Largest Autism Protein Interaction Map Reveals Shared Pathways That Could Guide Future Treatments

SAN FRANCISCO — Researchers have published what they describe as the largest molecular map of autism-related protein interactions, showing how many different genetic mutations can disrupt the same networks in the developing brain.

The study, released Thursday in the journal Science, charts more than 1,800 protein-protein interactions linked to 100 high-confidence autism risk genes. About 87 percent of those interactions had not been reported before. The authors say the work shifts attention from long lists of genes toward the protein complexes those genes encode — a level at which future drugs might act.

“You need this insight to ultimately develop drugs, and we’ve generated a map now that is providing essentially the molecular underpinnings of autism and pointing us in a multitude of different directions for ultimate drug discovery,” said Dr. Nevan Krogan, a University of California, San Francisco professor, director of the Quantitative Biosciences Institute and senior investigator at the Gladstone Institutes.

“The hope would be at some point you’d be looking back and saying, ‘Ah, this map led to X, Y, and Z, and therefore we have now the first-ever treatment to autism. That’s the vision, and I believe that’s going to come to fruition at some point in the future,” Krogan said.

More than 250 genes have been associated with autism spectrum disorder. That diversity has made it hard to imagine a single therapy. Proteins, however, often work in shared complexes. If many mutations damage the same partnerships, a smaller set of targets could matter for larger groups of patients.

“When you have the genes and the mutations, that’s just a list. That’s a parts list,” Krogan said. “What you need to do is have a wiring diagram of that parts list, and that’s where you need to go to the proteins and understand how the proteins talk to one another, and understand when you put a mutation in a protein, what does it do to the protein-protein interactions? So, understanding that will point you down therapeutic roads that you just could not have ever imagined if you were simply just looking at the genes and the mutation.”

The team used affinity purification–mass spectrometry to map interactions for the 100 genes, including versions carrying patient-derived missense mutations. They then used AlphaFold structural predictions to flag which contacts were likely direct and which mutations were most worth testing. Priority cases were examined in human forebrain organoids and in Xenopus embryos.

“AI is allowing us to study proteins that we could never have even dreamed of even a couple of years ago, and it’s providing an unprecedented light being shone on autism, the underlying biology behind autism,” Krogan said. He called the result “the largest that has ever been done on autism,” and “the largest map of its kind for any neuropsychiatric disorder,” as well as “the largest mutant map that’s ever been generated for any area of disease.”

One example in the paper involves FOXP1. Distinct mutations weakened its physical partnership with FOXP4. In organoids, that change altered how deep-layer cortical neurons developed. Other mapped proteins clustered around neurogenesis, tubulin biology, transcriptional control and chromatin modification.

Krogan said his group already has three programs underway to explore therapies based on the map.

Independent scientists not involved in the work described it as an important systems-level resource, while stressing that drugs are still far off.

“The new study is really exciting in that it represents the most systematic protein-level view of autism risk we’ve had to date,” Fikri Birey, an assistant professor of human genetics at Emory University School of Medicine, wrote in an email. “In the future, it may be possible to identify drugs that stabilize disrupted protein complexes or block pathological interactions, rather than trying to correct every individual autism-causing mutation.”

Alison Singer, president of the Autism Science Foundation, called the paper a turning point for families of people with profound autism. “Today’s study shows where different autism genes converge on shared biological pathways, and those shared pathways could ultimately give us therapeutic targets relevant to much larger groups of autistic people, including people without an identified genetic cause. That is nothing short of a huge watershed moment for autism science.”

She added: “For families of people with profound autism like mine, this is the kind of scientific advance we have been waiting for and praying for. Many people with profound autism have rare genetic variants and the promise of precision medicine has always been that understanding those variants would eventually allow us to move beyond treating symptoms and behaviors, and begin addressing the underlying biology.”

Singer also cautioned that discoveries still must become drug candidates and pass safety and efficacy tests.

Dr. Andy Shih, chief science officer at Autism Speaks, struck a similar note. “Translating these kinds of discoveries into therapies is a long process,” he wrote. “This study provides valuable insights into potential targets, but much more research is needed before these findings can lead to clinical applications.”

The authors themselves limit how far the map should be read. Krogan said it is most directly relevant to people with profound autism — those who often need around-the-clock care and who account for about 30 percent of people on the spectrum. He argued the same protein-network approach could still inform work on schizophrenia, obsessive-compulsive disorder, tic disorders and even cancer.

The Science paper, titled “A foundational autism protein interaction atlas reveals molecular convergence,” lists Belinda Wang, Rasika Vartak and Kelsey Hennick among co-first authors, with Kirsten Obernier, Tomasz J. Nowakowski and A. Jeremy Willsey among co-corresponding authors.

The publication arrives as federal autism research priorities are again under debate. The Interagency Autism Coordinating Committee has a draft strategic plan before it. Singer said she worries a broad “injury” framing could pull money away from genetics and brain biology. Shih said he expects the plan to keep room for large-scale work on how genetic factors shape development.

For now, the practical yield is a wiring diagram, not a pill. Hundreds of risk genes remain. Most of the newly listed interactions have not been tested as drug targets. Organoids and frog embryos are models, not patients. What the map does offer is a way to stop treating every mutation as a separate lock. If enough of those locks share the same bolts, the search for a key becomes smaller — and, researchers say, more realistic than it was when the field had only a parts list.