High Throughput Virtual Screening and Mutational Studies Using Docking and Molecular Dynamics on a Shared Memory GPU Supercomputer

Abstract

Human Fructose 1,6 Bisphosphatase (FBPase), an enzyme which serves as the control point in the gluconeogenesis pathway, was selected as a protein target for this study. FBPase is a putative target for drug development to combat Type II Diabetes. The FBPase protein structure has previously been solved (Zhang et al., 1995), and coordinates were readily available (PDB code IFTA) and downloaded for docking studies from the Rutgers Consortium of Structural Biology (RCSB) Protein Data Bank (PDB). The potential inhibitory molecules (IBS Natural Products Catalog) were downloaded from the ZINC Database (Irwin et al., 2012) in pdbqt file format. AutoDock Vina was selected as the ideal docking program for FBPase as determined by comparing co‐crystallized ligands to docked ligand positions. Selected compounds were chosen and docked into the FBPase active site and allosteric binding sites. Following the identification of promising theoretical binding constants for each molecule, molecular dynamics studies were performed for the FBPase protein target in the absence and presence of these promising molecules. Studies were caeried out using a NAMD/VMD (Nanoscale Molecular Dynamics with Visual Molecular Dynamics) system. NAMD implemented with CHARMM foce field is known to be highly efficient in simulating large systems for molecular motions. Dissociation constants (Ki) for the protein‐ligand complexes were calculated through NAMD runs, and respective conformational changes in the FBPase binding pockets were observed. Binding pockets were evaluated based on two factors: 1) pocket shape, and 2) pocket volume. Laboratory‐based site‐directed mutagenesis results revealed various levels of activation/inhibition of the enzyme as a result of mutations, an outcome with genetic implications for disease. Mutant enzymes resisted attempts at crystallization so molecular dynamics (MD) studies were performed on models of mutants. Focused on the active site of allosteric binding sites, further analysis was performed with key residues (virtually mutated) to initiate activation or inhibition of the enzyme (retention in T‐state or transition to R‐state) using NAMD. Resulting models of FBPase after MD runs shed light on possible interfacial structural changes produced in response to these mutations. Structural alterations in the models were observed in the active site, AMP (adenosine monophosphate) allosteric binding site, dimer interface allosteric binding site, and the interface between the active site and the AMP binding site. The signal from one part of the molecule is trasmittedto another part of the molecule through a known hydrogen‐bonding network (a sequential allosteric transition) vconnecting the active sites to the AMP allosteric binding sites. Novel hydrophobic networks were identified to connect the active sites to the allosteric binding sites using the NAMD protocol.This abstract is from the Experimental Biology 2019 Meeting. There is no full text article associated with this abstract published in The FASEB Journal.