Autism genes:  combinations and copy numbers

An international consortium led by researchers based at WTCHG has re-evaluated the genetic causes of autism and found new leads but no easy answers.

Autism is a disorder of brain development which clearly has a strong genetic component, but there has been no simple explanation of its genetic basis. In the past 15 years, researchers have identified more than 80 genes that have some kind of link to the disease, but different studies have often contradicted each other, and a clear-cut cause effect relationship has failed to emerge.

The team of Anthony Monaco at the WTCHG together with experts in other European countries and the EU Autism MOLGEN Consortium have now conducted a detailed study of many gene variants that have previously been linked to autism, using both published research results and new experimental analysis of numerous families with autistic children. 

What the researchers were looking for to indicate a link is known as "transmission disequilibrium."  "We looked at children with autism and their parents," explains WTCHG geneticist Richard Holt, first author of the new study. "If a genetic trait carried by one parent is transmitted to the affected children much more often than one would expect statistically, this indicates a possible association."

As always with such medical observations depending on assessments of probabilities, the trickiest part is to work out whether a deviation from the expected result is a significant result or just a fluke.  A commonly used threshold for significance of a statistical result is set at 5%, i.e. if the probability of the result occurring by chance is less than 5%, one assumes that it is probably significant. In large scale studies, however, the dice are thrown many times, meaning that even improbable results will eventually occur by chance. Therefore, the significance  threshold has to be lowered further by dividing the 5% value by the number of "throws of the dice."

With these caveats, the new study has reinforced the case for two of the previously known candidate genes for autism, known as RELN (reelin) and GRIK2 (glutamate receptor, ionotropic, kainite 2). The study also identified two new candidate genes.

None of these genes will act on its own in a simple causal connection, though. There are at least three complicating factors. Firstly, as Holt explains, copy number variations have recently emerged as an important factor in autism genetics. This means that duplication of a gene or loss of a duplicate copy may contribute to conditions like autism. Such variations are more difficult to spot than simple point mutations where a single letter of the DNA has been exchanged. Secondly, the gene variants involved in autism are likely to interact in a complex pattern, such that - to make up a hypothetical example -  the presence of mutations A, B and C and the simultaneous absence of gene variant D may trigger disease.  Finally, the importance of epigenetic markers, i.e. regulatory signals that are attached to the genes and may be inherited although they are not part of the fabric and can be removed again, in causing heritable disease is only beginning to emerge.

All these factors conspire to make the search for the causes of autism a significant challenge that will continue to keep researchers busy for a while.

Michael Gross is a science writer based at Oxford.