5 Questions: Dennis Wall on new discoveries in autism genetics

Wall discusses how he and his collaborators used whole-genome data from hundreds of families affected by autism to identify 16 new autism risk genes and a rare genetic syndrome that explains some cases of the disorder.

- By Erin Digitale

Dennis Wall

Autism is not one disease but several, a family of developmental disorders with similar symptoms. A rare, new subtype of autism was recently discovered by Dennis Wall, PhD, associate professor of pediatrics and of biomedical data science at the School of Medicine. Wall led research to analyze a large set of genetic data: complete genome sequences for 2,308 people from 493 nuclear families affected by autism.

Wall spoke with writer Erin Digitale about the findings, including the discovery of the rare syndrome and the identification of 16 genes that contribute to more-common forms of autism. Researchers from Stanford, UCLA, California Institute of Technology and the Karolinska Institute in Stockholm contributed to the work. Their study was published Aug. 8 in Cell.

Q. Prior research showed that autism is more than 80% genetic and may be as much as 60% heritable, meaning a large source of the disorder lies in the genetic code and may be passed from parent to child. How did we come to know that, and what does your new analysis add?

Wall: Heritability and genetic estimates come from pedigrees showing how often autism crops up in specific families, as well as from comparisons of autism rates in identical versus fraternal twins. There have been several compelling studies, including large cohort studies in Sweden and fantastic work by Stanford’s Joachim Hallmayer to establish the estimates.

But the specifics of how autism is inherited haven’t been well-characterized, such as which genes are important, where they’re located, what variation within those genes matters or how that’s transmitted from parents to offspring. Ours is the largest-ever study of multiplex families, with at least two children per family affected by autism plus an unaffected sibling. This unique dataset allows us to look at the contributions from mom and dad to all of their kids, and because we have both unaffected and multiple affected siblings, we can characterize statistically what specific inherited variants associate strongly with autism. 

We need huge data sets for this work because most cases of autism come from combinations of changes in several genes. The effect size of any single gene usually explains only a tiny percentage of autism symptoms, and people without autism often have a lot of the same genetic variants as those with the disorder.

Also, although inherited risk is a big contributor to the genes we highlight, the total picture of autism vulnerability comes from the combination of inherited risk plus new genetic changes that occur spontaneously in the child. Inherited and new variants combine to create the tipping point that manifests in autism.

Q. You identified 16 new autism risk genes. What’s the significance of this finding? 

Wall: We replicated others’ findings about which genes are important to autism risk and added some new ones. Replicating previous discoveries makes those 16 new genes more believable.

What is interesting about them is that they form a network, associating with each other more tightly than you’d expect by chance. The genes are talking to each other, and this implies that there is a cascade of genetic variation that lies at the root of at least some forms of autism.

Finding and understanding this network of genes helps us clarify why a common genetic change might contribute to autism in one person but not another — in a child from our data set but not his mom or brother, for example. 

We want to continue studying how different genetic changes link up with specific characteristics of people in our database, who have been very well characterized.  The network of 16 autism-risk genes is involved in chromatin remodeling during the growth of new neurons, so these genes are important to brain development, but we don’t yet know exactly how they lead to autism.

Q. Your work uncovered a new subtype of autism linked to mutations in a gene called NR3C2. Why is it valuable to pinpoint such a syndrome?

Wall: Although most cases of autism come from small changes in several genes, there are already a few known genetic syndromes that arise from single-gene mutations and cause autismlike behaviors, such as Rett’s syndrome, Fragile X syndrome and tuberous sclerosis. 

We found three families in our study that had a previously unrecognized, rare subtype of autism associated with mutations in just one gene, NR3C2. The similarities between affected individuals from these three unrelated families are really striking. They all have an abnormal shortening of the fourth and fifth fingers of both hands; a high, arched palate; sensory hypersensitivity; and abnormal speech rhythms and patterns. 

Identifying this rare syndrome is useful because it helps us parcel out our data. As we identify these rare, single-event mutations with larger genetic effects, we can essentially pull them from the “harder” group of cases, and then we should have an easier time figuring out which combinations of common genetic variants contribute to more common manifestations of autism.

And this discovery is clinically actionable. The NR3C2 mutation can be listed in genetic databases and included in genetic counseling evaluations — assuming it is replicated, of course.

Q. You also created a zebrafish model of the NR3C2 mutation. What did you learn from that?

Wall: We performed a knockout experiment in this gene and saw changes in zebrafish that are consistent with what we see in humans, including social deficits and sleep disturbances. In terms of social behavior, zebrafish usually show a preference for members of their own species, but the fish with the NR3C2 mutation don’t. The aberrations we saw are definitely not common in zebrafish; they don’t exhibit these behaviors unless there is something wrong with their brains. We were elated to see behaviors in the fish that are consistent with what we observed in humans, and this is worthy of further study for sure.

Q. Working with such a large genetic data set presents logistical challenges. How did you handle those?

Wall: These data files, consisting of whole-genome sequences for thousands of people, are big and expensive to store. It can take weeks to move them from one provider to another, and sharing data by normal means — if a scientist elsewhere says, “Hey, can you send me some data?” — can’t be done. The only way to share these data is on the cloud, but storing so much data could cost as much as $20,000 per month.

Stanford reached an agreement with Amazon Web Services, which is providing free hosting of these data for the next few years, so we’re in the lucky situation of being able to share data with the research community for free online.

That is really important because without data sharing, replication of findings is not possible. And we’re not the only experts who could find important things in these data. We have to make these data available so that everyone in the field has the opportunity to learn from them.

About Stanford Medicine

Stanford Medicine is an integrated academic health system comprising the Stanford School of Medicine and adult and pediatric health care delivery systems. Together, they harness the full potential of biomedicine through collaborative research, education and clinical care for patients. For more information, please visit med.stanford.edu.

2023 ISSUE 3

Exploring ways AI is applied to health care