Optimization of virtual screening against phosphoinositide 3-kinase delta: Integration of common feature pharmacophore and multicomplex-based molecular docking

Extensive research has accumulated which suggests that phosphatidylinositol 3-kinase delta (PI3K delta) is closely related to the occurrence and development of various human diseases, making PI3K delta a highly promising drug target. However, PI3K delta exhibits high homology with other members of the PI3K family, which poses significant challenges to the development of PI3K delta inhibitors. Therefore, in the present study, a hybrid virtual screening (VS) approach based on a ligand-based pharmacophore model and multicomplex-based molecular docking was developed to find novel PI3K delta inhibitors. 13 crystal structures of the human PI3K delta-inhibitor complex were collected to establish models. The inhibitors were extracted from the crystal structures to generate the common feature pharmacophore. The crystallographic protein structures were used to construct a naive Bayesian classification model that integrates molecular docking based on multiple PI3K delta conformations. Subsequently, three VS protocols involving sequential or parallel molecular docking and pharmacophore approaches were employed. External predictions demonstrated that the protocol combining molecular docking and pharmacophore resulted in a significant improvement in the enrichment of active PI3K delta inhibitors. Finally, the optimal VS method was utilized for virtual screening against a large chemical database, and some potential hit compounds were identified. We hope that the developed VS strategy will provide valuable guidance for the discovery of novel PI3K delta inhibitors.