Connecting conformational ensembles to properties for drugs that behave as molecular chameleons

SNIC 2019/3-295


SNIC Medium Compute

Principal Investigator:

Jan Kihlberg


Uppsala universitet

Start Date:


End Date:


Primary Classification:

10405: Organic Chemistry

Secondary Classification:

30103: Medicinal Chemistry



Our analyses suggest that drugs having a molecular weight >600-700 Da adopt a limited number of conformations in different environments (Nat. Chem. Biol., 2016, 12, 1065; J. Med. Chem., 2018, 61, 4189). This allows them to behave as molecular chameleons that adapt to the environment, thereby combining high solubility with permeability across cell membranes and potent target binding. However, the understanding of the relationships between chemical structure, populated conformations and properties is lacking for this class of larger drugs. This is mainly due to 1) lack of experimental data on which conformational ensembles drugs in this space populate in different environments, 2) that the lack of experimental data hinders development of reliable methods for prediction of conformational ensembles and 3) that conformational analysis is theoretically challenging and requires significant computational resource. We have recently taken a major leap by determining the solution ensembles of eight drugs from this high molecular weight class. Ensembles were determined in CDCl3, which has a polarity close to that of a cell membrane, and in D2O or DMSO-d6 which mimic plasma and cytosol (unpublished). The aqueous solubility, octanol-water partition coefficient and cell permeability of these drugs have also been determined. We investigated their conformations in silico and found that the sampled ensembles contained conformations determined experimentally by X-ray crystallography, but that low energy conformations from molecular mechanics often differed from the X-ray conformations (ACS Omega, 2018, 3, 11742). We believe that this discrepancy originates not only from poor sampling of conformational space, but also that implicit solvation models are imperfect and that entropy contributions are neglected to increase the speed of conformational sampling. In this project we will investigate 1) if quantum mechanical energy calculations allow more accurate identification of biologically relevant solution ensembles and 2) use all-atom molecular dynamics to simulate how the drugs partition from aqueous phase into a lipid membrane and across the membrane. The project will involve in-depth studies of three inhibitors of the hepatitis C virus NS3/4A protease (MWs appr 750 Da), one of which is linear (asunaprevir) and two that are macrocyclized in different ways (simeprevir and grazoprevir). The energies of the conformations in the in silico ensembles for the three drugs will be calculated using molecular mechanics (e.g. MMFF94) and quantum mechanics (e.g. DFT or wave functional). The results will then be compared with ensembles determined by NMR spectroscopy in apolar solution (CDCl3) to determine which method provides the best approximation of the ensembles relevant for cell permeability. In the second part of the project steered-MD simulations of the three drugs will be performed and correlated to the experimentally determined aqueous solubility, cell permeability and the solution ensembles determined by NMR spectroscopy. This project will take a major leap toward prediction of the conformational landscape of drugs that behave as molecular chameleons, and thereby towards providing the first scientific basis for rational design of orally absorbed drugs in this uncharted chemical space.