Theoretical studies of proteins and protein-ligand complexes: Part I. Toward the development of an efficient all-atom implicit-solvent model for use in protein structure prediction. Part II. Molecular dynamics of studies of trypsin-ligand complexes

Part I. Toward the development of an efficient all-atom implicit-solvent model for use in protein structure prediction. An efficient implicit-solvent model that includes the hydrophobic effect, hydrogen bonding, volume exclusion, bonded interactions, and aromatic-aromatic interactions was developed and applied to protein structure prediction. A new efficient approximate surface area method for modeling the hydrophobic effect was formulated, and the importance of electrostatic interactions between aromatic sidechains in conferring the geometry of inter-strand aromatic-aromatic interactions in the "trpzip" beta-hairpin was shown. The developed model was applied to the folding of the 20 amino-acid stable designed mini-protein "Trp-cage". Using replica-exchange molecular dynamics, Trp-cage was folded to its native state with near-experimental stability. The model predicts the folded structure Trp-cage as the global free-energy minimum while being two orders of magnitude faster than explicit-solvent simulation of this protein.; Part II. Molecular dynamics studies of trypsin-ligand complexes. The binding of two small-molecule competitive inhibitors, benzamidine and tranylcypromine, to trypsin was studied using all-atom explicit-water molecular-dynamics simulations. The weaker binding tranylcypromine exhibits significant orientational mobility when bound, in contrast to the stronger binding benzamidine. Binding free-energy profiles of these two small-molecule ligands with trypsin were able reproduce absolute binding free-energies to within 1 kcal/mol. 3 to 7 A from their bound positions, the ligands undergo a combined desolvation and loss of orientational entropy. Nearly all of the free energy of binding is gained as the ligands go from 3 A away from their bound positions to fully bound. However, even at large distances from the binding pocket, both ligands show a small degree of non-specific binding to the surface of the protein, suggesting that successful binding can result from a non-specific collision event followed by diffusion of the ligand on the protein surface. Direct simulation of benzamidine binding to trypsin shows that water molecules in the binding pocket do not have to work their way around benzamidine to the bulk solvent; rather, they can be expelled through a "back door" to the binding pocket formed by small fluctuations in loop regions of the protein.