The application of different solvation and electrostatic models in molecular dynamics simulations of ubiquitin: how well is the X-ray structure "maintained"?

We present molecular dynamics simulations on ubiquitin with explicit solvent molecules and investigate the influence of different force fields [Weiner et al. (J. Am. Chem. Soc. 106:765-784, 1984; J. Comput. Chem. 7:230-252, 1986) vs. Cornell et al. (J. Am. Chem. Soc. 117:5179-5197, 1995)], different treatments of the long-range electrostatic interaction (8 A cutoff vs. particle mesh Ewald), and different solvation models (periodic box vs. small shell of water molecules) on the structure and the dynamics of the protein. Structural data are monitored by atomic root mean square deviations (RMSDs) from the crystal structure, the radius of gyration, the solvent-accessible surface area, and the pattern of the backbone hydrogen bonds. The dynamic behavior is assessed by the atomic fluctuations and the order parameters of the N-H backbone vectors. With the Cornell et al. force field and a periodic box model, the simulated structures stay much closer to the experimental X-ray structure than with the older Weiner et al. force field. A further improvement of the simulation is found when the electrostatic interaction is evaluated with the particle mesh Ewald method; after 1.2 ns of simulation the backbone RMSD amounts to only 1.13 A. The analysis of the dynamic parameters shows that this good structural agreement is not due to a damping of internal motion in the protein. For a given length of simulation time, the shell models achieve an agreement between simulated and experimental structures that is comparable to the best models that employ a periodic box of solvent models. However, compared with the box models, the fluctuations of the protein atoms in the shell models are smaller, and only with simulation times as long as 2 ns do they become of comparable size to the experimental ones.