CAREER: New routes to designing dynamic protein structures

Natural proteins perform critical functions in our bodies. Proteins can be designed to perform new functions that help society. For example, designed proteins could detect diseases, transport drugs, generate clean fuels, and defuse toxins. However, designed proteins perform worse than natural ones. This project aims to use lessons from natural proteins to produce better designed proteins. This will be done by testing whether existing designed proteins move differently than natural ones. The work will also use artificial intelligence (AI) to design new proteins with different types of motions. Overall, this research will leverage AI to help design powerful new proteins. This will pave the way for the use of designed proteins in biotechnology, medicine, and chemical manufacturing. To connect with the public, AI-powered protein design competitions will be held in classrooms nationwide, along with science comedy events. This project advances NSF’s priorities in AI, biotechnology, and advanced manufacturing. Protein design has vast potential to generate new proteins that address unmet societal needs. For example, designed proteins could detect disease biomarkers, catalyze the creation of clean fuel sources, and defuse toxic chemicals in our environment. However, the functions of designed proteins to date pale in comparison to those of natural proteins, preventing us from achieving these goals. Notably, most protein design efforts to date have focused on stabilizing a single rigid target structure. By contrast, natural proteins must move to function, including induced fit upon ligand binding, catalytic motions in enzymes, and allosteric conformational changes. The hypothesis underlying this project is that existing designed proteins populate unnatural conformational landscapes that are too rigid, too dynamic, or otherwise different from those of highly functional natural proteins – and that explicitly designing for desired conformational landscapes will improve the functions of designed proteins. The recent revolution in fast, highly accurate AI-based protein design and structure prediction algorithms presents a prime opportunity to test these ideas. First, patterns of “hidden” conformational landscapes will be contrasted in crystal structures of traditional physics-based designed proteins, new AI-based designed proteins, and fold-matched natural proteins. Second, sequence space and conformational landscapes will be iteratively explored with AI-based methods for rigid vs. dynamic proteins. Third, the set of conformational inputs needed for efficient enzymatic function will be defined using crystallographic ensembles and AI-based multistate design, coupled to biochemical and biophysical experiments. This project will illuminate new paths forward for rational design of functional proteins to address myriad challenges in biotechnology, medicine, and other industries, and will help prepare the next generation of AI-aware protein scientists. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria. NSF Award ID: 2543428 | Program: 01003031DB NSF RESEARCH & RELATED ACTIVIT,01002627DB NSF RESEARCH & RELATED ACTIVIT | Principal Investigator: Daniel Keedy | Institution: Research Foundation CUNY - Advanced Science Research Center, NEW YORK, NY | Award Amount: $763,328 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2543428 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2543428.html