In this work, molecular docking, pharmacophore modeling and molecular dynamics (MD) simulation were rendered for the mouse P-glycoprotein (P-gp) (code: 4Q9H) and bioflavonoids; amorphigenin, chrysin, epigallocatechin, formononetin and rotenone including a positive control; verapamil to identify protein–ligand interaction features including binding affinities, interaction characteristics, hot-spot amino acid residues and complex stabilities. These flavonoids occupied the same binding site with high binding affinities and shared the same key residues for their binding interactions and the binding region of the flavonoids was revealed that overlapped the ATP binding region with hydrophobic and hydrophilic interactions suggesting a competitive inhibition mechanism of the compounds. Root mean square deviations (RMSDs) analysis of MD trajectories of the protein–ligand complexes and NBD2 residues, and ligands pointed out these residues were stable throughout the duration of MD simulations. Thus, the applied preliminary structure-based molecular modeling approach of interactions between NBD2 and flavonoids may be gainful to realize the intimate inhibition mechanism of P-gp at NBD2 level and on the basis of the obtained data, it can be concluded that these bioflavonoids have the potential to cause herb–drug interactions or be used as lead molecules for the inhibition of P-gp (as anti-multidrug resistance agents) via the NBD2 blocking mechanism in future.

Inhibition of P-glycoprotein (P-gp)'s function may conduct significant changes in the prescription drugs' pharmacokinetic profiles and escalate potential risks in taking place of drug/herb-drug interactions. Computational modeling was advanced to scrutinize some bioflavonoids which play roles in herb-drug interactions as P-gp inhibitors utilizing molecular docking and pharmacophore analyses. Twenty-five flavonoids were utilized as ligands for the modeling. The mouse P-gp (code: 4Q9H) was acquired from the PDB. The docking was operated utilizing AutoDock version 4.2.6 (Scripps Research Institute, La Jolla, CA) against the NBD2 of 4Q9H. The result illustrated the high correlation between the docking scores and observed activities of the flavonoids and the putative binding site of these flavonoids was proposed and compared with the site for ATP. To evaluate hotspot amino acid residues within the NBD2, Binding modes for the ligands were achieved using LigandScout to originate the NBD2-flavonoid pharmacophore models. The results asserted that these inhibitors competed with ATP for binding site in the NBD2 (as competitive inhibitors) including the hotspot residues which associated with electrostatic and van der Waals interactions with the flavonoids. In MD simulation of eight delegated complexes selected from the analyzed flavonoid subclasses, RMSD analysis of the trajectories indicated the residues were stable throughout the duration of simulations.

The data is obtained from exploring the modulatory activities of bioflavonoids on P-glycoprotein function by ligand-based approaches. Multivariate Linear-QSAR models for predicting the induced/inhibitory activities of the flavonoids were created. Molecular descriptors were initially used as independent variables and a dependent variable was expressed as pFAR. The variables were then used in MLR analysis by stepwise regression calculation to build the linear QSAR data. The entire dataset consisted of 23 bioflavonoids was used as a training set. Regarding the obtained MLR QSAR model, R of 0.963, R (2)=0.927, [Formula: see text], SEE=0.197, F=33.849 and q (2)=0.927 were achieved. The true predictabilities of QSAR model were justified by evaluation with the external dataset (Table 4). The pFARs of representative flavonoids were predicted by MLR QSAR modelling. The data showed that internal and external validations may generate the same conclusion.