Multidrug resistance-associated proteins (MRPs) can efflux structurally diverse drugs, drug conjugates, drug metabolites, as well as other small molecules out of the cells, and this is the main cause of producing multidrug resistance (MDR) of some anticancer drugs. Therefore, it is crucial to uncover the molecular features of MRPs substrates in developing anti-MDR cancer therapy.In the present study, common feature pharmacophore models were developed by employing CATALYST Pharmacophore Modeling and Analysis tools using substrates of MRPs, including MRP1, -2, -3, -4, -5, -6, -8 and MRPs family, respectively. The models were validated using independent decoy sets generated in DUD-E, and the ones with best AUC (area under the curve) scores were chosen to predict endogenous substrates by screening the Human Metabolome Database (HMDB). A number of molecules obtained by pharmacophore screening have been validated in the literatures. By comparing physical properties (ALOGP, Molecular_PolarSurfaceArea, Molecular_Volume, Molecular_Weight, Num_H_Acceptors, Num_H_Donors) and scaffold features of the screened candidates with the known substrates, we found that: 1) The two sets have consistent ALOGP, Molecule_Volume and Molecule_Weight distribution trend; 2) Substrates of MRP1 have a better lipophilicity than the other subtypes, which is consistent with the two hydrophobic centers on the MRP1 pharmacophore; 3) In the aspect of the scaffold structures, they have the identical or similar backbone fragments.
>
>