Fresh off winning a Nobel Prize last year, and previously tackling antibodies and peptides, the University of Washington’s David Baker and his team aim to make another breakthrough in artificial intelligence-powered molecule design—this time in constructing new enzymes from scratch, which can be built to catalyze nearly any scientific experiment.
“Enzymes are proteins that speed up chemical reactions under mild conditions, outperforming what even the best human chemists can do,” said Baker, who shared the 2024 Nobel Prize in Chemistry for his work in AI-generated proteins.
“Generating new enzymes for any chemistry challenge—whether building up pharmaceutical compounds or breaking down microplastics—would allow us to harness nature’s efficiency without relying on harsh solvents or fossil fuels,” he said in a statement.
According to the UW’s Institute of Protein Design, Baker has been working on creating never-before-seen enzymes for decades—while traditional methods have included modifying naturally occurring proteins and then nudging them toward their molecular goals over time.
These enzymes have been repurposed for nearly everything from biofuel production to laundry detergents, where they can help dissolve stains, as part of a broader multibillion-dollar industry.
“Traditional enzyme design is like buying a suit from a thrift store: the fit will probably be a little off. With AI, we can now tailor-make enzymes to ensure a perfect fit for every step of the reaction,” said Sam Pellock, an acting instructor in UW’s Baker Lab and a co-lead author of a study published today in the journal Science.
As an initial demonstration, researchers tasked their RFdiffusion AI model with generating enzymes designed to break ester bonds—the links that help form the underlying structure of many common substances.
Known as serine hydrolases, the success of the enzymes can also be easily measured, with the cleaving of bonds tracked through identifiable signals through fluorescence and other means. At the same time, such reactions can become valuable industrial processes, as in degrading plastics for recycling.
In their experiments, the researchers were able to develop enzymes that still employed a previously known mechanism of action, but by using molecular structures not found in nature.
“We tested our AI-designed enzymes in the lab and found they can be quite efficient,” said Anna Lauko, a co-lead author and Ph.D. graduate from the Baker Lab. “There is still room for improvement because these chemical transformations are complex, but I’m thrilled by what we can now accomplish with the latest generation of design tools.”