Proteins, polymer chains comprised of amino acids, are one of the fundamental machines that drive life. They perform a wide variety of functions within the cell such as nutrient transport, cell signaling, energy production, DNA repair, enzymatic reactions etc.. It was generally accepted that in order to perform these functions the sequence of amino acids would fold into a low energy structure, i.e. the “structure-function” paradigm. At the turn of the century, a new category of proteins was discovered which did not follow this paradigm, intrinsically disordered proteins (IDPs). In fact, we know now that IDPs perform a variety of functions through folding-upon-binding to partners, forming dynamic fuzzy-complexes, or liquid-liquid phase separation, while maintaining a unfolded/unstructured state in solution. These dynamic, unstructured ensembles are incredibly difficult to characterize using previous methods such as X-ray crystallography. Molecular dynamics (MD) simulations, with previous success modeling structured proteins, became one of the key computational methods for determining the conformational ensembles of IDPs. There are two main problems associated with MD simulations of IDPs: sampling and force-field selection. In this dissertation, we discuss these issues using two intrinsically disordered proteins/regions: Q15, a simple disordered peptide, and ChiZ-NT, disordered region from the TM protein ChiZ in Mycobacterium tuberculosis. In the first chapter, we compare the conformational ensembles of Q15 over a series of temperatures with multiple replicate conventional MD (cMD) simulations as well as two enhanced sampling methods, temperature replica exchange (TREMD) and replica exchange with solute tempering (REST), showing that the conformational ensembles of all three simulations are similar, but only the two enhanced sampling methods could reproduce physically correct temperature dependent helical melting. The second chapter focuses on solving the conformational ensemble and backbone dynamics of ChiZ-NT in solution using long-multiple replicate MD simulations. After extensive force-field validation against various experimental parameters, a viable force-field was found which could model the structural properties, and without initial validation, the backbone dynamics of ChiZ-NT. ChiZ-NT exhibits disparate dynamics in the two halves, and long-MD determined that the dynamics is due to sequence specific interactions. Finally, this force-field is used to sequentially model ChiZ in a Mtb mimetic membrane environment, a previously difficult simulation for IDPs due to both sampling and force field-selection. Again long-multiple replicate MD simulations accurately modeled the structure and dynamics of the disordered region of ChiZ as verified by solution and solid-state NMR. Several IDPs form amphipathic helices upon association with membranes, but ChiZ-NT remains disordered upon association with anionic membranes. The strict characterization by solid-state NMR in combination with MD simulations results in the first identification of a disordered, fuzzy complex between disordered regions and the membrane as has been observed between IDP-protein interactions.
