Multiscale Modeling of Peptides and Star Polymeric Systems

This work outlines two methods in the computational study of protein folding which aim to enhance conformational sampling while reducing thecomputational demands of the simulation. One common strategy of enhancing conformational sampling that has been incorporated into many simulation algorithms is to periodically afford the simulated molecule the opportunity to escape from energy minima and to thereby sample a much larger volume of phase space than by conventional methods. In the self-guided Langevin (SGLD) formalism, low-frequency modes of motion of the protein are enhanced in order to allow the protein to cross potential energy barriers. We have applied SGLD to three model peptides in implicit solvent in order to examine the effect of the method's two adjustable parameters on the peptides' resulting structural ensembles and folding rates. The model systems are of similar sizes but differing topologies, which allows for examination of transferral of parameters between systems.;Another strategy of enhancing sampling is an extension of parallel-tempering Monte Carlo to molecular dynamics. In this method, known as replica-exchange molecular dynamics (REMD), periodic attempts are made to exchange structures that are simulated at different temperatures, and a random walk in temperature space is achieved in order to surmount conformational barriers in the energy landscape. Variants of this technique have been developed over the years in order to increase the efficiency of REMD simulations of biomolecules. In particular, approaches have been developed in which a structural reservoir is used to decouple the high-temperature search for structures from the exchanges and annealing which occur at lower temperatures. It has been shown that the contents of this reservoir need not comprise a Boltzmann-weighted ensemble; any ensemble of structures may be used as long as its probability distribution is known. Expanding on this method, we have developed an algorithm to further enhance the efficiency of reservoir REMD through the inclusion of a weight factor that relates the relative probabilities of the highest-temperature replica structure and the structure in the reservoir under exchange. In this work, we outline attempts to apply this method to the model system alanine dipeptide, and discuss the results obtained using a coarse-grained model that considers only the potential energy of the dipeptide as a function of its dihedral angles and does not consider its atomistic degrees of freedom.;Finally, the application of simulation methodology to a non-biological self-assembling polymeric system on the nanoscale is demonstrated in this work, and its potential application to the field of targeted drug delivery is discussed. Diblock star copolymers are self-assembling nanoscale systems that have shown great potential in the field of targeted drug delivery in the human body. Intriguingly, these star polymer systems bear many important similarities in structure and composition to proteins, being composed of linear polymeric chains of repeating units which self-assemble with hydrophobicity as the driving force. These similarities allow for the application of many of the techniques of molecular modeling and simulation developed for proteins to these systems. At present, experimental imaging of star diblock copolymers and nanogel star copolymers, particularly in complex with drug molecules, has been limited, providing computational studies with the opportunity to predict the structures of these molecules in atomic detail, as well as their dynamic behavior. In this work, we describe a comparative study of three star block copolymer systems with varying hydrophobicity in their core regions. The goal of this work is to provide atomic-level information on star polymer structure and dynamic behavior, including the size and shape of the polymer, the details of its bonding patterns, and its potential for aggregation. Additionally, the kinetics of drug uptake and delivery, as well as the degradation profile of the delivery material, may also be examined. Because theoretical methods, in contrast to experiment, are often less expensive and more time-efficient, their systematic application may offer strategies at the molecular level by which to modify formulations of drug and polymer for optimal compatibility and delivery efficiency. (Abstract shortened by UMI.).