The long-range goal of this project is to build atomic-resolution molecular models of cystic fibrosis transmembrane conductance regulator (CFTR) and trace-amine associated receptor 1 (TAAR1), and design drugs using these molecular models. CFTR is a drug target for cystic fibrosis (CF) and secretory diarrhea. TAAR1 is a promising drug target for schizophrenia and drug addiction. Currently, TAAR1 lacks X-ray crystal structures, hampering structure-based drug design. Cryo-EM structures exist for human CFTR in the closed state and zebrafish CFTR in the semi-open state but homology modeling and refinement are required to generate human CFTR models useful for drug development. We have built homology models of CFTR and TAAR1. Refinement of these models will be achieved in two steps. In the first step, many plausible conformations are generated from the pre-refinement models through conformation sampling. Replica-exchange molecular dynamics (REMD), dynamic importance sampling (DIMS) and simulated annealing MD will be used in this step. In the second step, the models generated by conformation sampling will be assessed against functional and biochemical data using scoring functions. Models with the best scores will be selected as the final models. Conformation sampling usually generates models that are closer to the real native structures than the pre-refinement models. The challenge is to distinguish these better models from other models generated during conformation sampling. The scoring functions will be tested in mock (retrospective) modeling trials with members of the ATP binding cassette (ABC) transporter and G protein-couple receptor (GPCR) families, to which CFTR and TAAR1 belong, before they are used for the refinement of CFTR and TAAR1 models. Proteins with X-ray crystal structures in the Protein Data Bank are chosen for these trials. Homology models will be built and refined for these proteins so that the resulting refined models can be compared to the crystal structures. The scoring functions used for refinement of CFTR and TAAR1 models will be based on experimentally validated solvent accessible residues and inter-atomic distances experimentally estimated in this project. Thiol-reactive molecular linkers of different length can estimate the distance between sulfur atoms of two cysteines. Experimentally observed zinc binding indicates that atoms coordinating zinc are less than ~4 angstroms apart. After conformation sampling and scoring, the binding poses of CF drugs and experimental TAAR1 drugs will be predicted using a virtual docking program, Glide, with the final refined models of CFTR and TAAR1. We will validate the predictions using three empirical methods, 1) covalent attachment of a small chemical group in the binding pocket to significantly reduce the binding affinities of the drugs, 2) protection of an engineered cysteine from thiol-reactive reagents by the bound drugs, and 3) a designed single amino acid substitution to significantly increase the binding affinities of the drugs. The validated models will be useful for future drug development.
|Effective start/end date||9/1/18 → 8/31/21|
- HHS: National Institutes of Health (NIH): $46,606.00