Skip to Main Content

A firm foothold in the genetics of autism

New study applies advanced genomics in a carefully assembled population to yield some of the first solid data in autism genetics

Matthew State
Matthew State led one of three new genomic studies that provide some of the strongest evidence to date on genetic causes of autism.

Matthew W. State, M.D., Ph.D., had never seriously considered a career in genetic research until 1995, when he spent a few months in a child psychiatry ward during his residency at the University of California, Los Angeles. There he cared for a few children with Prader-Willi syndrome, a rare genetic disorder caused when all or part of a seven-gene stretch of chromosome 15 is missing. Children with Prader-Willi tend to be intellectually delayed and prone to compulsive behaviors such as skin-picking and uncontrollable eating.

“These kids just had one region lost, but it led to a variety of psychiatric manifestations,” State recalls. “It made me start thinking, quite naively, about studying rare mutations as a way of understanding disorders that are much more common in the population.” The notion led State, at age 35, to enroll in the graduate program in genetics at Yale, where others were thinking along the same lines. Most famously, faculty member Richard P. Lifton, M.D., Ph.D., now chair and Sterling Professor of Genetics, had discovered a trove of new genes involved in hypertension by screening families with rare, extreme blood pressure disorders.

In recent years, State, now the Donald J. Cohen Professor in the Child Study Center, professor of psychiatry and genetics, and co-director of the Yale Program on Neurogenetics at the School of Medicine, has been employing the latest genomic technologies to take the same general approach to the notoriously diverse autism spectrum disorders. Studies have found dozens of genetic blips in people with autism, which is characterized by repetitive behaviors and problems in communication and social interactions, but researchers have struggled to sort out which of these gene variants are harmful and which benign.

State and several other groups around the country have attacked this problem by focusing on an unusual group of study participants known as the Simons Simplex Collection (SSC). Assembled and maintained with funding from the New York City-based Simons Foundation, the SSC includes blood samples and extensive medical information from 2,700 families in which one child has been diagnosed with autism. The families are recruited from 13 clinics across the country, each using the same precise measures to confirm autism diagnoses. Researchers collect blood samples from not only the children with autism, but from parents and unaffected siblings. The term “simplex” in the SSC’s name refers to the fact that the collection includes only families with a single child with autism and at least one unaffected child, making it easier to draw precise genetic comparisons between affected and unaffected family members.

In a much-publicized study published April 4 in Nature, State’s team screened about 10 percent of SSC participants. But rather than sequencing the entire genome of each participant, which is for now prohibitively expensive, the researchers focused on the “exome,” those portions of the genome that code for proteins. By applying this technique to the SSC’s unique population, the research fingered three genes that almost certainly contribute to autism, and about a dozen others that are strong contenders. Two accompanying papers in Nature turned up yet more candidates.

As the cost of genome sequencing continues to plummet, State and others will screen more and more SSC children. Based on previous work and the findings published in Nature, he estimates that the total number of autism-linked genes will hit 500 or more—a greater cause for optimism than it might seem. “Even though there are many, many regions in the genome involved, I believe they will converge on a small number of pathways,” State says. And, he adds, those specific networks will make the best targets for new treatments.

The SSC was launched in 2008 to help solve a puzzle. “We knew that children with autism don’t usually get married and don’t have children, and yet the rate of autism isn’t decreasing,” says Gerald D. Fischbach, M.D., scientific director of the Simons Foundation’s Autism Research Initiative. “So where is it coming from?”

One leading hypothesis was that some portion of autism cases are caused by spontaneous—so-called de novo—genetic variations, rather than those inherited from parents. A 2007 Science paper reported that 10 percent of individuals with autism carry de novo copy number variations (CNVs), stretches of DNA that are either deleted or duplicated, compared with just 1 percent of controls. Last year, State’s team and an independent group screened the SSC samples for de novo CNVs and confirmed that these variations crop up more often in children with autism than in controls.

But CNVs are large, sometimes encompassing thousands of base pairs of DNA and dozens of genes, making it difficult to draw conclusions about which individual genes are implicated in a disease. The new studies, in contrast, read families’ genomes letter-by-letter, so researchers can zoom in on variants affecting just one letter.

After screening 928 SSC participants, State and colleagues found 125 single-letter de novo mutations in the children with autism, compared with 87 such mutations in unaffected siblings. That’s a statistically significant difference, but also highlights a basic problem: Since healthy people carry so many mutations that don’t seem to do any harm, how do we know which ones in the autism group are important? “Sorting out the stuff that’s damaging from the stuff that’s not is extremely difficult,” State says. “One of the major preoccupations of the paper was finding an unbiased way of doing that.”

The researchers reasoned that a gene was more likely to be meaningful if it carried new mutations in more than one unrelated child with autism. They also placed special focus on mutations that are highly damaging to protein production. Once those criteria were plugged into the proper statistical analyses, from over 238 families studied the researchers ended up with just one gene, called SCN2A, that is strongly associated with autism.

Another group of researchers, led by Evan Eichler, Ph.D., of the University of Washington, did a similar search in a separate group of 677 SSC participants. When State’s group applied their statistical approach to the combined data from both studies, three genes made the cut: SCN2A, KATNAL2 and CHD8.

“Each one of these genes is a whopper,” Fischbach says. “Because of the design of these experiments, using both parents and unaffected siblings, this is about as sure as one can be about a genetic cause.” These gene candidates have intriguing biological roles in brain cells. SCN2A, for example, makes a protein that helps control the transmission of electrical impulses across nerve cell membranes; a handful of people with epilepsy or intellectual disability have been reported to carry glitches in the gene.

State, Eichler, and Michael Wigler, Ph.D., of the Cold Spring Harbor Laboratory are working on sequencing the rest of the participants in the SSC, and based on the mutation rate reported in the Nature papers, they estimate they will find at least 25 genes solidly linked to autism in the SSC alone. What’s still unclear, however, is how much of the broader population of children with autism may also carry glitches in these genes—or how many of these genes will also figure into other psychiatric disorders, such as schizophrenia, bipolar disorder, and obsessive-compulsive disorder.

There’s also the more mysterious issue of environmental and developmental contributions to developmental disorders. And all three Nature papers shed light on a factor that’s been much talked-about: paternal age. The data show that as the age of the father increases, so does the number of spontaneous mutations carried by his child, which makes sense since sperm tend to acquire more genetic mutations as a man ages. “It’s a very strong correlation—there’s no question,” State says. Still, a father’s age probably plays a minor role in overall autism risk. The average age of men when they fathered children with autism was only 1.1 years higher than when they fathered unaffected siblings. And when paternal age was left out of the analysis, children with autism were still found to carry more mutations than unaffected siblings.

But State sees a broader lesson to be learned from this finding. “It shows that as you get more insight at the molecular level, you can begin to understand what mechanisms might be behind some of these environmental influences,” he says. “As you move forward in clarifying the genetics, you’re also going to clarify the environmental factors, and vice versa.”