Accurate Prediction of Protein NMR Spin Relaxation by Means of Polarizable Force Fields. Application to Strongly Anisotropic Rotational Diffusion.

Among the various biophysical methods available to investigate protein dynamics, NMR presents the ability to scrutinize protein motions on a broad range of time scales. H-N NMR spin relaxation experiments can reveal the extent of protein motions across the picosecond-nanosecond dynamics probed by the fundamental parameters N-, N-, and H-N NOE that can be well sampled by molecular dynamics (MD) simulations. An accurate prediction of these parameters is subjected to a proper description of the rotational diffusion and anisotropy. Indeed, a strong rotational anisotropy has a profound effect on the various relaxation parameters and could be mistaken for conformational exchange. Although the principle of NMR spin relaxation predictions from MD is now well established, numerous NMR/MD comparisons have hitherto focused on proteins that show low to moderate anisotropy and make use of a scaling factor to remove artifacts arising from water model-dependence of the rotational diffusion. In the present work, we have used NMR to characterize the rotational diffusion of the α-helical STAM2-UIM domain by measuring the N-, N-, and H-N NOE relaxation parameters. We therefore highlight the use of the polarizable AMOEBA force field (FF) and show that it improves the prediction of the rotational diffusion in the particular case of strong rotational anisotropy, which in turn enhances the prediction of the N-, N-, and H-N NOE relaxation parameters without the requirement of a scaling factor. Our findings suggest that the use of polarizable FFs could potentially enrich our understanding of protein dynamics in situations where charge distribution or protein shape is remodeled over time like in the case of multidomain proteins or intrinsically disordered proteins.