Efficient prediction of transmembrane binding sites for anesthetic ligands

点击翻译按钮获取中文摘要

Many anesthetics exert their action by binding to proteins embedded in the lipid membranes that encase cells. These proteins, including receptors and ion channels, allow cells to coordinate their action across the body. Explaining at the atomic level how binding to these proteins results in anesthesia requires knowing where on the protein the ligand actually binds. Determining this is a difficult problem that can be addressed with various methods, experimental and computational. The problem is made more difficult when the true binding sites are on a part of the protein that is actually in the lipid membrane (transmembrane domains), because of the complexity of the lipid environment. Computational methods to predict these sites that can accurately treat the membrane (e.g. flooding molecular dynamics) are also inefficient. But more efficient methods, particularly molecular docking, do not properly incorporate the effect of the membrane. This project seeks to improve docking specifically so it can predict anesthetic ligand binding sites in transmembrane domains. The overall goal is to create and calibrate a docking scoring function that takes the lipids into account, by conducting certain one-time preprocessing steps. This will be done by: 1) Predicting the microarchitecture of complex lipid membranes. Lipid membranes are composed of many different lipid types, and while the proportions of these lipids are known, the way they arrange themselves at the atomic level is not. This will be predicted using long-timescale molecular dynamics simulations. 2) Calculating the free energy profiles of insertion of selected anesthetics in these microarchitectures. It is necessary to know how favorable it is for the ligand in question to exist in the lipid membrane separately from the protein, so ligand free energy profiles as a function of depth in the membrane, as well as ligand rotation, will be calculated. 3) Identifying hydrophobic regions on the protein of interest. Traditional docking assumes that the protein is entirely solvated in water. Inhomogeneous solvation theory will be used to identify hydrophobic regions that do not contain water so they may be treated appropriately. 4) Constructing a modified docking scoring function that is parameterized by this data. The data calculated above will be fit to an efficient polynomial function for supplementing an existing docking scoring function. The project, by its completion, will have substantially improved docking methodology for this specific but important use case. It also will have served to improve the PI's ability to attack similar problems in the future, preparing him for a successful career as an independent physician-scientist.