On the development and performance of multiple-macromolecule Brownian dynamics simulations

This thesis is primarily concerned with the development and application of multiple-macromolecule Brownian dynamics (BD) simulations. The thesis is organized into five chapters starting with a literature review, then moving on to simulations, and concluding with future directions. Throughout this thesis, the complexity of the simulations increases from homogeneous solutions of proteins at dilute concentrations to heterogeneous solutions of proteins at high concentrations and culminates in a model of E. coli cytoplasm.;The BD simulation method reported here allows solutions of hundreds of macromolecules to be modeled in atomic detail. The intermolecular energy function incorporates electrostatic and hydrophobic interactions. The electrostatic component is well established but the strength of the hydrophobic component is calibrated to reproduce experimental thermodynamic information.;In Chapter 2, loose protein-protein interactions are investigated by simulating homogeneous protein solutions at low concentrations over a wide range of pH and ionic strengths using three types of proteins. The simulations reproduce known experimental trends in second virial coefficients (B 22) and translational diffusion coefficients. The structural resolution of the model enables it to reproduce experimental changes in B22 values due to single amino acid substitutions.;In Chapter 3, the effects of macromolecular crowding on protein diffusion are investigated by simulating heterogeneous protein solutions with tracer proteins in background protein solutions at increasingly high concentrations from 50 g/l to 150 g/l. The simulations quantitatively reproduce the experimental concentration dependence of a tracer protein's effective diffusion coefficient in seven different protein solutions.;In Chapter 4, crowding effects of E. coli cytoplasm are investigated with simulations that incorporate the 50 most abundant types of cytoplasmic macromolecules at their experimentally determined relative concentrations in a 1000-molecule model. The cytoplasm simulations are probed with a variety of green fluorescent protein (GFP) constructs, and their diffusion coefficients compared with experiments. Additionally, three cytoplasmic concentrations of 205 g/l, 240 g/l, and 274 g/l are simulated with wild-type-GFP molecules, and the simulations reproduce the observed concentration dependence of diffusion.