Structure-Based Pharmacophore Models Generation and Combinatorial Screening of ICE Inhibitors

Abstract

Eight structure-based pharmacophore models of Interleukin-1β converting enzyme (ICE) inhibitors were generated by LigandScout based on ICE inhibitor complexes from the Protein Data Bank (PDB). A combinatorial screening method based on multiple pharmacophore models were proposed in present study, since the binding pockets of different complexes were different, the structure-based pharmacophore models have a high specificity and cannot cover all the active molecules. Based on the screening results of MDDR and the metrics of E and A%, a new metric CAI (comprehensive appraisal index) was defined and used to determine the threshold of hit frequency of molecules which screened by the combinatorial screening method. According to the threshold, the potential ICE inhibitors were then identified from traditional Chinese medicine with the method of combinatorial screening with eight models. The molecules hit by not less than five pharmacophore models were taken as the screening objects of ICE inhibitor, and 781 molecules were obtained.