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.