08 Oct 2009

Semaphore may be a dying art, but the gene for semaphorin 5A (SEMA5A) could spark new life into the hunt for genetic causes of autism, which have been difficult to pinpoint. Results of a large, collaborative study reported in today’s Nature reveal a variation near the SEMA5A gene that associates with the disease. It is not clear how the single nucleotide polymorphism (SNP), which lies 80,000 base pairs upstream of the SEMA5A gene, relates to autism; however, semaphorin 5A is an attractive candidate gene because it plays a role in neuronal development and is downregulated in cell lines from autism cases (see Melin et al., 2006).

The finding comes from one of the largest genetic studies of autism to date. Researchers led by Mark Daly at the Center for Human Genetic Research, Massachusetts General Hospital, Boston, and Aravinda Chakravarti at Johns Hopkins University, Baltimore, Maryland, carried out both genomewide association (GWAS) and linkage studies on a large number of autism families. Samples from 780 families came from the Autism Genetic Resource Exchange (AGRE), a DNA repository and family registry, and 341 families from an NIMH repository. The primary GWAS covered just over 4,200 individuals and a total of 1,533 affected children.

Using Affymetrix platforms to identify 500,000 SNPs per sample, joint first authors Lauren Weiss and Dan Arking found that autism most strongly associated with eight different SNPs on six different chromosomes, but none of them reached the statistical threshold for genomewide significance. Nevertheless, the researchers re-analyzed the strongest candidates with additional data sets. SNP rs10513025 on chromosome 5p15 near SEMA5A turned out to be significantly associated with the disease in the combined analysis. Direct genotyping then confirmed that the nearby SNP rs10513026 also associated with autism. From the linkage analysis, the researchers identified a region on chromosome 20p13 that has genomewide significance, but there was little overlap between the strongest regions of SNP association. “This suggests that the regions of the genome showing linkage may harbour rare variation,” write the authors.

Whether the 5p15 SNPs influences SEMA5A expression remains to be determined, but Weiss and colleagues analyzed autism brain tissue samples and found that the gene was downregulated in 20 cases compared to 10 controls. “This is an attractive candidate gene given that its protein is a bi-functional guidance molecule, which is both attractive and inhibitory for developing neurons,” write the authors. Follow-up studies will be required to confirm these results and identify any functional genetic variation that contributes to autism.—Tom Fagan