Identification of Residues within the Drug-binding Domain of the Human Multidrug Resistance P-glycoprotein by Cysteine-scanning Mutagenesis and Reaction with Dibromobimane*

P-glycoprotein (P-gp) can transport a wide variety of cytotoxic compounds that have diverse structures. Therefore, the drug-binding domain of the human multidrug resistance P-gp likely consists of residues from multiple transmembrane (TM) segments. In this study, we completed cysteine-scanning mutagenesis of all the predicted TM segments of P-gp (TMs 1–5 and 7–10) and tested for inhibition by a thiol-reactive substrate (dibromobimane) to identify residues within the drug-binding domain. The activities of 189 mutants were analyzed. Verapamil-stimulated ATPase activities of seven mutants (Y118C and V125C (TM2), S222C (TM4), I306C (TM5), S766C (TM9), and I868C and G872C (TM10)) were inhibited by more than 50% by dibromobimane. The activities of mutants S222C (TM4), I306C (TM5), I868C (TM10), and G872C (TM10), but not that of mutants Y118C (TM2), V125C (TM2), and S776C (TM9), were protected from inhibition by dibromobimane by pretreatment with verapamil, vinblastine, or colchicine. These results and those from previous studies (Loo, T. W. and Clarke, D. M. (1997)J. Biol. Chem. 272, 31945–31948; Loo, T. W. and Clarke, D. M. (1999) J. Biol. Chem. 274, 35388–35392) indicate that the drug-binding domain of P-gp consists of residues in TMs 4, 5, 6, 10, 11, and 12.

The human multidrug resistance P-glycoprotein (P-gp) 1 is located in the plasma membrane and uses ATP to pump a wide variety of structurally diverse cytotoxic compounds out of the cell (1,2). Expression of P-gp is relatively high in the epithelial cells of the gastrointestinal tract, renal proximal tubules, biliary tract, and capillaries of the brain and testes (3,4). The pattern of P-gp expression in tissues and studies on P-gp knockout mice indicate that the main physiological role of P-gp is to protect the organism from toxic xenobiotics (5,6). The protective role of P-gp contributes to the phenomenon of multidrug resistance during cancer and AIDS chemotherapy because many of the therapeutic compounds are also substrates of P-gp (7)(8)(9). P-gp is a member of the ATP-binding cassette family of transporters (10,11), and its 1280 amino acids are organized as two repeating units joined by a linker region of about 60 amino acids (12). Each repeat consists of an NH 2 -terminal hydrophobic domain containing six transmembrane (TM) segments followed by a hydrophilic domain containing an ATP-binding site (13,14).
The exact mechanism of how P-gp functions is unknown. It is known, however, that both halves of the molecule are essential for activity (15) and that both nucleotide-binding domains can bind and hydrolyze ATP and are essential for function (15)(16)(17)(18)(19). The drug-binding domain is located in the TM domains of P-gp because drug substrates will bind to a deletion mutant lacking both nucleotide-binding domains (20).
An important step in understanding the mechanism of P-gp is to determine the residues in the drug-binding domain. A common method for identifying residues in a membrane transporter that are critical for substrate binding and/or transport is to use alanine-scanning mutagenesis. This has been used successfully for transporters such as bacteriorhodopsin (21) and the SERCA1 calcium pump (22). Recent crystal structures of these two transporters showed that the amino acids involved in ligand binding are in agreement with those identified through mutational analyses (23,24). The residues in P-gp that are involved in drug binding, however, have been difficult to characterize because a large number of mutations throughout the molecule can alter the substrate specificity (25)(26)(27)(28)(29)(30)(31). It has been difficult to determine whether mutations that affected activity were actually close to the drug-binding site or whether they affected the global structure of the protein (32).
To avoid these difficulties, we used a direct assay involving cysteine-scanning mutagenesis and modification with a thiolreactive substrate to identify residues in the TM segments that are critical for drug binding. The rationale is that the thiolreactive substrate, dibromobimane (dBBn), will enter the drugbinding site of P-gp, covalently label any adjacent cysteine residue, and inhibit activity. Inhibition by dBBn should be preventable by pretreatment with other substrates such as verapamil, colchicine, and vinblastine if the reactive residue is in the drug-binding domain. We have used this method and identified the residues important for drug binding in TMs 6, 11, and 12 (33,34). In this study, we identify important residues in the remaining TM segments (TMs 1-5 and 7-10) that are important for activity.

EXPERIMENTAL PROCEDURES
Construction of Mutants-Wild-type P-gp has cysteine residues at positions 137, 431, 717, 956, 1074, 1125, and 1227. None of these cysteines are important for activity because mutation of all cysteines to alanine (Cys-less P-gp) resulted in an active molecule (13). The Cys-less P-gp cDNA was modified to code for 10 histidine residues at the COOH end of the molecule (Cys-less P-gp(His) 10 ). This facilitated purification of the Cys-less P-gp by nickel-chelate chromatography (35). Cysteine residues were then introduced into the Cys-less P-gp(His) 10 as described previously (36). The integrity of the mutated cDNA was confirmed by sequencing the entire cDNA (37).
Expression and Purification of P-glycoprotein-Expression and purification of histidine-tagged P-gp mutants were done as described previously (35). Briefly, fifty 10-cm-diameter culture plates of HEK 293 cells were transfected with the mutant cDNA. After 24 h, the medium was replaced with fresh medium containing 10 M cyclosporin A. P-gp was expressed in the presence of cyclosporin A because it is a drug substrate that promotes maturation of the protein (38). The transfected cells were harvested 24 h later and solubilized with 1% (w/v) n-dodecyl-␤-D-maltoside, and the mutant P-gp was isolated by nickel-chelate chromatography (Ni-NTA columns; Qiagen, Inc., Mississauga, Canada).
Measurement of Drug-stimulated ATPase Activity-The P-gp(His) 10 mutants were eluted from the nickel columns with buffer containing 10 mM Tris-HCl, pH 7.5, 500 mM NaCl, 300 mM imidazole (pH 7.0), 0.1% (w/v) n-dodecyl-␤-D-maltoside, and 10% (v/v) glycerol and mixed with an equal volume of 10 mg/ml sheep brain phosphatidylethanolamine (Type II-S; Sigma-Aldrich) that was washed and suspended in 10 mM Tris-HCl, pH 7.5, and 150 mM NaCl. The P-gp:lipid mixture was then sonicated for 45 s at 4°C (bath-type probe; maximum setting, Branson Sonifier 450; Branson Ultrasonic, Danbury, CT). An aliquot of the sonicated P-gp:lipid mixture was assayed for drug-stimulated ATPase activity by addition of an equal volume of buffer containing 100 mM Tris-HCl, pH 7.5, 100 mM NaCl, 20 mM MgCl 2 , 10 mM ATP, and the desired drug substrates (2 mM verapamil, 0.2 mM vinblastine, or 10 mM colchicine). The samples were incubated for 30 min at 37°C, and the amount of inorganic phosphate liberated was determined by the method of Chifflet et al. (39).
For inhibition with dBBn, the P-gp:lipid mixture was preincubated with 2 mM dBBn (Molecular Probes, Inc.) for 5 min at 37°C. The reaction was stopped by the addition of cysteine, pH 7.5, to a final concentration of 40 mM. Drug-stimulated ATPase activity was then determined as described above. In the protection experiments, the P-gp:lipid samples were treated with 2 mM verapamil, 0.2 mM vinblastine, or 10 mM colchicine for 15 min at 4°C before the addition of dBBn (the final concentration of dBBn was 0.2 mM or 2 mM). These were saturating substrate concentrations for stimulation of ATPase activity, whereas 0.2 mM is the K m for stimulation by dBBn (33,34). The samples were then incubated with 0.2 or 2 mM dBBn for 5 min at 37°C, and the reaction was stopped by the addition of cysteine, pH 7.5 (final concentration, 40 mM). Drug-stimulated ATPase activity was then determined with the same drug used for protection (final concentrations were 1 mM verapamil, 0.1 mM vinblastine, or 5 mM colchicine).
Immunoblot Analysis-The purified histidine-tagged P-gp was subjected to SDS-polyacrylamide gel electrophoresis, transferred onto a sheet of nitrocellulose, and probed with a rabbit polyclonal antibody against P-gp, followed by enhanced chemiluminescence (36).

Verapamil-stimulated ATPase Activity of the Cys Mutants-
There is increasing evidence that the drug-binding site is likely to be within the TM segments (20, 40 -44). The amino acids predicted to be in TM segments are shown in Fig. 1. We had reported that residues within TMs 6, 11, and 12 were impor-tant for drug binding (33,34). In this study, we used cysteinescanning mutagenesis of the remaining TM segments and inhibition by dBBn to identify residues that contribute to the drug-binding domain. Accordingly, each residue in TMs 1, 2, 3, 4, 5, 7, 8, 9, and 10 was changed to cysteine, and then we determined whether the mutant protein was active. The mutant P-gps were expressed in HEK 293 cells, purified by nickelchelate chromatography, and assayed for verapamil-stimulated ATPase activity. We chose verapamil as the substrate because it had the highest stimulation of the ATPase activity of Cys-less P-gp (about 13-fold; Ref. 33). Substrate-stimulated ATPase activity has been shown to correlate well with transport because the turnover numbers are similar (45). Fig. 2 shows the verapamil-stimulated ATPase activity of each Cys mutant after treatment with dibromobimane. A total of 189 Cys mutants were analyzed. Six mutants, V52C (TM1), G54C (TM1), G62C (TM1), G122C (TM2), G763C (TM8), and P866C (TM10), were not analyzed because of low expression or very low activity. The mutants V52C, G54C, and G62C appeared to be defective in processing and were not efficiently expressed compared with Cys-less P-gp(His) 10 . Blots of whole cell extracts of HEK 293 cells expressing these mutants showed very low amounts of the immature (150-kDa) form of P-gp (data not shown). Expression of these TM1 mutants in the presence of cyclosporin A induced maturation of the 150-kDa protein to the mature 170-kDa P-gp, but the yields were too low to allow accurate measurement of drug-stimulated ATPase activity. The remaining 186 P-gp mutants were expressed as well as Cys-less P-gp(His) 10 . Three of these mutants (G122C, G763C, and P866C) exhibited less than 20% of the verapamil-stimulated ATPase activity of Cys-less P-gp(His) 10 . Mutants G122C (TM2), G763C (TM8), and P866C (TM10) had 9%, 11%, and 6%, respectively, of the verapamil-stimulated ATPase activity of Cys-less P-gp(His) 10 . It is interesting that mutants G122C (TM2) and G763C (TM8) occupy equivalent positions in each half of P-gp. We reported previously that mutation of P866A caused drastic alteration in the substrate specificity of P-gp (25).
Inhibition by Dibromobimane-The 183 mutants that had expression levels similar to that of Cys-less P-gp(His) 10 and had greater than 20% of the verapamil-stimulated ATPase activity of Cys-less P-gp(His) 10 were tested for inhibition by dBBn. Dibromobimane is a substrate of P-gp, and its reaction with a cysteine within the drug-binding site would be expected to result in the inhibition of verapamil-stimulated ATPase activity. The mutant P-gps were isolated by nickel-chelate chromatography, mixed with lipid, sonicated, and incubated for 5 min at 37°C with dBBn. The reaction was stopped by the addition of cysteine to quench the reactivity of dBBn and con- Histidine-tagged mutants were purified by nickel-chelate chromatography, mixed with lipid, sonicated, and then incubated for 5 min at 37°C with or without 2 mM dBBn. The reaction was quenched by the addition of cysteine, and then the verapamil-stimulated ATPase activity was determined. The results (percentage of untreated) are expressed relative to that of a mock-treated sample. The verapamil-stimulated activity (with no dBBn) of each mutant relative to Cys-less P-gp(His) 10 is shown in the lower box. Each value is the average of two different purifications. The deviation from the mean is shown. A-I show the activities of mutants in TMs 1, 2, 3, 4, 5, 7, 8, 9, and 10, respectively. ND, not done because of very low activity.
vert it to a compound that was no longer a substrate of P-gp (33). The verapamil-stimulated ATPase activities of the dBBntreated samples were assayed and compared with that of mocktreated samples.
The results for each TM segment are shown in Fig. 2, A-I. In TM1 ( Fig. 2A), the active mutants retained more than 80% of their ATPase activity after treatment with dBBn. Two mutants in TM2 (Fig. 2B) were substantially inhibited by dBBn. Mutants Y118C and V125C were inhibited 84% and 67%, respectively, by dBBn. The results for TM3 (Fig. 2C) were similar to those in TM1 in that all of the mutants retained more than 80% of their activity after treatment with dBBn. In TM4 (Fig. 2D), the activity of one mutant, S222C, was inhibited by 75% after exposure to dBBn. In TM5 (Fig. 2E), there were modest decreases in activity (about 25-30%) for mutants A295C and S298C, whereas the remaining mutants retained more than 80% of their activity after treatment with dBBn. Although mutant I306C (TM5) had only 26% of the verapamil-stimulated ATPase activity of Cys-less P-gp(His) 10 , its basal ATPase activity was increased (about 2-fold) upon treatment with dBBn (data not shown). Its verapamil-stimulated ATPase activity also increased to 130% after treatment with dBBn. TM7 (Fig. 2F), the first TM in the COOH half of P-gp, showed results similar to those for TM1 in that all the mutants were quite resistant to inhibition by dBBn (all retained greater than 80% of their activity). Similarly, the pattern of inhibition in TM8 mutants (Fig. 2G) was very similar to that seen with TM2 mutants (Fig. 2B). Mutating the residues at the fifth position, G122C (TM2) and G736C (TM8), resulted in significant reductions in the verapamil-stimulated ATPase activities (9% and 11%, respectively), whereas mutating the residues at the eighth position, V125C (TM2) and S766C (TM8), resulted in mutants that were sensitive to inhibition by dBBn (67% and 81% inhibition, respectively). The Cys mutants in TM9 (Fig.  2H) were relatively resistant to inhibition by dBBn, with the most sensitive mutants, A834C and V835C, showing about 25% inhibition. Two mutants, I868C and G872C in TM10 (Fig. 2I), were very sensitive to inhibition by dBBn. Their activities were inhibited by 80% and 75%, respectively. The concentrations of dBBn required to give 50% inhibition of verapamil-stimulated ATPase activity for mutants Y118C (TM2), V125C (TM2), S222C (TM4), S766C (TM8), I868C (TM10), and G872C (TM10) were 740, 870, 340, 92, 80 and 62 M, respectively. Immunoblot analysis of cells expressing these mutants P-gps showed that the expression was similar to that of Cys-less P-gp (Fig. 3). In all cases, the major product was the mature 170-kDa protein.
Protection from Inhibition by dBBn-Although the initial results identified residues within the predicted TM segments that were sensitive to inhibition by dBBn, it was possible that these cysteines were very accessible to dBBn and may not be close to the drug-binding site. To ensure that inhibition was indeed due to inhibition of a residue in the drug-binding site, it was important to show that the presence of a substrate would protect the mutant P-gp from inhibition by dBBn. Accordingly, three substrates, verapamil, vinblastine, and colchicine, were selected for the protection experiments. Verapamil was selected because it was used in the inhibition assays and showed the highest stimulation of activity. Vinblastine and colchicine were also included because they are structurally different from verapamil and have traditionally been used to study mutants of P-gp (25)(26)(27)46). We have used colchicine and vinblastine to characterize the residues in TMs 6, 11, and 12 that were sensitive to inhibition by dBBn (33,34). Six mutants that were inhibited by dBBn, Y118C (TM2), V125C (TM2), S222 (TM4), S766C (TM8), I868C (TM10), and G872C (TM10), were tested for their ability to be protected from inhibition by dBBn in the presence of substrate. Mutant I306C (TM5) was also chosen because its basal ATPase activity was increased after treatment with dBBn (data not shown). Fig. 4A shows that dBBn treatment also inhibited the vinblastine-and colchicine-stimulated ATPase activities of the mutants, with the exception of mutant I306C. Dibromobimane inhibited the vinblastine-stimulated ATPase activity of I306C, whereas the verapamil-and colchicine-stimulated activities were increased (130% and 125%, respectively).
To test for protection by drug substrates, the mutants were preincubated with verapamil (2 mM), vinblastine (0.2 mM), or colchicine (10 mM) and then treated with dBBn for 5 min at 37°C, and the reaction was quenched with cysteine. The mutants (S766C, I868C, and G872C) that were more sensitive to low concentrations of dBBn were treated with 0.2 mM dBBn, whereas the others (Y118C, V125C, S222C, and I306C) were treated with 2 mM dBBn. Drug-stimulated ATPase activity was then determined with the same drug used for protection. The results are shown in Fig. 4B. Most protection by substrate was observed for mutant S222C (TM4). The mutant was almost completely protected from inactivation by dBBn in the presence of vinblastine. Similarly, vinblastine protected mutant G872C (TM10) from inactivation by dBBn because more than 80% of the activity remained. G872C (TM10) was modestly protected by verapamil, whereas colchicine offered little protection. All three substrates also showed modest protection of mutant I868C (TM10). Some protection from dBBn inhibition by vinblastine was observed for mutant I306C. There was little or no protection by any of the drug substrates for mutants Y118C (TM2), V125C (TM2), and S766C (TM8). We were not able to accurately determine whether the basal ATPase activity of mutants Y118C, V125C, or S766C was inhibited dBBn because the basal ATPase activities were too low.
To test whether differences in membrane fluidity affected the drug protection characteristics of the mutants, the assays were repeated at room temperature. The samples were preincubated with verapamil, vinblastine, or colchicine for 15 min at room temperature and treated with dBBn for 15 min at room temperature, and then the ATPase activity was measured at 37°C. The results were similar to those shown in Fig. 4 (data not shown). DISCUSSION We studied drug-binding by measuring inhibition of ATPase activity of P-gp mutants containing a single cysteine. Cys-less P-gp is a good model system because the mutant could still confer resistance to a wide variety of cytotoxic substrates (13), and it retains about 80% of the drug-stimulated ATPase activity of wild-type enzyme (35). A potential drawback of this approach is that the assay measures only inhibition of ATPase activity and may not reflect drug transport activity. However, drug-stimulated ATPase activity appears to reflect transport activity because the turnover number for vinblastine-stimulated ATPase activity correlates with vinblastine transport out of the cell (45).

FIG. 4. Protection of mutants from dBBn inhibition by drug substrates.
A, purified histidine-tagged Cys-less P-gp(His) 10 and mutant P-gp(His) 10 (Y118C, V125C, S222C, I306C, S766C, I868C, and G872C) were mixed with lipid, sonicated, and then incubated with or without 2 mM dBBn (Y118C, V125C, S222C, and I306C) or 0.2 mM dBBn (S766C, I868C, and G872C) for 5 min at 37°C. The reactions were quenched with cysteine and then verapamil (Ver.)-, vinblastine (Vin.)-, or colchicine (Colch.)-stimulated ATPase activity was determined. The results are expressed relative to that of a sample that was not treated with dBBn. B, the purified Cys-less and mutant P-gps were mixed with lipid, sonicated, and then incubated for 15 min at 4°C without drug or with 2 mM verapamil (Ver.), 0.2 mM vinblastine (Vin.), or 10 mM colchicine (Colch.). The samples were then treated for 5 min at 37°C with or without 2 mM (Y118C, V125C, S222C, and I306C) or 0.2 mM dBBn (S766C, I868C, and G872C), and the reaction was quenched by the addition of cysteine. The ATPase activity was then determined, and the results are expressed relative to that of a sample that was not treated with dBBn. Each value is the average of four different experiments. The specific activity for Cys-less P-gp was 1.4, 0.8, and 0.7 mol/min/mg P-gp with verapamil, vinblastine, and colchicine, respectively. This study shows that residues S222C (TM4), I306C (TM5), I868C (TM10), and G872C (TM10) may be important for drug binding. We have reported that residues L339C and A342C in TM6, F942C and T945C in TM11, and L975C, V982C, and A985C in TM12 are important for drug binding (33,34). In trying to understand how residues that are quite distant from each other could form the drug-binding site, we have aligned the residues in each TM segment as ␣-helical wheels. In Fig. 5, we have aligned the residues that are sensitive to inhibition by dBBn such that they face toward the center of the molecule. It is interesting to note that the dBBn-sensitive residues also lie on one face of the helix. Fig. 5 also takes into consideration the results from disulfide cross-linking studies that show that TMs 4 -6 and 10 -12 are close to one another (47).
Residues S222 (TM4) and I868 (TM10) were protected from inhibition by dBBn by verapamil, vinblastine, and colchicine. This indicates that S222 and I868 may be common to the binding of all three substrates. In contrast, residue G872 (TM10), which lies on the same helical face as I868 (TM10), was protected by verapamil and vinblastine but not by colchicine and may be involved in the binding of verapamil and vinblastine. Residue I306 may be important for drug binding because of the effect of dBBn on the vinblastine-stimulated ATPase activity of mutant I306C. Dibromobimane inhibited the vinblastine-stimulated activity of this mutant, but not its verapamil-or colchicine-stimulated activity. Vinblastine also protected the mutant from inactivation by dBBn. Disulfide crosslinking data support TM5 being part of the drug-binding domain. Increases in verapamil-and colchicine-stimulated activities after reaction of residues I306C with dBBn could also be an indication that I306 is close to or within the drug-binding site. It is also possible that modification of a residue within or close to the drug-binding domain could lead to inhibition or stimulation of ATPase activity. Residues Ala 295 (TM5) and Ile 299 (TM5) lie on either side of Ile 306 in the helical wheel model (Fig. 5) and could be cross-linked to residues in TM12 (47). Cross-linking of I299C (TM5) to residues in TM12 was inhibited by vinblastine and promoted by verapamil and colchicine.
Assignment of TMs 4, 5, and 10 as part of the drug-binding domain of P-gp is consistent with the results obtained from mutation and labeling studies with radioactive analogs of photoactive substrates. Mutation of prolines at equivalent positions in TM4 (P223) and TM10 (P866) caused large alterations in the substrate specificity of the protein (25). The mutants L210I, K209E, and I214T in TM4 of mouse mdr3 (31) and I299M (48) and L305A, S, or T in TM5 of human P-gp also changed the substrate specificity of the transporter (49). The yeast homolog of human P-gp is Pdr5. It is also an ATP-binding cassette transporter, and many of the substrates of P-gp are also substrates of Pdr5 (50). Random mutagenesis studies on Pdr5 showed that residue 1360 in predicted TM10 was important for interaction with FK506 (51). This residue aligns with residue Leu 861 (TM10) of human P-gp and faces toward the drug-binding domain in Fig. 5. TMs 4 and 5 are also labeled by photoactive derivatives of iodomycin (43) and forskolin (52).
DBBn also inhibited P-gp by reacting with cysteines in TM2 (Y118C and V125C) and TM8 (S766C). The drug substrates verapamil, vinblastine, and colchicine did not protect these mutants from inhibition by dBBn. It is possible that some of these residues do not lie within the drug-binding site for these compounds because relatively high concentrations of dBBn were needed to inhibit 50% of the activity of mutants Y118C (TM2) and V125C (TM2) (740 and 870 M, respectively). Another possibility is that modification of mutant Y118C, V125C, or S766C blocks an essential conformational change during coupling of drug binding to ATPase activity. Residue Tyr 118 lies close to a conformationally sensitive residue, Arg 113 (53). It is also possible that these residues lie within another drug-binding site or modulator site. The results from several studies FIG. 5. Proposed working model of the drug-binding domain of P-gp. The residues in TMs 4 -6 and 10 -12 are arranged as ␣-helical wheels as viewed from the cytoplasmic side of the membrane. The residues that are sensitive to inhibition by dBBn and that are protected from inhibition by drug substrates are shown as black-filled circles. The predicted drugbinding domain for dBBn is shown. The dBBn-sensitive residues in TMs 6, 11, and 12 were identified in previous studies (33,34).
suggest that P-gp may contain up to four different drug interaction sites (54 -56). In one model of transport, the substrates occupy two different drug-binding sites during transport (54). In other models, different substrates interact with separate binding sites (56). In both models, ATP hydrolysis leads to drug efflux. It has been reported that two to three molecules of ATP are hydrolyzed for every molecule of vinblastine transported (45).
Another way of explaining how P-gp can transport such a large variety of structurally different compounds is to assume that different residues in the TMs contribute to the binding of a particular substrate. This may be possible because of the flexibility of the TM segments. Studies have shown that the TMs are quite flexible at 37°C because cross-linking results indicate that residues in TM6 can cross-link residues in TMs 10, 11, and 12, whereas residues in TM12 can cross-link residues in TMs, 4, 5, and 6 (36,47). It is possible that binding of a particular compound causes an "induced-fit" binding site by reducing the flexibility of the TMs and thereby contributing different residues to the binding of a particular substrate. Such an induced-fit binding site model has also been proposed for the BmrR transcription factor that can bind a wide variety of compounds (57). This model is also consistent with the observation that various structurally diverse drug substrates can induce correct folding of P-gp-processing mutants that have mutations in all parts of the molecule (38). The drug substrates can do this by inducing superfolding of the transmembrane domains (58). The ability to rescue P-gp-processing mutants with different substrates indicates that there is at least one site or overlapping site that can be occupied by most substrates. This is supported by the finding that several residues (this study and Refs. 33 and 34) can be protected from inhibition by dBBn with verapamil, vinblastine, and colchicine.
Further work with other substrates and compounds that are not transported by P-gp will be needed to clarify whether P-gp contains one drug-binding site with broad specificity or multiple sites with distinct specificities.