Home > News & Opinion > News > 2012 > Autism risk gene found to alter brain wiring

Autism risk gene found to alter brain wiring

Popularity tracker
Virginia Hughes
13 September 2012

Telling tracts: Some long white-matter bundles (orange) are stronger in children who do not carry MET variants than in those who do.

MET, a leading candidate gene for autism risk, influences the strength of connections between brain regions involved in social behaviors, and this effect is especially prominent in people with the disorder. The findings are from a large study using several imaging techniques, published 6 September in Neuron1.

The new report finds that children carrying a common, autism-linked variant in MET have weaker connections — as measured in both activity patterns and anatomy — between distant brain regions. This broadly agrees with a popular hypothesis that autism results from weak long-range connections in the brain, although that theory has lately been called into question.

The differences in connectivity appear in children with and without autism, but are more prominent in those with the disorder, the study found. Still, the data also indicate that MET is not the sole or even a primary factor in autism, as many study participants with the disorder don’t carry the variant.

“It highlights how all of these genetic factors are contributing to the disorder but they’re not necessarily leading to the disorder,” says lead investigator Mirella Dapretto, professor of psychiatry and biobehavioral sciences at the University of California, Los Angeles.

The study is powerful, according to several experts, because it suggests that combining discoveries in genetics with technological advances in brain imaging can help researchers parse autism’s diversity.

“What the study offers us is the first glimpse into the fact that there is a discoverable relationship between genetic risk and changes in the brain structure and function,” notes Marlene Behrmann, professor of cognitive neuroscience at Carnegie Mellon University in Pittsburgh, who was not involved in the work.

Multiple hits:

The study is significantly larger than other analyses of its type, analyzing 75 children who have high-functioning autism and 87 controls overall.

Using different imaging techniques, the researchers measured brain activity in these individuals and tracked the nerve bundles, or white matter, that connect one brain region to another. They also analyzed DNA from the participants’ saliva to determine the number of copies they carry of a specific MET variant — denoted with the letter C.

In one set of experiments using functional magnetic resonance imaging, the children lay in a brain scanner as they looked at pictures of faces showing different emotions. Regardless of whether they have an autism diagnosis, ‘CC’ participants show over-activation in the amygdala, a brain region involved in emotion processing, and the striatum, which is important for movement planning and memory.

In children with two C variants of MET, activity in the default mode network is not synchronized as tightly as in children who carry no MET variants, the study found.

Finally, the scientists used a technique called diffusion tensor imaging, which tracks the flow of water in nerve fibers to measure the strength of the brain’s anatomical bridges. Children with the CC type have less robust connections in several long white-matter tracts compared with children who carry other variants, the researchers found.

In all the imaging experiments, the risk variant’s effects are more pronounced in participants with autism than in typically developing children. Among children who carry a single C variant, those with autism show patterns similar to children (with or without autism) who have the CC variant, whereas those without autism are more similar to children who don’t have any C variants at all.

“[The MET variant] takes a bigger toll, if you will, in the brain of a child who develops autism,” Dapretto says.

This complexity, Dapretto adds, highlights one of the biggest unanswered questions in the autism field: why some carriers are vulnerable to autism whereas other carriers are protected. “It could be a multiple-hit story,” she says.

So far, common variants in only a handful of genes have been linked to autism: MET, the oxytocin receptor OXTR, and CNTNAP2, which is involved in communication between brain cells.

In 2010, Dapretto and her colleagues used functional magnetic resonance imaging to show that individuals who carry certain CNTNAP2 variants — regardless of whether they have autism — have weak connections between regions in the front and back of the brain4.

Connectivity confusion:

Both the CNTNAP2 and MET studies partially support the ‘connectivity theory’ of autism, which says that individuals with the disorder have weak connections between distant regions and unusually strong connections between nearby regions.

Mice lacking CNTNAP2 or MET also add heft to the argument. Last year, a study reported that mice missing MET have unusually strong short-range connections in the outer layers of the brain.

Still, precisely how connectivity is disrupted in the brains of children with autism is unclear, experts say.

Earlier this year, Kennedy reported at the International Meeting for Autism Research in Toronto that people with autism and typical controls have largely similar patterns of brain connectivity.

The connectivity hypothesis has come under fire in the past couple of years partly because of a common methodological issue. Children with autism are more likely than controls to move their heads while they are in a brain scanner, which can make long-range connections appear weaker than they actually are.

In the new study, the researchers note that the autism and control groups showed a similar range of head motion. Still, even tiny movements of a few individuals can affect imaging data, and the researchers did not use a new ‘scrubbing’ method recommended by some experts to control for the head motion.

“It worries me a little,” notes Damien Fair, assistant professor of behavioral neuroscience and psychiatry at Oregon Health and Science University, who was not involved in the study. “It’s one of the big problems that has kind of crept into the literature these days and has not been fully resolved.”



1: Rudie J.D. et al. Neuron 75, 904-915 (2012) PubMed

2: Campbell D.B. et al. Proc. Natl. Acad. Sci. U. S. A. 103, 16834-16839 (2006) PubMed

3: Campbell D.B. et al. Ann. Neurol. 62, 243-250 (2007) PubMed

4: Scott-Van Zeeland A.A. et al. Sci. Transl. Med. 2, 56ra80 (2010) PubMed


Name: Brian Scott
13 September 2012 - 8:50PM

Hi Virginia Hughes -

I wonder, does the study indicate if the detection of connectivity differences based on allele type gives us any insight into the head movement / confounder problem?

Also, there a study on the horizon that utilizes a combined MET-C/CNTNAP2 detection to try to detangle low penetrance participation of genes?

Very nice write up. Thank you.


Name: Virginia Hughes
13 September 2012 - 9:10PM

Hi Brian,

Your first question is interesting. If you take as a given that kids with autism tend to move more than kids without autism, then you'd expect that any head motion artifacts would show up in all of the autism groups, regardless of genotype. Here, they showed these connectivity differences between kids with autism who carry MET variants and kids with autism who do not. Still, head motion is a really big problem because even teeny tiny movements (less than a millimeter, I believe) of a couple of individuals can affect data....so, I guess the unsatisfying answer to your question is maybe.

As to your second question, yes, Dr. Dapretto mentioned that she would like to do a study of people who carry both CNTNAP2 and MET risk alleles. You can imagine, though, that finding people (with and without autism) who carry both will make recruitment much more challenging. Still, given that the field is moving toward more large-scale sharing of imaging data, I'm sure it will happen soon enough.

Thanks for reading SFARI.


Name: RAJensen
14 September 2012 - 2:13PM

Common variants in the CNTNAP2 and MET genes penetrate throughout the general population. Common variants in these genes have been described as ‘autism’ susceptibility genes when they are studied in autistic people only. In general population studies common variants in these genes are associated not with autism as such, but more robustly with a wide array of broader autism phenotype (BAP) social and non-social features including self reported schizotypal personality features in healthy young males, obsessive compulsive disorders, immune deficiencies, sensory abnormalities and anxiety disorders.

These findings in general population studies might be explained by Rutter’s ‘two-hit’ hypothesis which states that the genes underlying the broader autism phenotype are not the same as the genetic and environmental factors involved in the transition to ‘autism’.


To be continued below

Name: raj
14 September 2012 - 2:15PM

Increased paternal age and increased maternal age are both associated with autism risk. However, several studies have found that there is no additive effect when both parents are aged. This might be explained by Rutter’s ‘two-hit’ hypothesis which states that the genes underlying the broader autism phenotype may not be the same as the genetic and environmental factors involved in the transition to autism.


Ghaziuddin tested the Rutter hypothesis in Down’s syndrome with and without autism featuring the interaction between a de novo genetic mutation, Trisomy 21, and a background broader autism phenotype genetic effect and when both independent component parts are present a developmental trajectory to an autism diagnosis is followed. In Down’s syndrome with autism there was an excess of unaffected first degree relatives (parents and siblings) who exhibited an increased family loading of broader autism phenotype features compared to first degree relatives in Down’s syndrome children without autism who did not. Autism was not present in first degree relatives, parents and siblings, and the genes underlying the broader autism phenotype component part are independent of and are a background genetic effect secondary to the disruption of early brain development in Down’s syndrome and the transition to autism as predicted by the Rutter hypothesis.




Name: Greg Bull
15 September 2012 - 10:45PM

There may be a further connection to consider which might possibly contribute to understanding why the risk variant's effects are more pronounced in participants with autism. L. Heur et al Translational Psychiatry (2011), show that the functional MET promoter C allele is associated with the presence of ASD associated maternal antibodies to fetal brain proteins....maybe another multiple hit?

Name: Virginia Hughes
17 September 2012 - 3:06PM

That's an interesting idea, Greg -- thanks for pointing out the reference.

Add a Comment

You can add a comment by filling out the form below. Plain text formatting.

Question: What is 10 + 4 ?
Your answer: