Research into protein interactions sheds new light on causes of autism

Research Research into protein interactions sheds new light on causes of autism By Marissa Russo Aug 27, 2026 2:00pm autism Protein Interactions University of California San Francisco drug discovery Studying protein-protein interactions might hold the key to identifying druggable targets for autism, a group of scientists at the University of California, San Francisco (UCSF) has suggested. The researchers created a molecular map of autism, which used 100 autism risk genes to identify over 1,800 protein interactions. In a study published Aug.

27 in Science, the scientists calculated that 87% of these interactions had not previously been reported. “We have laid out a blueprint on one way drug discovery could be done, going from genes, mutations and generating these protein interaction maps,” Nevan Krogan, Ph.D., director of the Quantitative Biosciences Institute at UCSF, told Fierce. It's estimated that one in four children on the spectrum meet the criteria for profound autism, a severe presentation of autism spectrum disorder where children are typically non-verbal, have cognitive impairments and require lifelong, around-the-clock care.

The researchers at UCSF identified highly interconnected cell hubs that converged on shared protein pathways, and used AlphaFold, a Google DeepMind artificial intelligence program that predicts 3D protein structures, to help narrow down which connections were direct interactions. One interaction they identified was between the forkhead box P1 (FOXP1) mutation, a high-risk mutation in autism, and the FOXP4 protein. In a human organoid model, the scientists showed that the interaction of FOXP1 and FOXP4 caused the cells to exhibit autism-like characteristics.

When FOXP4 was deleted, the normal cell phenotype was rescued, Krogan explained. “If you just looked at the genes and mutations, there is no way you could have figured out that you need a molecular glue here, or you may need an inhibitor—not of FOXP1, but of an interacting protein,” Krogan said. Proteins are the end products of what our cells express and these proteins carry out the functions of the cell.

Proteins are also not static—they can go through modifications after they are translated that can alter the expression or function of the original protein, Krogan detailed. They can also chat with other nearby proteins and form protein-protein interactions. The many personality traits of proteins make them both interesting—and challenging—to study for drug discovery, Krogan explained.

“Drug discovery is not exploiting protein interaction data like it should,” he said. Back in 2021, Insilico Medicine partnered with Chinese pharma Huadong Medicine to prioritize protein-protein interactions to uncover undruggable targets in cancer. More recently, the London-based AI drug discovery firm, Isomorphic Labs—founded by Google parent company Alphabet—has been using AlphaFold to discover small molecules in collaboration with Big Pharmas such as Eli Lilly and Novartis.

Krogan told Fierce that he is looking to start a company and is in “deep talks” with Big Pharmas to use these findings to develop a drug discovery platform. “Let's do drug discovery with this protein-protein interaction lens or that protein complex lens,” Krogan said. “We hope that this [study] will change the way drug discoverers look at early-stage drug discovery.” autism Protein Interactions University of California San Francisco drug discovery Biotech Research