Researchers use AI to design proteins that control hard-to-target drug receptors
Scientists at UW Medicine and Skape Bio have used artificial intelligence to design miniproteins that can turn cell signaling on or off for a class of receptors implicated in dozens of diseases but largely resistant to conventional drug development.
The work, published May 21 in Nature, demonstrates that computationally designed proteins can activate or block G protein-coupled receptors, or GPCRs - molecules that sit on cell surfaces and regulate vision, smell, hormone response, and countless other bodily functions.
The challenge with GPCRs
GPCRs have proven difficult targets for drug makers. Their signaling switches sit in deep, flexible pockets that are hard to access with traditional small-molecule drugs or antibodies. Existing drugs often bind to these receptors but fail to reliably turn signaling on or off.
The UW Medicine Institute for Protein Design and Skape Bio developed specialized design strategies to build miniproteins - proteins smaller than 100 amino acids - that can fit into these hard-to-reach sites. The approach allowed the team to create molecules that either activate or block GPCR signaling with precision.
From design to testing
The researchers also invented a new screening system to test designed proteins directly in living human cells, rather than in isolated, altered receptor preparations. This approach allowed them to evaluate tens of thousands of proteins against GPCRs while keeping the receptors in their natural state within cell membranes.
Structural studies confirmed that several designed miniproteins matched their computational models. In a mouse study, one designed miniprotein performed comparably to a clinically approved drug while producing fewer side effects.
Moving toward therapies
Skape Bio, founded by researchers from the Baker Lab and the BioInnovation Institute, plans to use these methods to develop drug candidates for metabolic, inflammatory, and neurologic diseases where GPCR-targeting treatments have not previously been available.
The work involved collaborators from Monash University, the MRC Laboratory of Molecular Biology, Johns Hopkins University, the University of North Carolina, Novo Nordisk, Lundbeck, and several other institutions.
Researchers working in drug discovery and protein engineering may find additional resources in AI for Science & Research or the AI Learning Path for Biochemists.
Your membership also unlocks:
