Computational Modeling of Protein Structure, Function, and Binding Hotspots

Mixed-solvent molecular dynamics (MixMD) is a cosolvent mapping technique for structure-based drug design. MixMD simulations are performed with a solvent mixture of small molecule probes and water, which directly compete for binding to the protein’s surface. MixMD has previously been shown to identify active and allosteric sites based on the time-averaged occupancy of the probe molecules over the course of the simulation. Sites with the highest maximal occupancy identified known biologically relevant sites for a wide range of targets. This is consistent with previous experimental work identifying hotspots on protein surfaces based on the occupancy of multiple organic-solvent molecules. However, previous MixMD analysis required extensive manual interpretation to identify and rank sites. MixMD Probeview was introduced to automate this analysis, thereby facilitating the application of MixMD. Implemented as a plugin for the freely available, open-source version of PyMOL, MixMD Probeview successfully identified binding sites for several test systems using three different cosolvent simulation procedures. Following identification of binding sites, the occupancy maps from the MixMD simulations can be converted into pharmacophore models for prospective screening of inhibitors. We have developed a pharmacophore generation procedure to convert MixMD occupancy maps into pharmacophore models. Validation of this procedure on ABL kinase showed good performance. Additionally, we have identified characteristic occupancy levels for non-displaceable water molecules so that these sites may be incorporated into structure-based drug design efforts. Lastly, we have explored the potential for accelerated sampling methods to be used in tandem with MixMD to simultaneously capture conformational changes while mapping favorable interactions within binding sites. These developments greatly extend the utility of MixMD while also simplifying its application. In addition, two exploratory studies were completed. First, traditional MD simulations were performed to understand the dynamics of NSD1. Crystal structures of NSD1 capture the post-SET loop in an autoinhibitory position. MD simulations allow conformational sampling of this loop, yielding insight into its dynamic behavior in solution. Second, an epidemiological study was conducted which was aimed at understanding the transmission and sequence variation of CTX-M-type β-lactamases, in fulfillment of the clinical research component of the MICHR Translational Research Education Certificate.