Using NMR-derived restraints in combination with molecular dynamics simulations to derive thermodynamic parameters for biomolecular interactions.

Complex biological processes involve the binding of one molecule by another. The binding process can be thought of as a 'shape' problem, whereby the strength of binding depends critically on shape complementarity between the ligand and the binding pocket on the protein. However, the binding process is more complicated than this, since it is also determined by the extent of dynamics ('floppiness') of the interacting partners. A valuable technique to measure the contribution of protein dynamics to binding concerns high-resolution nuclear magnetic resonance (NMR) relaxation-time measurements. However, the intepretation of such measurements requires some knowledge of the nature of the molecular motions, and in the past it has been common practice to assume a simplified model for these motions. Here, we propose to make use of theoretical molecular dynamics simulations which are restrained by the experimental NMR data. Such simulations will offer a much more realistic picture of the motional processes that take place in the protein before and after ligand binding, and will permit a more accurate quantitation of the dynamic factors that influence the binding process. Such information is of significant importance in the context of structure-based discovery of lead compounds for novel drugs.

Complex biological processes involve the binding of one molecule by another. The affinity of a given interaction is governed by the standard free energy of binding which in turn comprises enthalpic ('structural') and entropic ('dynamic') factors. High-resolution NMR has become an established tool to probe protein dynamics through relaxation-time measurements. Moreover, it is possible to estimate conformational entropies for molecular motions on appropriate time-scales via the order parameter for these motions. The latter can be derived from NMR relaxation-time measurements using the so-called 'model-free' formalism. However, the estimation of entropies is dependent on the nature of the motions, and it is commonpalce to assume simple models. The derivation of binding entropies in turn assumes that the model for the motions is unchanged before and after ligand binding, but the validity of this assumption remains untested. Here, we propose to apply recently described replica-exchange molecular dynamics simulations incorporating order parameters as restraints, in a model protein-ligand interaction. Such simulations will provide a much more realistic model of the nature of protein motional modes before and after ligand binding, and will permit more accurate conformational entropies to be obtained.