(449h) Designing Proteins with Enhanced Antifreeze Activity Using Simulated Directed Evolution

Found in a variety of organisms, from cold water fish to arctic beetles, antifreeze proteins (AFPs) are a diverse class of proteins that depress the non-equilibrium freezing point of water. AFPs have been of scientific interest for over 30 years due to their role in organism survival at low temperatures, and have potential applications in cryopreservation, cryosurgery, clathrate formation inhibition, and specialty coatings. Despite a large body of experimental and theoretical literature accumulated on AFPs, there exists no widely applicable method for accurately predicting AFP activity - making intelligent design of new antifreeze materials challenging. To remedy this, we perform molecular dynamics simulation with a collection of different types of AFPs (ranging from ~30 to 300 residues in length) in liquid water just below the freezing point. By analyzing the dynamics of hydrogen bonds in water near the protein in conjunction with the geometry of the protein structure, we establish a general and quantitatively accurate correlation between simulation and experimentally measured antifreeze activity. Implementing this correlation in a directed evolution algorithm, we identify several key mutations that exhibit more than five time the predicted activity than the wild type protein.