Exploring the boundaries of molecular modeling : a study of nanochannels and transmembrane proteins

Many interesting physical and biological phenomena can be investigated using molecular modeling techniques, either theoretically or by using computer simulation methods, such as molecular dynamics and Monte Carlo simulations. Due to the increasing power of computer processing units, these simulation methods allowed over the last decades for the dramatic increase in knowledge of the behavior of systems at the molecular level. In the first part of this thesis the foundations of molecular modeling techniques are revisited. Empirical force fields and the physical background between thermodynamics and individual particles are discussed. The applicability of molecular modeling techniques is shown by two representative cases. First, the molecular dynamics simulation method is used to understand the dynamics of specific proteins at the molecular level. This is important, because drug design efforts are increasingly laborious, especially with the paucity of available structural information. Therefore, computational methods are helpful in predicting the structure of proteins, and, more importantly, to predict conformational dynamics leading to protein activation. To that end a specific asthma-related protein, the beta2-adrenergic receptor, is investigated in atomistic detail together with the molecules that can bind to the protein to cause activation or inhibition. Clearly, molecular dynamics simulations are an important tool to provide further knowledge on the activation pathway of this protein. Although these all-atom simulations give some insight on the dynamics, the computational demand does not allow for systems much larger than several nanometers or time scales exceeding several nanoseconds. An attempt to overcome these problems is presented by the development of a coarse grained description of the transmembrane proteins. Because coarse graining reduces the number of degrees of freedom, the computational demands decrease, and larger systems can be investigated. However, to maintain the specific characteristics of transmembrane proteins, the general force field used in molecular modeling techniques needs to be extended with hydrogen bonding capabilities and helical backbone stabilization. This new coarse grained model is applicable to transmembrane proteins, and is used to investigate two independent cases: WALP-peptides and antimicrobial peptides. The first serve as a model system for both experiments and theory to investigate the interaction between transmembrane peptides and lipid membranes, whereas the latter are antibiotics whose pore-forming capacities are of great interest to act as target-specific drug candidates. From the molecular dynamics simulations of the WALP-peptides it is shown that the apparent hydrophobic mismatch between peptide and membrane can be resolved by two mechanisms (membrane thickness adaptation and peptide tilting) and that these two mechanisms occur sequential and not in parallel. In the case of the antimicrobial peptides it is shown that many of the orientations found with the molecular simulation techniques are in agreement with experimental observations. The second case to show the applicability of the molecular modeling techniques is that of the heat transfer characteristics of gas flows in nanochannels. Understanding these characteristics is important, because these very small channels are considered to be promising devices to locally cool systems (such as computer processing units) or to be used in lab-on-chip devices for at home medical diagnostics. Thus, understanding the interactions between the channel walls and the gas flow is of great importance. Unfortunately, the computational cost involved in simulating the solid wall, currently restrains the size of the systems that can be investigated using molecular dynamics simulations. Therefore, instead of the explicit modeling of the solid wall, appropriate boundary conditions are used, such as wall potentials or stochastic models. Both of these boundary conditions are examined in great detail and a new wall potential is presented. Also the investigations of a specific case of a channel with platinum walls with a noble gas (argon or xenon) in between allows to introduce a new method to compute an important heat transfer determining parameter. Furthermore, it is shown that both boundary conditions have their benefits and drawbacks, and that the use of either one depends heavily on the application under consideration. Both cases used to show the applicability of molecular modeling techniques, although very different from each other, indicate the importance of particle simulation methods. Investigating the interactions at the molecular level, and the development of new models allows for an even better understanding of underlying molecular processes