Design and Development of Drugs that Target Virus Ion Channels

Abstract

more rigorous physical model combined with effective sampling of molecular configurations is critical for binding affinity prediction to chemical accuracy, which is defined as within one order of magnitude of the true equilibrium dissociation constant. We have demonstrated that electrostatic interactions, especially electronic polarization, are critical for protein-ligand recognition due to the significant change in electrostatic environments between bulk water and protein pockets and have achieved encouraging success in treating charged species using the polarizable Atomic Multipole Optimized Energetics for Biomolecular Applications (AMOEBA) force field. To maintain accuracy while also achieving efficiency, AMOEBA has been combined with the Orthogonal Space Random Walk enhanced alchemical free energy algorithm. Here we present applications of this strategy for the computation of protein-ligand binding affinities and, for the first time, drug solubility from alchemical simulations using the Force Field X software. 2083-Plat MloK1 LigandBinding Simulations Indicate an Induced-Fit Mechanism Many ion channels such as the MloK1 channel are steered by ligand binding via conformational changes. Mainly two binding mechanisms have been proposed, induced fit and conformational selection. Using molecular dynamics simulations, we studied ligand binding of cyclic adeonise monophosphate at the cyclic nucleotide binding domain of the MloK1 ion channel of Mesorizobium loti. For this binding domain, both crystal and NMR structures have been determined for both the ligand free as well as for the ligand bound conformation. In the simulations, spontaneous binding was observed, which enabled us to determine reaction coordinates for the ligand binding as well as for the associated conformational change of the protein. We used a combination of force probe simulations, umbrella sampling, and unbiased simulations to determine potentials of mean force along these reaction coordinates, transition rates, as well as free energy differences and barriers between the most relevant substates. Our results are compared with measured affinities and kinetics, and suggest an induced fit-mechanism.