 6,000 genes that carry fewer missense mutations than would be expected by chance. In roughly one-quarter of these genes, the missense mutations are unevenly distributed, with some subregions particularly ‘constrained’ from containing missense mutations.
For example, one gene has only half of the expected amount of variation overall. But on closer analysis, they found that two of the gene’s exons carry 144 mutations compared with the 158 mutations expected, whereas the other two exons contain only 44 mutations compared with an expected 233.
These constrained regions together add up to 14 percent of the exome. They tend to be rich in harmful mutations. For example, a missense variant included in ClinVar — an aggregate of all variants found in people with some disorder — is roughly 30 times more likely to fall into one of these regions than elsewhere in the exome.
To explore the role of these regions in disorders, the researchers looked at a separate set of sequences from 5,622 people with autism, developmental delay, intellectual disability or epilepsy, and 2,078 controls.
They found that any mutation that is more common among people with intellectual disability or developmental delay than in controls falls into one of the constrained regions. They saw the same trend, but to a lesser extent, in people with autism or epilepsy.
The next step, says Daly, is to compile a list of the missense mutations that fall into these regions in people with autism. The list might point to new autism candidates that wouldn’t emerge from studies of loss-of-function mutations. “We’d like to end this work with actually identifying, for the first time, genes enriched specifically with missense variation,” says Daly. “I think the exciting thing is that we have an opportunity now to really find those.”
The massive size of the EXAC dataset is also helping to refine researchers’ understanding of de novo mutations. Roughly one-third of the de novo mutations found in people with autism are also present in EXAC.
This subset of de novo mutations is also no more common in people with autism than in their unaffected siblings. Jack Kosmicki, a graduate student in Daly’s lab, presented those unpublished findings in a poster Wednesday.
The observations suggest that this subset of mutations is not harmful and so has not been selected out of the genome by evolution, says Brian O’Roak, assistant professor of molecular and medical genetics at Oregon Health & Science University in Portland. O’Roak is not involved in this study, but has found a similar trend in his data. Certain sites in the genome may also be hypermutable and so more likely to populate with mutations over time, O’Roak says.
These sorts of findings help to further refine where in the genome researchers should focus their efforts, says Samocha. “It’s another way to say, if you see this in another person [who does not have autism], it can’t be that important to autism.”
For more reports from the 2015 American Society of Human Genetics Annual Meeting, please click here.