Accurate Prediction of Core-Hopping Transformations Using Molecular Dynamics-Derived Conformational Ensembles: Application to the Discovery of Novel P2X3 Antagonists

Traditional three-dimensional core-hopping methods aim to substitute the central core of a reference ligand conformer with alternative scaffolds that exhibit similar pharmacophoric features. As such, these algorithms are not trained to perceive the effect of a new core on the conformational equilibrium of ligand substituents. With the aim of evaluating the ability of a new scaffold to preserve the conformational dynamics of the reference compound, we developed DynaCore, a novel bespoke algorithm designed to compute the similarity of conformer ensembles collected from molecular dynamics simulations of a ligand before and after core replacement. DynaCore was subsequently used to prospectively design novel P2X3 antagonists using privileged fragments from previous medicinal chemistry campaigns. The algorithm recapitulated known conformation-activity relationships and successfully distinguished new scaffold replacements that retained or abolished ligand potency, leading to the identification of multiple potent hit series. DynaCore can potentially be applied to any compound pair sharing a common set of substituents.