Two large-scale genetic analyses have turned up a trio of new sites associated with autism, including a large-effect allele that seems to reduce the risk of developing the debilitating brain disorder, researchers reported today (Nov. 12) at the American Society of Human Genetics meeting in Philadelphia.

Last year, the Autism Genome Project Consortium performed the largest genome-wide linkage scan to date with around 10,000 SNPs in 1,181 families with at least two affected individuals. The group flagged a handful of genomic regions harboring autism susceptibility genes, although none of the linkage results were statistically significant (Nat Genet, 39:319-328, 2007).

Now, a team led by Dan Arking, a geneticist at Johns Hopkins University, has ramped up the SNP count to include around 500,000 markers in 802 affected pairs of siblings. They then eliminated all the error-prone or uninformative SNPs to amass a collection of 180,000 high-quality markers for their analysis. "It's the cleanest best set of markers you can imagine," Arking said at a press conference.

This enhanced genome-wide scan proved effective. Arking's team discovered two regions of significant linkage that had not been implicated before with the disease -- one at the tip of chromosome 20's short arm, and one at the end of chromosome 6's long arm.

Arking, together with Lauren Weiss, a molecular geneticist at the University of California, San Francisco, also used the SNP dataset to perform familial association mapping in 1,594 parent-offspring trios to hunt for common variants of major effect linked to the disorder. At first, they did not find any genome-wide significant results. Additional assays, however, revealed a hitherto unidentified site on chromosome 5 where one particular allele was transmitted less often than expected to autistic individuals whose parents carried the allele. Thus, this allele, although only found in 4% of the population as a whole, likely confers some protection against autism, Arking and Weiss argued.

This "protective allele" fell near the semaphorin 5A (SEMA5A) gene, which is involved in axonal guidance during neural development. The researchers compared brain slices of 20 autistic individuals with 10 controls and found that SEMA5A had much lower expression levels in the autistic brains, further implicating this novel locus with autism.

Arking and Weiss will present their findings in a talk on Saturday (Nov. 15) and in a poster on Friday (Nov. 14).