Computational probe of cavitation events in protein systems.

Previous all-atom simulations have identified several classes of proteins where hydrophobic de-wetting (cavitation) is at play. Here we develop and validate a computationally fast method that predicts in which protein systems water spontaneously cavitates. We implement a cubic lattice model, which incorporates the protein shape from crystallographic data and the protein-water interactions from thermodynamic data. Combining it with the previously developed coarse-grained model for water, we determine the extent of occupancy of water at protein-protein interfaces and in protein-ligand cavities. The model captures essential findings from all-atom molecular dynamics studies on the same systems by distinguishing protein cavities that dry from those that remain wet. We also interpret the origin of the cavitation inside the melittin tetramer on simple thermodynamic grounds, and show that part of the mellitin surface is sufficiently hydrophobic to trigger cavitation. Using Glauber/Kawasaki dynamics we obtain the time-scales for de-wetting events that are in agreement with those from all-atom simulations. The method can serve as an intermediate step between the necessary initial screening that identifies proteins with abundance of hydrophobic patches using bioinformatics tools (L. Hua, X. H. Huang, P. Liu, R. H. Zhou and B. J. Berne, J. Phys. Chem. B, 2007, 111, 9069), and computationally extensive studies that need to incorporate molecular details (e.g. single mutation studies of amino acid residues).