Influence of internal dynamics on accuracy of protein NMR structures: derivation of realistic model distance data from a long molecular dynamics trajectory.

In order to study the effect of internal dynamics on the accuracy of NMR structures in detail, we generated NOE distance data from a long molecular dynamics trajectory of BPTI. Cross-relaxation rates were calculated from the trajectory by analysis of the appropriate proton-proton vector autocorrelation functions. A criterion for the convergence of correlation functions was developed, and the analysis was restricted to those correlation functions that had converged within the simulation time. Effective distances were determined from the calculated cross-relaxation rates. Internal dynamics affected the derived distances in a realistic way, since they were subject both to radial averaging (which increases the cross-relaxation rate) and angular averaging (which decreases the cross-relaxation rate). The comparison of the effective distances with average distance between the protons during the trajectory showed that for most the effects of angular and distance averaging essentially cancel out. For these distances, the effective distance derived from an NOE is therefore a very good estimate of the average distance, or the distance in the average structure. However, for about 10% of the distances, the effective distance was more than 10% larger than the average distance, while for about 5%, it was more than 10% smaller, in some cases by more than 2 A. Little correlation is observed between the effects on cross-relaxation rates to different protons of the same residue. The results of this analysis have implications for the way structures are calculated from NOE distance data. For many distances, the assumption of a rigid structure is valid, and large error bounds would result in the loss of too much information content. On the other hand, the error bounds very often employed are not wide enough for some of the effects seen in our study.