Screen for autism risk genes finds two new candidates

Scientists have found a handful of genes — including two that had not previously been associated with autism — that may increase risk of the disorder, according to a study published today in Molecular Psychiatry ¹.

Using a strategy called high-density genetic association screening, the international team of researchers conducted an exhaustive search of every possible gene in two chromosomal regions previously implicated in autism susceptibility. But they did not find any common genetic variants that could account for the risk of developing the disorder in a majority of the autism families tested.

Rather, the team found weak genetic signals, present in only a few of the families with autism spectrum disorders. Although the genes found warrant further investigation, this study further strengthens the notion that no single gene makes a person susceptible to autism spectrum disorders.

“Our work adds to the increasing evidence that autism is due to many genes of small effect, as well as rare variants that increase susceptibility in small numbers of cases,” says lead investigator Anthony Monaco, head of neurodevelopmental and neurological disorders at the Wellcome Trust Centre for Human Genetics in Oxford, U.K.

Researchers studying the complex genetics of autism have so far completed 13 whole-genome linkage scans, which trace the transmission of risk genes in affected families².

Although these studies have not consistently confirmed a particular chromosomal region, several of them implicated a region located on chromosome 7, called AUTS1, and another on chromosome 2, dubbed AUTS5.

Six years ago, the International Molecular Genetic Study of Autism Consortium, a collective of two dozen autism research groups worldwide, used the linkage results to launch an in-depth search for autism risk genes in these chromosomal regions.

Pinpointing risk:

The team screened more than 6,000 single-nucleotide polymorphisms (SNPs), or individual DNA sequence variations, in individuals from 127 families most likely to harbor genetic risk variants in AUTS1, 126 families with potential for genetic risk variants in AUTS5, and 188 control families.

The researchers found that an increased risk of developing autism is associated with variations in two genes on AUTS1, known as IMMP2L and DOCK4, neither of which had previously been implicated in autism. But these variants are not common enough to be a robust signal of disease risk in larger studies of the general population, in which the overall effect of each genetic variant is diluted by the presence of the others.

Even weak signals can point to potential candidate genes, however, and could help uncover common pathways that lead to the development of such a heterogeneous disorder.

“We still know very little about the genetic causes of autism. In that regard, the unusual suspects can be more informative than usual suspects,” says Aravinda Chakravarti, a professor of molecular biology and genetics at Johns Hopkins University in Baltimore.

The usual suspects, in this case, are any SNPs of genes previously implicated in autism susceptibility. Even though SNPs make up 90 percent of all human genetic variation, they are evolutionarily stable — which makes them easy to trace through populations.

Linkage studies identify and narrow the chromosomal regions that predispose a person to a disease by exploiting the tendency for nearby DNA segments to be inherited together. The strength of the linkage signal, which is quite high for AUTS1 and AUTS5, reflects the degree to which identical DNA in these regions is inherited in pairs of siblings with autism.

“By looking at the SNPs, we’ve done a thorough study of common variation in this region,” says Alistair Pagnamenta, a postdoctoral fellow in Monaco’s lab. “Unfortunately, none of these have a large enough effect to fully account for the original linkage signal highlighting the AUTS1 and AUTS5 chromosomal regions.”

Finding all of the genetic aberrations that account for a linkage signal is notoriously difficult — especially for complex spectrum disorders such as autism. Genes linked to autism typically have such a low prevalence across study groups that large sample sizes are needed to tease out an effect.

Still, the genes with the strongest association are interesting. The IMMP2L gene is part of a complex that forms mitochondrial proteins. One research group reported in 2001 that the gene is disrupted in an individual with Tourette syndrome, a complex neuropsychiatric disorder that, like autism, affects behavior, social interaction, movement and language³.

The DOCK4 gene is perhaps even more intriguing because a 2008 study in which researchers deleted the gene showed that it is needed to form dendrites, the branching fibers that transmit neuronal signals⁴.

Alternatively, the linkage signal could be due to a number of different genetic abnormalities that accumulate to make a chromosomal region significant.

Rita Cantor, a professor of human genetics at the University of California, Los Angeles, says the worst-case scenario for researchers trying to trace autism through families would be if every family had a unique genetic story. “We can’t identify the risks statistically because there are not enough individuals with the same story,” Cantor says.

Multiple approaches:

Because of the increasing interest in the role that rare deletions or duplications of DNA — called ‘copy number variations’ (CNVs) — may play in autism⁵, the researchers also looked for these structural variants in the AUTS1 and AUTS5 regions.

They found CNVs in the region containing IMMP2L and DOCK4 in four of the families — which could also have contributed to the linkage signal for this chromosomal region.

Much larger sample populations will be needed to detect the genetic abnormalities — whether genes, CNVs, or other rare genetic variants that have only a small effect — that make individuals susceptible to autism, says Pagnamenta.

For instance, a study last year found three structural variations associated with schizophrenia, but only after a genome-wide search comparing 1,433 schizophrenia cases with 33,250 controls⁶.

Genome-wide association studies can pick up smaller signals because they use microarrays to directly compare thousands of SNPs between populations to find suspect patterns of variation⁷. These studies are being widely adopted — in part because of their rapidly decreasing cost and ability to search more broadly for risk genes.

For example, the new study suggests that classical linkage and association approaches alone are not sufficient to deal with the genetic heterogeneity seen in autism.

The Autism Genome Project, an international research partnership created to assemble and study the world’s largest gene bank for autism, is completing the largest genome-wide association study to date. Monaco says his group is also independently studying DOCK4 variants that regulate gene expression in order to understand how the variants confer susceptibility to autism.

Still, although genome-wide studies are effective at finding less widespread, yet common variants, linkage studies, whole-genome sequencing and analyses of CNVs will all be important in understanding the genetics of autism, notes Chakravarti.

“We don’t rely on one telescope to conduct astronomy experiments,” Chakravarti says. “We’re too early in this game to say which types of studies will pan out or not.”

References: