AI-Designed Proteins Enable Scientists to Watch Multiple Cellular Processes Simultaneously
Researchers at the Howard Hughes Medical Institute are combining artificial intelligence with fluorescent dyes to create imaging probes that allow biologists to observe a dozen proteins at work inside a cell at the same time-experiments currently impossible with existing tools.
The project pairs RFdiffusion, an AI model developed by HHMI Investigator David Baker's lab, with bright fluorescent dyes created by Janelia Senior Group Leader Luke Lavis. Together, they're producing a new class of probes called NovoTags that could accelerate biological discovery by revealing how multiple proteins interact within cells.
The Current Limitation
Fluorescent dyes have long helped researchers visualize cellular structures, but they require molecular tags-such as HaloTag or SNAP-tag-to bind to specific proteins. The problem: each tag uses the same chemical connector, limiting scientists to imaging one or two proteins at a time.
Studying complex biological pathways requires researchers to image a few proteins, then repeat the experiment multiple times and manually piece together the results. "Right now, it's a heroic experiment to do more than three or four colors," Lavis said.
How AI Solves the Problem
RFdiffusion generates novel proteins that don't exist in nature and bind directly to specific fluorescent dyes-eliminating the need for traditional chemical connectors. The AI model predicts how a new protein will fold and bind to a dye, then generates thousands of possible designs for researchers to test.
"Building a three-dimensional protein that surrounds a small molecule is just something you can't do without AI," Lavis said.
Unlike naturally occurring proteins that require extensive engineering to fit a particular dye, RFdiffusion creates proteins that bind tightly to a specific dye from the start. Long Tran, a graduate student in the Baker Lab who co-led recent research, compared the approach to creating "a perfect catcher's mitt for a particular dye molecule."
From Lab to Scientific Community
The team recently demonstrated three unique proteins that selectively bind three different color dyes. They tested these systems with advanced microscopy experiments at EMBL-Heidelberg.
The next phase involves developing NovoTags for about a dozen different color dyes commonly used in microscopy. This would allow researchers to image many proteins simultaneously-capturing biological interactions that current systems cannot detect.
The team also plans to create more advanced tools, including binders for dyes that change color or blink on and off, and new probes that measure physiological signals like calcium levels or metabolites.
"You can have this great imaging toolkit that could image several organelles, several proteins, several compartments, at once," Tran said. "It will really capture a lot of biological interactions that HaloTag and SNAP-Tag and fluorescent proteins are currently not able to do."
The researchers plan to make NovoTags available to the broader scientific community. The project is part of AI@HHMI, the institute's $500 million initiative to support AI-driven research projects.
Learn more about AI for Science & Research, or explore the AI Learning Path for Research Scientists.
Your membership also unlocks:
