MODELING POLARIZATION IN PROTEINS AND PROTEIN-LIGAND COMPLEXES: METHODS AND PRELIMINARY RESULTS

This chapter discusses methods for modeling electronic polarization in proteins and protein-ligand complexes. Two different approaches are considered: explicit incorporation of polarization into a molecular mechanics force field and the use of mixed quantum mechanics/molecular mechanics methods to model polarization in a restricted region of the protein or protein-ligand complex. A brief description is provided of the computational methodology and parameterization protocols and then results from two preliminary studies are presented. The first study employs quantum mechanics/molecular mechanics (QM/MM) methods to improve the accuracy of protein-ligand docking; here, incorporation of polarization is shown to dramatically improve the robustness of the accuracy of structural prediction of the protein-ligand docking by enabling qualitative improvement in the selection of the correct hydrogen bonding patterns of the docked ligand. The second study discusses a 2-ns simulation of bovine pancreatic trypsin inhibitor (BPTI) in water using a variety of fixed charge and polarizable models for both the protein and the solvent, analyzing observed root mean square deviations (RMSD), intra-protein hydrogen bonding, and water structure and dynamics. All of these efforts are in a relatively early stage of development, the results are encouraging in that stable methods have been developed, and significant effects of polarization are seen and (in the case of the QM/MM-based docking) improvements have been validated as compared to experiment. With regard to accuracy and robustness of full simulations, a great deal more work needs to be done to quantitate and improve the present models.