Computational approaches to understanding the role of hydration forces in protein folding

We develop and employ multiple computational models of hydration and protein folding.; The highest level of detail belongs to the atomistic models used in our molecular dynamics simulations of pure water and aqueous solutions of amino acid analogues. X-ray solution scattering experiments provide detailed measurements of the structure of matter in the liquid phase; the combination of the predictions of molecular dynamics simulations with the results of scattering experiments provides a powerful tool for determining the structure of aqueous solutions in detail. We have further applied this powerful combination to undertake the complex question of the structure of liquid water. In this application, we have developed new techniques for analyzing the predicted and observed x-ray scattering from polar molecular fluids.; To better characterize the essential effects of solvation on the kinetics of protein folding, we have developed a solvation model in the context of a simplified lattice model of protein folding. This solvation model incorporates both multi-bodied and long-range aspects of hydration forces and provides a simple model for testing hypotheses regarding the role of hydration forces on protein folding. Through the use of a simplified model we are able to conduct thousands of simulations to obtain statistically reliable results over a wide range of models and find generalizable insight in these results. The use of a more realistic hydration model is shown to improve the folding of a popular, yet problematic, lattice model and points towards the importance of hydration forces in governing proper biomolecular recognition in solution.; The final emphasis of this research has been the construction and validation of more detailed, yet still simplified, off-lattice models of protein folding, suitable for simulating larger proteins, simulating with solvent molecules present, and providing experimentally-verifiable pictures of the folding process. Our first accomplishment towards this goal was the development of a novel method for designing protein sequences which will fold to a target structure in the context of our simplified model. This design method was shown to produce significantly improved sequences compared to previously used off-lattice models for simulating the folding of small all-β proteins. The method also motivates the construction of more complex models, by providing a way to productively design them.; To better match experimental details of the folding process, we have extended this model to enable the simulation of the folding of proteins with mixed α/β content. Simulation of this new model has shown significant experimental agreement with the folding of the small α/β proteins L and G. The details of protein engineering experiments on protein L are also well reproduced. Finally the utility of the model has been shown by extending it to model the larger protein ubiquitin, too large of a protein for simulation by other conventional methods. (Abstract shortened by UMI.)...