Abstract
There is ample computational, but only sparse experimental data suggesting that pico-ns motions with 1 Å amplitude are pervasive in proteins in solution. Such motions, if present in reality, must deeply affect protein function and protein entropy. Several NMR relaxation experiments have provided insights into motions of proteins in solution, but they primarily report on azimuthal angle variations of vectors of covalently-linked atoms. As such, these measurements are not sensitive to distance fluctuations, and cannot but under-represent the dynamical properties of proteins. Here we analyze a novel NMR relaxation experiment to measure amide proton transverse relaxation rates in uniformly (15) N labeled proteins, and present results for protein domain GB1 at 283 and 303 K. These relaxation rates depend on fluctuations of dipolar interactions between (1) HN and many nearby protons on both the backbone and sidechains. Importantly, they also report on fluctuations in the distances between these protons. We obtained a large mismatch between rates computed from the crystal structure of GB1 and the experimental rates. But when the relaxation rates were calculated from a 200 ns molecular dynamics trajectory using a novel program suite, we obtained a substantial improvement in the correspondence of experimental and theoretical rates. As such, this work provides novel experimental evidence of widespread motions in proteins. Since the improvements are substantial, but not sufficient, this approach may also present a new benchmark to help improve the theoretical forcefields underlying the molecular dynamics calculations.