Incorporating Explicitly Treated Surface Water Produces Better Alignments Between Experimental and Predicted Anisotropic B-Factors

Abstract

X-ray crystallography has been the gold standard for the determination of protein structures. Other than providing the definition of the positions for each atom, it also yields information on protein dynamics but this aspect has been under appreciated. The B-factors in high-resolution structures are anisotropic and reflect not only the amplitude but also the direction of the atomic vibrations. This trove of information on protein dynamics presented by experimental anisotropic B-factors (or anisotropic displacement parameter, ADP) is invaluable for the study of the intriguing relationship among structure, dynamics, and function. Here starting from the definition of atom positions, we apply normal mode analysis (NMA), a computational approach for studying protein dynamics, to predict the ADPs and then align them with experimental measurements. We use both coarse-grained NMA approaches and the NMA based on canonical all-atom force fields. By comparing the accuracy of these methods, we find that the all-atom NMA and the NMA based on portioning the all-atom Hessian matrix give the best results. Moreover, adding a layer of explicitly treated water molecules on protein surface consistently makes the energy minimized structure closer to the native structure and improves the accuracy of the ADP results. The above observations could be attributed to the significant impacts on protein dynamics by surface water and the intrinsic differences in physical-chemical properties from bulk water. This study illustrates the importance of incorporating surface water in the study of protein dynamics and provides useful information for the emerging technique of ADP-based refinement of medium-to-low resolution structures.