9603556 Jardetzky The goal of this research is to (1) To develop a physical model to relate the motion of the protein backbone to the ensemble average of auto-correlation functions for NH and CH bond vectors, which are measured using NMR. In order to overcome the time limit (1 ns) on calculating molecular dynamics of proteins, molecular dynamics with random walks in dihedral angle space - the MCDIS-MD (Monte Carlo in Dihedral Angle Space with Molecular Dynamics) procedure recently reported by Zhao and Jardetzky will be used. Correlation functions based on the investigator's theoretical results will be calculated from the trajectory, which can be Fourier transformed to spectral density functions. Probability distribution and correction matrices among the various dihedral angles will also be constructed. (2) To identify relevant time scales by carrying out a detailed analysis of relaxation data, using the formalisms developed by Lefevre et al. and King and Jardetsky. Three proteins on which extensive relaxation data are already available are chosen for the initial analysis: lysozyme staphylococcal nuclease and trp-repressor. (3) To use the percolation formalism of Hoch to identify clusters of trajectories that reflect conformational transitions that may be expected to be slow and to match the spectral density terms calculated from each cluster against the spectral density terms independently derived from experiment. Modeling the types of motions that can account for the slow motion terms in NMR data would be a significant advance in the understanding of protein dynamics. Understanding a protein as a molecular mechanism - which it is - requires detailed and accurate knowledge of its structure and dynamics. Knowledge of protein structure is by now very extensive - over 1000 structures at atomic resolution have been determined by x-ray and NMR. In contrast, knowledge of protein dynamics is still rather fragmentary. Theoretical Molecular Dynamics simulation s of protein motions have contributed greatly to the understanding of proteins, but remain limited to the study of very fast (piconanosecond) motions for both theoretical and practical reasons. To interpret the information on slow motions ( on nano-to microsecond or longer time scales) contained in NMR relaxation and proton exchange measurements, more powerful statistical mechanical tools must be applied, such as the MCDIS-MD method to be developed under the terms of this grant.
