A gene-sequencing study of children with autism, described in an advance online publication in Nature Genetics on 15 May, offers a sneak peek at a technique which, combined with other approaches, may explain 40 to 50 percent of the genetic causes of the disorder within just a few years, proposes the study's lead investigator. This approach, says Evan Eichler of the University of Washington in Seattle, will potentially allow clinicians to "lift the hood on what has gone wrong in each individual child with autism," with the hope of ultimately devising individually-tailored drug therapies.
Autism spectrum disorders manifest themselves in a wide variety of ways, and researchers believe that they are highly genetically diverse, involving mutations in any of several hundred genes. While studies of twins suggest that as much as 90 percent of autism is genetically based, large-scale genetic screens over the last decade that searched for common genetic variants underlying the disorder have been disappointing. A growing body of evidence suggests that, especially in families with no prior history of the disorder, autism results not from the inheritance of an unfortunate combination of common gene variants, but from rare, spontaneous -- or de novo -- mutations in the egg or sperm.
Over the past few years, this theory has been supported by numerous microarray studies showing that children with spontaneous autism are more likely than their unaffected siblings to have de novo copy number variants, mutations in which a large chunk of DNA is duplicated or deleted.
Now, in work funded in part by the Simons Foundation, Brian O'Roak, a joint postdoc in Jay Shendure's and Eichler's labs at the University of Washington, has sequenced the exome -- the protein-coding regions of the genome -- of 20 families consisting of one child with an autism spectrum disorder and unaffected parents and siblings. In contrast to most previous studies, which had sufficient resolution to detect only large copy number variants, the new study could detect even point mutations, in which just a single DNA nucleotide is affected. "Our approach has the advantage of taking a snapshot of an individual's protein-coding genome and quickly identifying the one or two new sporadic mutations they carry," O'Roak says.
The families in the study were drawn from the Simons Simplex Collection, a large repository of genetic, phenotypic and biological data from families with just one affected child and unaffected parents and siblings. The collection was created for the express purpose of facilitating the search for rare, de novo autism mutations.
While the 20 children with autism did not have significantly more de novo point mutations than would be expected in the population at large, their mutations were much more disruptive to the proteins they encoded than is typical. What's more, a significant number of the mutations occurred in regions of the genome in which mutations are rarely found, probably because these regions are so crucial to bodily functioning that individuals with defects in those regions usually die without reproducing.
In four children, the researchers identified de novo mutations that are so deleterious that they likely play a causative role in these children's autism. Probably not coincidentally, these four children are among the most severely affected in the study group.
Three of the four mutated genes -- FOXP1, GRIN2B and SCN1A -- have previously been implicated in autism, and are thought to play roles in speech and language disorders, intellectual disability and epilepsy, respectively. The fourth gene, LAMC3, has not previously been linked to autism, but is known to be expressed in many areas of the cortex and limbic system. "Finding a LAMC3 mutation will probably set the stage for some new research agendas," Eichler says.
Two of the four children appear to have experienced a genetic double-whammy, having inherited a deleterious mutation from a parent in addition to having a de novo mutation. The child with a FOXP1 mutation also inherited a defective copy of CNTNAP2, another gene that may be involved in language development. "It's like getting hit by lightning twice," Eichler says. That child has severe autism and the greatest language deficit of any individual in the study.
The child with the epilepsy-related SCN1A mutation also inherited from his mother a deletion that increases the risk for epilepsy; and indeed, that child has been diagnosed with epilepsy. The findings support the 'multi-hit' theory of autism, the idea that it may take a combination of mutations in the same pathway to cause severe autism or related disorders.
Studying 20 families is just a start -- "a teaser," as Eichler puts it. At the same time, the study offers two important proofs of principle: It provides compelling evidence that de novo point mutations may underlie many cases of autism, and it shows that exome-sequencing is an effective way to discover which of the more than 20,000 genes in the human genome are responsible for autism spectrum disorders.
"It's like having a dartboard with 20,000 candidates -- the fact that we could pick off four outstanding candidate genes is a great success," Eichler says. "It's proof on the ground that this technique is fruitful."
The Simons Foundation is providing funding for Eichler's team and several other groups to do whole-exome sequencing of several hundred families in the Simons Simplex Collection over the next few months. As whole-exome and eventually whole-genome sequencing become more accurate and affordable, it won't be long before it will be possible to sequence several thousand families, which should be enough data to provide statistical arguments about which genes are responsible for autism spectrum disorders, Eichler says.
"Within a couple of years, we should have a pretty comprehensive view of the genes that cause autism," he says.