Molecular motors perform central functions in fundamental biological processes, including cell division, DNA replication, muscle contraction, and intracellular transport. Molecular motors operate via mechanochemical cycles in which long-range conformational changes are coupled to chemical events at an enzymatic active site. Experimental and computational approaches are needed to uncover the mechanisms by which molecular motors convert chemical energy into mechanical work. The goal of this dissertation was to develop methods and software to generate structurally realistic models of molecular motor conformations compatible with experimental data from different sources.

A software system called Protein Mechanica has been developed to facilitate the construction of models of protein geometry from atomic resolution structures, lowerresolution electron microscopy data and parametric solids. Coarse-grained models of molecular structures are constructed by combining groups of atoms into a system of rigid bodies connected by joints. Contacts between rigid bodies enforce excluded volume constraints, and spring potentials model system elasticity. This simplified representation allows the conformations of complex molecular motors to be simulated interactively, providing a tool for hypothesis building and quantitative comparisons between models and experiments. This software is freely available at www.simtk.org.

Protein Mechanica was used to build an atomic-resolution model of a mouse brain myosin V, a dimeric cellular transport motor, in pre- and post powerstroke conformations from partial X-ray crystal structures. Normal mode analyses and molecular dynamics simulations were performed on the myosin V head-neck region model to understand its mechanical behavior. We found that calmodulin molecules bound to the neck move as rigid units. Two regions of reduced stiffness were identified between IQ-CaM pairs 2-3, and 4-5. These mechanical analyses guided construction of a coarse-grained model of myosin V that had 24 rigid bodies and 33 degrees of freedom.

The coarse-grained model of myosin V enabled examination of its conformations bound to its actin track. The simulations also allow us to calculate and compare elastic strain energies for these conformations, with implications for the mechanism of step size selection by myosin V. Elastic strain energy calculations showed a preferred binding to actin subunits 11 and 13. These simulation results are compatible with observations of single myosin V motors traversing suspended actin filaments that infer that the motor uses a combination of 11 subunit and 13 subunit stride sizes. These calculations extend previous simple mechanical models for step size selection and processivity, and provide atomically detailed models for comparison with future experiments.

The Protein Mechanica software described in this dissertation provides a new tool for structural modeling of many different molecular machines, basic understanding and engineering design.