In lab, UCSF maps 1,800 protein links in autism

A decade of work at the University of California, San Francisco has produced a molecular map with more than 1,800 protein–protein interactions linked to autism. The team studied 100 high-confidence autism risk genes and found that 87% of the interactions had never been reported. The map turns a long list of genetic risks into a diagram of the protein machinery those genes help build.

The key shift is from genes to their working parts. Using affinity purification-mass spectrometry, or AP-MS—a laboratory method for identifying proteins that bind together—the researchers tracked how autism-related proteins interact. They then examined 54 patient-derived mutations and found that distinct genetic variants can rewire these networks in similar, convergent ways.

That convergence may matter more than the sheer number of risk genes. The study found that many genetically different forms of autism disrupt the same protein complexes, suggesting that a drug aimed at a shared molecular hub could potentially help several groups of patients. In lab-grown brain organoids, separate mutations in FOXP1 and FOXP2 disrupted the same FOXP1–FOXP4 interaction, with effects including premature cortical-neuron development and increased neural-circuit excitability.

The researchers combined their interaction maps with AlphaFold structural predictions, an artificial-intelligence system for predicting protein structures, to pinpoint mutation-affected interfaces. Those locations offer starting points for drug discovery. Compared with gene-specific approaches such as antisense oligonucleotides—short genetic medicines designed to alter RNA—or CRISPR-based therapies, drugs acting on shared protein interactions could potentially offer broader reach, as well as advantages in brain delivery, tolerability, scalability and manufacturing. Those advantages remain possibilities, not demonstrated clinical outcomes.

In practical terms, the work gives researchers a way to search for treatments that do not require a separate medicine for every mutation. It is most directly relevant to the approximately 30% of people with profound autism who carry rare, high-impact mutations in established risk genes. But the map is a laboratory framework for connecting genetic variation to protein networks and therapeutic targets.