ELSEVIER Computer Physics Communications 91 (1995) 305-319 Computer Physics Communications Computer simulation of protein motion W.F. van Gunsteren, P.H. Hiinenberger, A.E. Mark, P.E. Smith, I.G. Tironi Laboratory of Physical Chemistry, ETH Ziirich, CH-8092 Ziirich, Switzerland Received 23 December 1994 Abstract The application of molecular dynamics computer simulation methods to study the dynamics of proteins is reviewed with an eye to its possibilities and limitations. Examples are given, mainly using nanosecond trajectories of the proteins bovine pancreatic trypsin inhibitor and lysozyme, of the different protein properties, of which the dynamics can be or cannot be sampled on a nanosecond time scale. It is concluded that the major asset of the simulation technique is that the different factors contributing to the dynamics of a particular process can be analyzed at atomic detail, as long as one has sampled the appropriate time scale. Keywords: Protein dynamics; Molecular dynamics simulation 1. Introduction Our knowledge of biomolecular systems and pro- cesses is steadily increasing due to the continuous advance of experimental techniques that reveal atomic properties of biomolecules such as proteins. X-ray diffraction provides a detailed, but static pic- ture of the spatial atomic structure and an indication of the extent of atomic motion or disorder. Energetic information at the atomic level is largely inaccessible to experimental probes. Information concerning the dynamics of particular atoms, bond vectors or aro- matic groups can be obtained by different spectro- scopic techniques, including 13C and 15N nuclear magnetic resonance (NMR) relaxation, fluorescence depolarization or infrared absorption [1]. The mecha- nism of specific processes, such as protein folding, may be analyzed indirectly by trapping and charac- terizing folding intermediates [2]. Yet, the experi- mental characterization of the dynamics of protein atoms on the time scales ranging from femtoseconds to seconds is far from complete, due to the difficulty of measuring dynamics into atomic detail. An alternative method to study protein motion is to simulate protein dynamics on a computer. In the molecular dynamics (MD) simulation method, New- ton's equations of motion dri(t)/ dt= vi(t ) (1) and dPi(t)/ dt = m;1Fi(t) (2) with Fi(t) = -aV(r 1, r 2 .... , rN) / Ori (3) for the N atoms with coordinates r - (rl, rE,... ,rN) , velocities v--(v 1, v 2..... v N) and masses m i (i = 1, 2 ..... N) of a molecular system are solved nu- merically by integration in time t. The force F i on atom i is obtained by taking the negative gradient of the potential energy function V(r) = V(rl, r e .... , rs), (4) 0010-4655/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSDI 0010-4655(95)00055-0