Rapid Report pubs.acs.org/biochemistry
Drugs Modulate Interactions between the First Nucleotide-Binding Domain and the Fourth Cytoplasmic Loop of Human P‑Glycoprotein Tip W. Loo and David M. Clarke* Departments of Medicine and Biochemistry, University of Toronto, Toronto, Ontario M5S 1A8, Canada S Supporting Information *
ABSTRACT: Drug substrates stimulate ATPase activity of the P-glycoprotein (P-gp) ATP-binding cassette drug pump by an unknown mechanism. Cross-linking analysis was performed to test if drug substrates stimulate P-gp ATPase activity by altering cross-talk at the first transmission interface linking the drug-binding [intracellular loop 4 (S909C)] and first nucleotide-binding domains [NBD1 (V472C or L443C)]. In the absence of lipid (inactive P-gp), only V472C/S909C showed cross-linking. Drugs blocked V472C/S909C cross-linking. In the presence of lipids (active P-gp), drug substrates promoted only L443C/S909C cross-linking. This suggests that drug substrates stimulate ATPase activity through a conformational change that shifts Ser909 away from Val472 and toward Leu443.
P-Glycoprotein (P-gp, ABCB1) is a member of the ABC (ATPbinding cassette) family of proteins, the largest class of membrane transport proteins.1 It is a drug pump that catalyzes the ATP-dependent efflux of lipophilic compounds such as hydrophobic drugs, steroids, peptides, and detergents. The physiological role of P-gp is to protect us from cytotoxic compounds as it is highly expressed in the epithelium of the liver, kidney, and gastrointestinal tract and at the blood−brain or blood−testes barrier where it blocks entry or mediates export of toxins. It is clinically important because it affects absorption, distribution, and clearance of a wide range of drugs and contributes to multidrug resistance in diseases such as cancer. Because of its clinical importance, intensive efforts have been made to understand its structure and mechanism. Human P-gp is a single polypeptide of 1280 amino acids that is organized into four domains: two nucleotide-binding domains (NBDs) and two transmembrane (TM) domains (TMDs) (Figure 1A). Drug substrates bind to sites in the TMDs,2,3 while ATP molecules bind to a pair of sites at the NBD interface between the opposing Walker A and LSGGQ signature sequences.4 Binding of drug substrates to the TMDs stimulates ATPase activity by an unknown mechanism. Crosstalk between the NBDs and TMDs is mediated by transmission interfaces. Transmission interfaces are composed of cytoplasmic extensions of TM segments [intracellular loops (ICLs)] that contact the NBDs to form ball-and-socket joints.5 The NBD1/ICL4 and NBD2/ICL2 contact points (Figure 1A) appear to be critical for ABC drug pumps as some drug pumps like BCRP (ABCG2) lack the NBD1/ICL1 and NBD2/ICL3 contacts.6 In © XXXX American Chemical Society
Figure 1. Cross-linking inhibits the ATPase activity of mutant L443C/ S909C. (A) Model of human P-gp showing the locations of crosslinkable cysteines at the NBD1/ICL4 interface. (B) Histidine-tagged L443C/S909C in membranes treated with (+) and without (−) copper phenanthroline (CuP) was isolated by nickel-chelate chromatography, mixed with sheep brain lipid, and assayed for ATPase activity in the absence (None) or presence of 0.1 mM vinblastine (Vin), 0.3 mM verapamil (Ver), or 1 mM rhodamine B (Rhod) before (−) and after (+) addition of DTT. The inset shows an immunoblot showing that reduction of cross-linked (X-link) product after treatment with DTT. An asterisk indicates a significant difference compared to uncross-linked P-gp (P < 0.001; n = 4).
addition, molecular dynamic studies7 have predicted that in human P-gp there were more contact points between coupling helices at the NBD1/ICL4 and NBD2/ICL2 interfaces than at the NBD1/ICL1 and NBD2/ICL3 interfaces (Figure 1A). It was postulated that the NBD1/ICL4 and NBD2/ICL2 contacts were structurally more important for driving conformational changes needed for drug export.7 We predict that the presence of substrate in the drug-binding pocket in the TMDs stimulates ATPase activity in the NBDs through conformational changes at the NBD1/ICL4 and NBD2/ICL2 interfaces. Received: March 14, 2016 Revised: April 24, 2016
A
DOI: 10.1021/acs.biochem.6b00233 Biochemistry XXXX, XXX, XXX−XXX
Biochemistry
Rapid Report
mutant P-gp was treated with dithiothreitol (DTT). These results suggest that conformational changes at the NBD1/ICL4 interface are critical for activity. We then tested whether drugs could affect cross-linking of the isolated mutants in detergent. The histidine-tagged mutants V472C/S909C and L443C/S909C were purified and then treated with copper phenanthroline in the absence or presence of 0.3 mM verapamil or 0.1 or 1 mM rhodamine B. Samples were then subjected to immunoblot analysis. Mutant V472C/ S909C showed ∼75% cross-linking in the absence of drugs (Figure 2). The level of cross-linking was reduced by >10-fold
Here we tested whether the mechanism of drug stimulation of ATPase activity involves conformational changes at the NBD1/ICL4 ball-and-socket joint using a cysteine mutagenesis and cross-linking approach. The rationale was that drug substrates might alter the cross-linking pattern between cysteines introduced at the NBD1/ICL4 interface as a result of conformational changes in the TMDs in the drug-binding pocket. The NBD1/ICL4 interface was examined because activity is much less sensitive to point mutations than that of the NBD2/ICL2 interface.8,9 For example, point mutations of 18 of 20 eight residues in ICL2 inhibited folding of P-gp compared to 1 of 28 mutations in ICL4.8 In addition, molecular dynamics studies predicted that the largest conformational change during the catalytic cycle occurred at the NBD1/ICL4 interface.10 Reporter cysteines at the NBD1/ICL4 interface were introduced into histidine-tagged Cys-less P-gp to create mutants L443C/S909C and V472C/S909C. The S909C mutation is located in ICL4, while the L443C and V472C mutations are located in NBD1 (Figure 1A). The V472C/ S909C and L443C/S909C mutants were used because they cross-link under different conditions.11 The V472C/S909C mutant shows robust cross-linking in detergent,11 while L443C/S909C P-gp shows cross-linking only in the presence of lipid. Human P-gp is not active in detergent alone.11 Mutant V472C/S909C, however, was included because the equivalent residues in the crystal structures of Caenorhabditis elegans (V496/A950) and mouse (V468/S905) P-gp5,12 are close together (α-carbons are 7.5 and 7.4 Å apart, respectively). When purified, the mutant can be cross-linked in detergent.11 Lipids can activate human P-gp and shift cross-linking of S909C from V472C to L443C.11 Lipids also appear to alter the structure of ABC exporters by bringing the NBDs closer together than what has been reported for the crystal structures of P-gp.13,14 Lipids also appear to act as weak substrates because P-gp can translocate a broad range of lipids such as phosphatidylcholine. P-gp, however, cannot compensate for the lack of MDR3 (phosphatidylcholine transporter) in knockout mice.15,16 Because lipids can act as P-gp substrates, we predicted that drug substrates would also induce conformational changes to inhibit V472C/S909C cross-linking in detergent and promote L443C/S909C cross-linking in lipid. Because both L443C/S909C and V472C/S909C mutants are active (see Figure S1),17 we first tested whether cross-linking at the NBD1/ICL4 interface affected ATPase activity. Membranes prepared from cells expressing histidine-tagged L443C/S909C were treated with or without copper phenanthroline oxidant and P-gp isolated by nickel-chelate chromatography. Crosslinked product can be readily detected on sodium dodecyl sulfate−polyacrylamide gel electrophoresis gels as it migrates with a slower mobility (inset, Figure 1B). By contrast, no crosslinked product was detected when the single cysteines in membranes or in detergent (purified) were treated with oxidant (see Figure S2). The isolated P-gps were then mixed with lipid and assayed for drug-stimulated ATPase activity. Sheep brain lipid was used because it yields basal ATPase activity lower than that seen with Escherichia coli lipid18,19 and contains important lipids such as phosphatidylcholine found in mammalian cells but not in E. coli (Avanti Polar Lipids, Alabaster, AL). Figure 1B shows that cross-linking of L443C/S909C inhibited the ability of verapamil, vinblastine, or rhodamine B to stimulate its ATPase activity. The activities were restored when cross-linked
Figure 2. Drug substrates inhibit V472C/S909C cross-linking in detergent. Histidine-tagged mutant V472C/S909C or L443C/S909C in detergent was cross-linked without (−) or with (+) copper phenanthroline (CuP) in the absence (None) or presence of 0.3 mM verapamil (Ver), 0.1 mM vinblastine (Vin), or 1 mM rhodamine B (Rho). The positions of cross-linked (X-link) and mature P-gp (170 kDa) are shown. An asterisk indicates a significant difference relative to the untreated mutant (P < 0.001; n = 4).
(to ∼6%) by drug substrates. By contrast, mutant L443C/ S909C showed little cross-linking in the absence of lipids. Inhibition of cross-linking was not due to drugs acting like amphiphiles because high levels of the amphiphile DMSO (up to 20% final concentration) did not inhibit cross-linking (see Figure S3). Drug inhibition of V472C/S909C was likely caused by long distance conformational changes rather than binding to the NBD1/ICL4 interface because the drug-binding sites are located within the TM segments.2 Binding of drug substrates has been shown to alter packing of the TM segments and induce helical movements20,21 that are then transmitted to the NBDs.18 We then tested the effect of drug substrates on cross-linking of P-gp in lipid. Histidine-tagged mutants V472C/S909C and L443C/S909C were isolated by nickel-chelate chromatography, mixed with sheep brain lipid, and cross-linked with copper phenanthroline in the absence or presence of drug substrates (Figure 3A). Mutant V472C/S909C showed no detectable cross-linking in the presence of lipid regardless of whether drug was present (Figure 3A). By contrast, drug substrates increased the level of cross-linking of mutant L443C/S909C from ∼10% (no drugs) to ∼60% (verapamil), ∼45% (vinblastine), and ∼38% (rhodamine B) (Figure 3A). The increase in the level of cross-linking was due to drug substrates binding to the drugbinding pocket because only rhodamine B enhanced crosslinking of mutant I306R/L443C/S909C (Figure 3A). It was previously reported that the I306R mutation inhibits binding of verapamil and vinblastine but not rhodamine B.22 The results suggest a mechanism that is responsible for drug stimulation of P-gp ATPase activity is activation of a B
DOI: 10.1021/acs.biochem.6b00233 Biochemistry XXXX, XXX, XXX−XXX
Biochemistry
Rapid Report
In conclusion, we have identified a novel conformational switch at the NBD1/ICL4 interface that is sensitive to both lipid and drug substrates. Coupling of drug binding to stimulation of ATPase activity appears to involve conformational changes at the NBD/ICL interfaces as well as movement between the two wings of the protein. This might be a common mechanism in all ABC proteins.
■
ASSOCIATED CONTENT
* Supporting Information S
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.biochem.6b00233. Details of experimental procedures (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*Department of Medicine, University of Toronto, Toronto, Ontario, Canada. Telephone and fax: 416-978-1105. E-mail:
[email protected]. Funding
This work was supported by a grant from the Canadian Institutes of Health Research (MOP-102620) and Cystic Fibrosis Canada (3014). D.M.C. is the recipient of the Canada Research Chair in Membrane Biology. Notes
Figure 3. Drug substrates promote L443C/S909C cross-linking in lipid. (A) Cross-linking of isolated histidine-tagged mutants V472C/ S909C, L443C/S909C, and I306R/L443C/S909C with (+) or without (−) copper phenanthroline (CuP) in the absence (None) or presence of 0.3 mM verapamil (Ver), 0.1 mM vinblastine (Vin), or 1 mM rhodamine B (Rhod) after addition of lipid. An asterisk indicates a significant difference compared to the mutant cross-linked with no drug (P < 0.001; n = 4). The positions of cross-linked (X-link) and mature P-gp (170 kDa) are shown. (B) Drug substrates cause Ser909 to shift away from Val472 (relaxed state) and move closer to Leu443 (active state).
The authors declare no competing financial interest.
■
REFERENCES
(1) Sharom, F. J. (2006) Biochem. Cell Biol. 84, 979−992. (2) Loo, T. W., Bartlett, M. C., and Clarke, D. M. (2003) J. Biol. Chem. 278, 39706−39710. (3) Loo, T. W., and Clarke, D. M. (1999) J. Biol. Chem. 274, 24759− 24765. (4) Loo, T. W., Bartlett, M. C., and Clarke, D. M. (2002) J. Biol. Chem. 277, 41303−41306. (5) Jin, M. S., Oldham, M. L., Zhang, Q., and Chen, J. (2012) Nature 490, 566−569. (6) Rosenberg, M. F., Bikadi, Z., Chan, J., Liu, X., Ni, Z., Cai, X., Ford, R. C., and Mao, Q. (2010) Structure 18, 482−493. (7) Wise, J. G. (2012) Biochemistry 51, 5125−5141. (8) Loo, T. W., and Clarke, D. M. (2015) J. Biol. Chem. 290, 16954− 16963. (9) Loo, T. W., Bartlett, M. C., and Clarke, D. M. (2013) J. Biol. Chem. 288, 20326−20333. (10) Prajapati, R., and Sangamwar, A. T. (2014) Biochim. Biophys. Acta, Biomembr. 1838, 2882−2898. (11) Loo, T. W., and Clarke, D. M. (2016) Biochem. Biophys. Res. Commun. 472, 379−383. (12) Li, J., Jaimes, K. F., and Aller, S. G. (2014) Protein Sci. 23, 34− 46. (13) Verhalen, B., Ernst, S., Borsch, M., and Wilkens, S. (2012) J. Biol. Chem. 287, 1112−1127. (14) Zoghbi, M. E., Cooper, R. S., and Altenberg, G. A. (2016) J. Biol. Chem. 291, 4453−4461. (15) van Helvoort, A., Smith, A. J., Sprong, H., Fritzsche, I., Schinkel, A. H., Borst, P., and van Meer, G. (1996) Cell 87, 507−517. (16) Romsicki, Y., and Sharom, F. J. (2001) Biochemistry 40, 6937− 6947. (17) Zolnerciks, J. K., Wooding, C., and Linton, K. J. (2007) FASEB J. 21, 3937−3948. (18) Loo, T. W., Bartlett, M. C., and Clarke, D. M. (2003) J. Biol. Chem. 278, 1575−1578. (19) Loo, T. W., Bartlett, M. C., and Clarke, D. M. (2008) J. Biol. Chem. 283, 28190−28197.
conformational switch at the NBD1/ICL4 interface (Figure 3B). In the absence of lipid and drug substrates, P-gp is in a relaxed state where S909C cross-links to V472C. Drug substrates alter the structure to inhibit V472C/S909C crosslinking but promote L443C/S909C cross-linking in the presence of lipid (active state). Alterations in the structure of the NBD1/ICL4 interface would be expected to have an impact on ATPase activity because it is in the proximity of residues involved in ATP binding and hydrolysis [Walker A site and Qloop (Gln475)] (Figure 3B). A recent study of the Q-loop glutamines reported that the NBD1/ICL interfaces were critical for coupling drug binding to the ATP catalytic cycle.23 Conformational changes at the NBD/ICL interfaces have also been predicted in molecular dynamics studies. For example, Wise7 reported that the NBD1/ICL4 coupling site showed a twisting motion during the catalytic cycle. Similarly, Prajapati et al. reported that the largest angle change between the α-carbons of the NBD/ICL residues was between ICL4 and NBD1.10 During simulation from an open (NBDs apart) to a closed conformation (NBDs close together), intracellular helix 4 (ICH4) (intracellular helix interacting with NBD1) underwent a rotation of approximately 82° about the horizontal axis, while the changes in other ICHs interacting with the NBDs were approximately 20−40°. C
DOI: 10.1021/acs.biochem.6b00233 Biochemistry XXXX, XXX, XXX−XXX
Biochemistry
Rapid Report
(20) Loo, T. W., Bartlett, M. C., and Clarke, D. M. (2003) J. Biol. Chem. 278, 13603−13606. (21) Szewczyk, P., Tao, H., McGrath, A. P., Villaluz, M., Rees, S. D., Lee, S. C., Doshi, R., Urbatsch, I. L., Zhang, Q., and Chang, G. (2015) Acta Crystallogr., Sect. D: Biol. Crystallogr. 71, 732−741. (22) Loo, T. W., Bartlett, M. C., and Clarke, D. M. (2004) Biochemistry 43, 12081−12089. (23) Zolnerciks, J. K., Akkaya, B. G., Snippe, M., Chiba, P., Seelig, A., and Linton, K. J. (2014) FASEB J. 28, 4335−4346.
D
DOI: 10.1021/acs.biochem.6b00233 Biochemistry XXXX, XXX, XXX−XXX
