Computational Insights into Ligand Selectivity of Estrogen Receptors from Pharmacophore Modeling

Abstract Estrogen receptors (ERs) belong to the nuclear receptor superfamily, which play crucial roles in the human body. To date, two subtypes, ERα and ERβ, have been identified. The function and expression of both subtypes are various, stimulating the interest for discovering subtype selective ligands. However, ERα and ERβ are highly homologous. They share 56 % sequence identity in the ligand binding domain and only two pairs of residues differ in the ligand binding pocket. In this study, pharmacophore models were built for both subtypes based on 23 ERα and 24 ERβ selective ligands, respectively. It was found from these pharmacophore models that hydrophobic and hydrogen bonding interactions were essential for the subtype selectivity. The hydrogen bond donor feature in the phenol part was a favorable contribution to ERβ selectivity, whereas the one in the benzopyran part of the ligand was important for ERα selectivity. Furthermore, simulated virtual screening was performed with both subtype specific pharmacophore models. The results confirmed the reliability and discrimination ability of both models. These findings could not only be helpful for the understanding of a possible mechanism underlying ligand selectivity of ERs, but could also pave a new avenue for the discovery of novel selective ER ligands.