Studying the protein-water interface by molecular dynamics simulations

The interaction between proteins and water is important for the functioning of these biological macromolecules. Water fulfills multiple tasks in this context. Individual water molecules can have a clear functional role by participating in hydrogen bonds that are either crucial for the three dimensional structure and/or the biological functionality of the protein. Since charged and polar amino acids are preferentially on the outside of proteins, water is necessary to solvate these functional groups. These water molecules in turn dampen the strong electrostatic fields originating from the dipolar and charged sidechains on the protein surface. The biomolecule exerts a strong ordering influence on the waters surrounding it. There are clear indications that proteins are covered by a layer of so-called hydration waters that behave very differently from normal water molecules. Thus, a potential ligand (at least at first) does not "see" its receptor directly; it rather encounters a protein-water complex. Although a variety of experimental techniques is available to study biomolecule-solvent interactions, the results of such measurements are often indirect and difficult to interpret. It is one of the strengths of molecular dynamics based simulation methods to aid the interpretation of such experiments. This project proposes a systematic study of protein solvation based on molecular dynamics simulations. We plan to accumulate trajectories (positions of atoms over time) of at least 20 nanoseconds length for four proteins (bovine apo-calbindin D9K, human interleukin-4 R88Q mutant, domain IIA of bacterial glucose permease, and ubiquitin) which represent a cross section of the structural variety of (small) globular proteins found in the database of known protein structures. From these trajectories, we intend to compute quantities which are relevant for the interpretation of inelastic neutron scattering experiments, dielectric relaxation and nuclear magnetic relaxation measurements. In addition, the computation of so-called orientational correlation functions is expected to yield additional insight into the structure of water near the surface of a biomolecule. To the best of our knowledge this is the first time that such a large variety of properties would be calculated not only for a single protein, but for four proteins of rather different fold type. The results are expected to enhance considerably our understanding of biomolecular solvation on the molecular level. A close connection to experimental data (in particular, to dielectric relaxation and nuclear magnetic relaxation measurements) will be maintained through collaborations.

A central goal of molecular biology is to understand how cells work. This, of course, requires a detailed understanding of the cell`s constituting parts, such as proteins. The interaction between proteins and water is important for the functioning of these biological macromolecules. Our computational study enlightens the structure and dynamics of the complex interaction pattern of hydrogen bonding networks of the protein and water, respectively. In particular, the drastic changes of the water network at the protein surface is analysed at two levels of resolution: The atomistic and the mesoscopic one. At the atomistic level the spatial arrangement of solvent water molecules is described in terms of the local density as well as the mutual orientation of groups of the solute(protein) and nearby water molecules. In this way a pronounced shell structure is found. The first shell being in direct contact with the protein surface is dominated by shielding of charged and polar protein groups, while in the second solvation shell the transition and integration into the water network is realized. At the protein surface a retardation of water dynamics is observed independently of the idiosyncrasy of the underlying protein. For example, the mean residence time of water in the proximity of charged amino acids is much longer compared to apolar ones. Although the dynamics in the second shell becomes more uniform, the influence of the protein on the water dynamics is fairly long ranged. At the mesoscopic level, a very coarse-grained decomposition into protein, first and second water shell and bulk water enables the interpretation and analysis of experimental dielectric spectra of protein solutions. The principal structure of this spectra is characterized by a low-frequency protein peak and a high-frequency bulk water peak. A more detailed analysis, however, reveals the importance of cross terms between dielectric components: While the major contribution comes from the coupling of protein and bulk water, the interactions between the protein and the first two shells as well as the interaction between these shells is directly visible. Altogether, our computer simulations provide valuable tools for the analysis of the complex experimental data. From the principal agreement of experimental and simulated data, simulation can infer additional, detailed information of high reliability. A typical example is the retardation factor of mean residence times in the first and second shell compared to bulk water as extracted from magnetic resonance dispersion (MRD) experiments.