Studies are proposed to address the two major difficulties in application of molecular simulations to protein systems and further develop one area of major successes. The major difficulties hampering the application of molecular simulations to a broader range of important biological problems, including predicting more effective enzyme inhibitors for therapeutic use and predicting protein structure from amino acid sequence are: 1) the representation of the energy and (2) the sampling (global minimum) problem. To improve the representation of energy, we propose to develop new force fields to improve the energy function for description of non-covalent interactions and conformational analysis in proteins and protein-ligand complexes. A significant improvement in the representation of electrostatic and van der Waals energies in additive force fields is proposed, as is the development of the first general, self-consistent non-additive protein force filed. To improve the energy representation in simulations of covalent processes, we propose a new methodology for combined quantum/molecular mechanical methods to improve the representation of the energetics of enzyme catalysis and to apply this to the mechanism of catalysis by the serine and cysteine proteases and triose phosphate isomerase. Accurately characterizing all the thermally populated local minima of a molecule in solution is what we mean by the """"""""sampling problem"""""""". This problem has been solved for 18- crown-6 in vacuo and we propose to develop approaches to allow it to be solved for small organic molecules and peptides in aqueous solution. This is a very small step toward solving the """"""""protein folding"""""""" problem, but it is of importance per se in aiding in """"""""rational"""""""" drug design, since the low energy conformations of strong binding small molecules must fit into the protein active site with little or no strain. A major area of usefulness of simulation methods in understanding protein stability and ligand binding has been the use of free energy perturbation methodologies. We plan to further develop and refine such methods and to apply them to some of the most important and interesting examples of protein-ligand recognition: (a) biotinavidin, the strongest protein- small molecule interaction, (b) sequence selectivity of trimethoprim for bacterial over mammalian DHFR which is the most well characterized example of a clinically useful drug emerging from species selective differences in a key metabolic enzyme; and (c) design of a more thermally stable T4 lyzozyme as a paradigm for sequence selective increases in protein stability.
Showing the most recent 10 out of 81 publications
