Rapid Ranking of Heterocycle-Side Chain Stacking Interactions Based on New Molecular Descriptors

Abstract

Despite the importance of stacking interactions between ligands and aromatic amino acid side chains in binding sites, it is still unclear how such interactions can be optimized in the context of design. I will present highly-accurate computed interaction energies for parallel stacked dimers of aromatic amino acid side chains and pharmaceutically relevant heterocycles. The trends in binding energies that are observed can be described using a combination of known descriptors of dispersion and new electrostatic potential (ESP) based heterocycle descriptors we have developed which have been used to generate a multi-parameter predictive model for the interaction energy. These electrostatic descriptors are based on statistical features of the electric field and ESP in the plane near the heterocycle and can be directly related to the number and arrangement of the heteroatoms. This relationship provides a means for rapidly evaluating these descriptors, and consequently the stacking interaction energies, directly from simple molecular representations (e.g. SMILES, connectivity matrix). This enables the rapid ranking of large sets of heterocycles based on their stacking ability.