Most hunts for genes that might heighten the risk of autism involve tediously scanning the genomes of several families in which the disorder is prevalent and looking for multiple cases in which a single base in the genetic code has been substituted or deleted. Over the years, scientists have linked increased risk for the mental disorder—characterized by asocial and obsessive behavior—to genetic regions on 20 chromosomes. With so many different possible culprits, it is clear that there is not just one gene (or even just a few) that causes the illness now diagnosed in one of every 166 children born in the U.S.

"The challenge of finding genetic risk factors for autism spectrum disorders is the diversity of genes that are involved in cellular and morphological processes responsible for 'normal' human social skills and behavior," says Sabine Klauck, a molecular geneticist at the German Cancer Research Center in Heidelberg.

Now, a new study, led by a group at Cold Spring Harbor Laboratory (CSHL) in Long Island, N.Y., has a wider scope, viewing the genome as a whole and looking for copy number variations (sections of the genome, often spanning multiple genes, where there is an extra or missing piece) that could flag unstable regions, which may be risk factors in patients without a family history of autism. The report, which comprises research from 14 institutions and appears in Science, could help researchers better pinpoint their searches for autism-related mutations.

"Our results show conclusively that these tiny glitches are frequent in autism, occurring in at least 10 percent of cases, and primarily in the sporadic (nonfamilial) form of the disease, which accounts for 90 percent of affected individuals," says lead study author Jonathan Sebat, an assistant professor at CSHL.

Using high-resolution technology developed at CSHL, Sebat and his colleagues scanned the genomes of 264 families: 118 "simplex" families (those with only one child diagnosed with autism); 47 "multiplex" families (those with multiple autistic children); and 99 control families with no reported cases of autism. When mutations were found, they were checked against the subject's parents to make sure the change was de novo (newly arisen, or not found in either parent).

The team found 17 de novo copy number variations (CNVs)—affecting as few as one gene or as many as 69—in 16 individuals, 14 of them autism patients. The mutations were 10 times more likely to appear in patients who were part of a simplex family (12 of 118, or 10 percent of patients) and five times more prevalent in the multiplex patients than they were in the control families. This provides evidence that familial and sporadic cases of the disorder are indeed distinct, and that at least some sporadic autism instances result from CNVs—some 15 percent, according to the study.

"This is one of the first studies to examine families with only one child with autism," says Geraldine Dawson, director of the University of Washington's Autism Center "The fact that these de novo mutations were more frequent in these families suggests that they are contributing to autism."

Interestingly, the two variations observed in control subjects were duplications of genetic material, while the mutations in autism patients were all deletions. Four of the deleted CNVs impacted only a single gene. Among the genes affected by these mutations were ones previously tied to mental health disorders such as impaired speech processing, retardation and epilepsy.

Changes in the "structures of genes within a lesion could have quantitative effects on gene function," the authors write. "A genomic rearrangement may also disrupt regulatory elements that influence the expression of neighboring genes; thus, in some cases, a gene related to autism may lie adjacent to the deletion or duplication."

German Cancer Center researcher Klauck says the new study adds information necessary to "pinpoint mutation-prone genomic regions in parallel to other approaches like whole-genome association studies." The chromosomal regions identified here, he says, "are an excellent starting point for further fine-scale analyses." The University of Washington's Dawson agrees that these fragile regions likely hide some genes involved in autism. "The goal," she says, "is to eventually be able to identify infants at risk for autism at birth so we can begin treatment as soon as possible."