Membrane-assisted tariquidar access and binding mechanisms of human ATP-binding cassette transporter P-glycoprotein

Abstract

The human multidrug transporter P-glycoprotein (P-gp) is physiologically essential and of key relevance to biomedicine. Recent structural studies have shed light on the mode of inhibition of the third-generation inhibitors for human P-gp, but the molecular mechanism by which these inhibitors enter the transmembrane sites remains poorly understood. In this study, we utilized all-atom molecular dynamics (MD) simulations to characterize human P-gp dynamics under a potent inhibitor, tariquidar, bound condition, as well as the atomic-level binding pathways in an explicit membrane/water environment. Extensive unbiased simulations show that human P-gp remains relatively stable in tariquidar-free and bound states, while exhibiting a high dynamic binding mode at either the drug-binding pocket or the regulatory site. Free energy estimations by partial nudged elastic band (PNEB) simulations and Molecular Mechanics Generalized Born Surface Area (MM/GBSA) method identify two energetically favorable binding pathways originating from the cytoplasmic gate with an extended tariquidar conformation. Interestingly, free tariquidar in the lipid membrane predominantly adopts extended conformations similar to those observed at the regulatory site. These results suggest that membrane lipids may preconfigure tariquidar into an active ligand conformation for efficient binding to the regulatory site. However, due to its conformational plasticity, tariquidar ultimately moves toward the drug-binding pocket in both pathways, explaining how it acts as a substrate at low concentrations. Our molecular findings propose a membrane-assisted mechanism for the access and binding of the third-generation inhibitors to the binding sites of human P-gp, and offer deeper insights into the molecule design of more potent inhibitors against P-gp-mediated drug resistance.

Keywords: human P-glycoprotein; mechanism of action; membrane lipids; molecular dynamics simulations; tariquidar; third-generation inhibitors.

FIGURE 1

Structural comparison of mouse and human P-gp. (A) Cartoon and surface representation of the X-ray structures of mouse P-gp (PDB ID: 4M1M, 4Q9H). (B) Cartoon and surface representation of the cryo-EM structure of human P-gp in apo state. (PDB ID: 7A65, 7A6E). The N- and C-half of P-gp are shown in cyan and pink, respectively. The approximate position of a lipid bilayer is indicated by gray box. The opening extent of the proposed intramembranous gates are indicated by orange triangles. (C) Cartoon representation of the cryo-EM structure of human P-gp in tariquidar-bound state. The tariquidar molecules bound in the central pocket and access tunnel are shown as green and yellow spheres. (D) Structural comparison of TM4 and TM10 between the apo (gray) and tariquidar-bound (orange) structures of human P-gp. Tariquidar molecules bound in the transmembrane cavity are shown as green and yellow sticks.

FIGURE 2

Structural alignments of the snapshots after 500 ns MD simulations of APO1/2, TCP and TAT onto the cryo-EM structures of human P-gp in apo and singly tariquidar-bound states. (Top panel) Cartoon representation of the MD snapshots (cyan and orange) and cryo-EM structures (white). The approximate position of a lipid bilayer is indicated by gray box. Tariquidar molecules bound in the transmembrane cavity are shown as spheres. (Bottom panel) Extracellular view of the NBDs after structural superposition of TMDs from the extracellular side. For clarity, TM helices are not shown.

FIGURE 3

Conformational dynamics of human P-gp from unbiased MD simulations. (A) The distance between the Cα atoms of residue pair N81TM1−F739TM7 indicates the conformational dynamics of the periplasmic TMD gate region. (B) The distance between the Cα atoms of residue pair D177TM3−N820TM9 indicates the conformational dynamics of the cytoplasmic TMD gate region. (C) The distance between the center of mass of the NBD1 (residues L392 to K619) and NBD2 (residues V1035 to H1254) indicates the opening/closing of the NBD dimer. (Left panel) Cartoon representation of apo human P-gp (PDB id: 7A65) with the initial distances highlighted in black lines. (Middle panel) Time series of the distances from the MD simulations. (Right panel) Distance distributions from the equilibrium states of both simulated replicas (i.e., from 200 to 500 ns) for each system. Initial distances in the cryo-EM structure are indicated by black dashed lines.

FIGURE 4

MD characterization of dynamics in the transmembrane sites. (A) Close-up view of the central binding pocket. (B) Close-up view of the access tunnel site. For clarity, only the cryo-EM structure of human P-gp (PDB id: 7A6E) is shown as cartoon. Tariquidar molecules from the cryo-EM structure (yellow) and the last MD snapshots (magenta and purple) from two simulated replicas are shown as sticks. Residues within 3 Å of tariquidar are shown as white sticks and labeled. Hydrogen bonds observed in the cryo-EM conformation are shown as yellow dashed lines. (C) Time evolution of Cα-RMSD of the residues lining the central pocket (CP) and access tunnel (AT) during the courses of 2 × 500 ns MD simulations for each system. (D) Conformational stability of tariquidar per se in the binding sites indicated by two-dimension distribution of the distance and internal angle indicated in the insets. (E) Time series of the distance between the center of mass of tariquidar and the contacting residues in the corresponding binding site.

FIGURE 5

Environment-dependent conformational preference of tariquidar. (A) Simulation of tariquidar in water. (Left) Time evolution of the radius of gyration (Rgyr) of tariquidar. (Right) Conformational ensembles of tariquidar generated by a total of 3 μs MD simulations in water. (B) Simulation of tariquidar in lipid membrane. (Left) Time evolution of the radius of gyration (Rgyr) of tariquidar. (Right) Conformational ensembles of tariquidar generated by a total of 3 μs MD simulations in lipid membrane. In a and b, three simulations of tariquidar initiated with the folded conformation is shown in the upper panel, and the other three with the extended conformation is shown in the lower panel. (C) The tariquidar conformation and orientation in lipid membrane. The population of each state is indicated. (D) Representative conformations of tariquidar showing the flexibility of the tetrahydroisoquinoline ring.

FIGURE 6

Free energy profiles of tariquidar binding along the putative pathways with different starting points. (A) The two potential binding paths proposed in this work are indicated by black and purple arrows. The gate region is pointed by an orange ball. G, gate region; T, access tunnel; P, central pocket. (B) Path detection by CAVER 3.0 software with a 0.9 Å probe radius, 3 Å shell radius, and 4 Å shell depth. The detected path from the gate region to the central pocket via the access tunnel is shown in purple. The two proposed drug-entry portals laterally open to the membrane are shown in yellow and cyan, respectively. (C) The cytoplasmic entrance gate formed by the kinking of TM4, TM10 and TM12 is indicated by an orange diamond. (D) The starting point structures with different conformations adopted by tariquidar shown as sticks. (E, F) The free energy profiles for tariquidar entering into the transmembrane sites along the PNEB-generated paths.

FIGURE 7

PNEB-optimized pathways. (A) Front view of the two time-independent binding paths of tariquidar are indicated by snapshots. Tariquidar is shown in sticks. For clarity, TM 10 and TM12 are not shown. (B) Conformational transitions of tariquidar along the binding paths indicated by the internal angle as shown in Figure 4. (C) Conformational changes of P-gp TMDs along the binding paths indicated by the Cα-RMSD. (D, E) Conformational changes of the gate-involved TMs in P_G→T→P (D) and P_G→P (E).

FIGURE 8

A membrane-assisted P-gp access and binding model for the third-generation inhibitors. The free tariquidar in aqueous solutions exists in a dynamic equilibrium between the extended and folded conformations, which shifts toward predominantly extended conformations in the lipid membrane. Dual asymmetric binding pathways from the cytoplasmic gate region to the central drug-binding pocket are initiated with an extended tariquidar conformation. At high concentrations, two inhibitor molecules simultaneously bind in the drug-binding pocket and the access tunnel, resulting in inhibitory effects by blocking the conformational changes required for substrate transport. At lower concentrations, the inhibitor only binds at the drug-binding pocket through the two binding pathways identified, leading to the subsequent transport. According to this model, we suggest a two-in-one strategy for the molecular design of more effective P-gp inhibitors.