This project focuses on the study of membranes, proteins and carbohydrates by molecular dynamics computer simulation. ? ? Force field (FF) development (1) continued for three classes of compounds: lipids, carbohydrates, and ethers. All involved extensive ab initio calculations on small model compounds, followed by extensive molecular dynamics simulations (MD) on the target systems to ensure that a variety of target data were reproduced. The revised CHARMM monosaccharide FF (2) shows excellent agreement with experimental pucker angles, and the distributions and relaxation rates of the exocyclic torsion hydroxyl group, and solution density data. ? ? A revision (5) to the CHARMM ether FF based on target data for dimethoxy ethane and subsequent MD simulations on polyethylene oxide (PEO) and polyethylene glycol (PEG) resolved an apparent paradox in membrane biophysics: the observation that the solution hydrodynamic radius of the largest PEG able to diffuse though a membrane pore is nearly equal to the radius of that pore. Simulations indicated that the hydrodynamic radius is nearly identical to length of the middle dimension of the PEG, implying that the polymer orients in the pore and diffuses though without appreciable distortion. This result provides a physical basis for why PEG and related polymers are so useful for sizing membrane pores, and explains observed differences in diffusion rates between smooth and irregularly shaped pores (the polymer must distort when passing through the latter). A combined simulation and experimental project in collaboration with Sergey M. Bezrukov of NICHD is underway to explore membrane transport of PEG and assorted dendrimers. Development of a coarse grained (CG) FF of PEG is also underway. Such a model, in combination with CG protein FF, will allow simulations of transport through large membrane pores. ? ? The results from simulations of GC membrane models are being compared with those from all-atom simulations and macroscopic measurement to determine the validity of CG models in the high and low frequency regimes, respectively. These results will be used to develop multiscale models for simulations of large and complex membranes.? ? High level quantum mechanical calculations with and without implicit solvent demonstrated how solvent modulates the torsional surface of the model carbohydrate, 5-(hydroxymethyl) tetrahydropyran (8). In addition to providing insight into effects of stereoelectronic and entropic stabilization, the method can be applied to FF development to adjust torsional surfaces for missing solvent effects. ? ? Simulations of Helix-0 of the N-BAR domain in SDS micelles showed how the peptide conforms to the highly curved surface of micelles, and aided the interpretation of experimental measurements (6). The combined theoretical and experimental findings support models for membrane curving by BAR domains, where helix-0 increases the binding affinity to the membrane and enhances curvature generation.? ? Hydrodynamic calculations of a variety of proteins were used to support the development of a Langevin network model of myosin (7).? ? Extensive testing of the revised IPS (Isotopic Periodic Sum) method of Wu and Brooks was also carried out, and led to substantial improvements of the method. Simulations of DPPC and DMPC monolayers indicated that the surface tension obtain with the current CHARMM FF are too high. This further demonstrates the importance of long-range Lennard Jones interections, and provides a critical piece of data for the reparametrization efforts underway (1).
Showing the most recent 10 out of 13 publications
