Across multiple mouse models carrying different autism-linked variants, gene activity in the prefrontal cortex fell into two opposing patterns, a large transcriptomic study shows.
In one, genes involved in communication between neurons were less active than usual, whereas genes involved in regulating DNA and processing RNA were more active. The second group showed the reverse pattern.
The work is a step forward in the effort to organize the more than 100 autism-linked genes into biologically meaningful groups that could eventually guide targeted treatments. The findings also reinforce the idea that some of the genetic diversity associated with autism converges on a small number of recurring molecular states.
“The main novelty is the scale and comprehensiveness of the comparison,” says Kaustubh Supekar, clinical associate professor of psychiatry and behavioral sciences at Stanford University, who was not involved in the research.
Earlier mouse studies typically looked at autism-linked variants in a single brain region or developmental window. By comparing 17 mouse lines under the same conditions, across both sexes and different developmental stages and drug exposures, this work could identify molecular patterns shared across very different genetic forms of autism, Supekar says.
T
he primary distinction between the groups was the direction of the gene-expression changes. But the two cortical transcriptome groups also responded differently to pharmacological intervention that modified autism-like behaviors in mice in past research. For example, when the mice received fluoxetine or lithium shortly after birth, some disrupted gene-expression programs moved back toward a more typical pattern in one of the two groups.However, the two-group classification was not fixed. For example, in 7 of the 17 mouse models, males and females carrying the same variant fell into different groups, and some mice switched groups as they matured. And when the researchers looked in the hippocampus, they did not see the same clear two-group split.
For that reason, it is too early to call these groups “subtypes,” says study investigator Eunjoon Kim, professor of biological sciences at the Korea Advanced Institute of Science and Technology. At this point, he adds, “they are best understood as temporary biological states.”
An analysis of gene-activity data from the prefrontal cortex of 40 autistic people and 17 non-autistic people showed that the autistic samples also separated into two groups with broadly opposite patterns of gene activity, although the split was not as clear as in the mice. The team reported the findings last month in Science.
The human pattern is “relatively weak,” Kim says, and it could reflect the greater variability in the human samples, which differed in age, sex and seizure history. Still, the pattern’s presence suggests that the mouse findings capture something relevant to autism in people—a link that more brain samples, including those from other cortical and subcortical regions, could confirm, he says.
A
utism is a developmental condition, so the fact that mice carrying the same variant can fall into different groups depending on sex, brain region or developmental stage does not undermine the findings, says Alessandro Gozzi, senior researcher at the Istituto Italiano di Tecnologia, who was not involved in the study. “We should expect [autism’s] biology to change as the brain develops.”What stands out, he adds, is that even when some mouse models move from one group to the other over time, the same two broad patterns appear. The result fits with other research—including his own work on brain connectivity—suggesting that different autism-linked variants may converge on a smaller number of recurring biological patterns. “Autism-related biology may be less chaotic than the enormous genetic diversity would make us expect,” he adds.
The work also has implications for future efforts to define and identify autism subgroups. Based on these findings, “autism heterogeneity may be better understood as something that can shift across biological context and development instead of a set of completely stable subtypes,” Supekar says.
Most efforts to define biological subgroups in autism have relied on brain imaging or other broad measures, but this study suggests molecular data could provide a complementary approach, at least in mouse models, he says. If similar patterns exist in people, he adds, combining molecular, genetic, imaging and clinical data could help identify subgroups that may change over time, and show how molecular differences relate to development, behavior and treatment response.
