Defining the Drug-binding Site in the Human Multidrug Resistance P-glycoprotein Using a Methanethiosulfonate Analog of Verapamil, MTS-verapamil*

Defining the residues involved in the binding of a substrate provides insight into how the human multidrug resistance P-glycoprotein (P-gp) can transport a wide range of structurally diverse compounds out of the cell. Because verapamil is the most potent stimulator of P-gp ATPase activity, we synthesized a thiol-reactive analog of verapamil (MTS-verapamil) and used it with cysteine-scanning mutagenesis to identify the reactive residues within the drug-binding domain of P-gp. MTS-verapamil stimulated the ATPase activity of Cys-less P-gp and had a K m value (25 m M ) that was similar to that of verapamil. 252 P-gp mutants containing a single cysteine within the predicted transmembrane (TM) segments were expressed in HEK 293 cells and purified by nickel-chelate chromatography and assayed for inhibition by MTS-verapamil. The activities of 15 mutants, Y118C (TM2), V125C (TM2), S222C (TM4), L339C (TM6), A342C (TM6), A729C (TM7), A841C (TM9), N842C (TM9), I868C (TM10), A871C (TM10), F942C (TM11), T945C (TM11), V982C

The human multidrug resistance P-glycoprotein (P-gp) 1 uses ATP to pump a wide variety of cytotoxic compounds out of the cell (1,2). Overexpression 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 (3)(4)(5).
Although P-gp is normally expressed in many tissues, its physiological function is unknown. The pattern of P-gp expression in tissues and studies on P-gp "knock-out" mice indicate that it may protect the organism from toxic compounds in our diet (6 -8).
P-gp is a member of the ABC (ATP-binding cassette) family of transporters (9,10). Its 1280 amino acids are organized as two repeating units of 610 amino acids that are joined by a linker region of about 60 amino acids (11). There are six transmembrane (TM) segments and a hydrophilic domain containing an ATP-binding site in each repeat (12,13).
The exact mechanism of how P-gp functions is unknown, but P-gp has been studied quite extensively and serves as a model for understanding the mechanism of other ABC transporters. It is known, however, that both halves of P-gp are essential for activity (14) and that both nucleotide-binding domains can bind and hydrolyze ATP and are essential for function (14 -18).
In understanding the mechanism of P-gp, it is important to determine the location of residues that contribute to the drugbinding domain. Important clues that the transmembrane domains may be important for drug binding were obtained from mutational studies and from labeling studies with photoactive substrate analogs (19 -26). It was later shown that the TM domains alone were sufficient for drug binding, because a deletion mutant lacking both nucleotide-binding domains could still bind drug substrates (27).
Initial attempts to identify the residues that line the drugbinding pocket within the TM domains involved the use of cysteine-scanning mutagenesis and reactivity with a thiol reactive substrate, dibromobimane (dBBn). Several residues in TMs 4, 5, 6, 10, 11, and 12 reacted with dBBn, suggesting that these TMs contribute to the binding of substrate (28 -30).
It is not known if P-gp has many distinct or overlapping binding sites (31)(32)(33)(34)(35). One step toward understanding this problem is to test whether other substrates of P-gp react with the same residues in the predicted drug-binding domain. Verapamil is a particularly important substrate used in studying P-gp. The advantage of using verapamil is that it shows the greatest ability to stimulate the ATPase activity of P-gp. Development of a thiolreactive derivative of verapamil would be a useful tool for characterizing the drug-binding domain of P-gp, because reactive cysteines can be specifically protected with verapamil. In this study, a thiol-reactive methanethiosulfonate analog of verapamil (MTS-verapamil) was synthesized and used to identify residues in the drug-binding domain of P-gp.
Expression and Purification of P-glycoprotein-Expression and purification of histidine-tagged P-gp mutants were done as described previously (36). Briefly, 50 culture plates (10 cm diameter) of HEK 293 cells were transfected with the mutant cDNA, and the medium was replaced 24 h later with fresh medium containing 10 M cyclosporin A. Cyclosporin A is a substrate of P-gp and acts as a powerful chemical chaperone for promoting maturation of P-gp (38). The transfected cells were then harvested 24 h later, solubilized with 1% (w/v) n-dodecyl-␤-Dmaltoside, and the mutant P-gp isolated by nickel-chelate chromatography (nickel-nitrilotriacetic acid columns, Qiagen, Inc., Mississauga, Ontario, 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 2 mM verapamil. The samples were incubated for 30 min at 37°C, and the amount of inorganic phosphate liberated was determined previously (39).
For inhibition with MTS-verapamil (synthesized by Toronto Research Chemicals, Toronto, Ontario), the P-gp/lipid mixture was preincubated with 0.5 mM MTS-verapamil for 10 min at 22°C. The unreacted MTS-verapamil was then removed by gel filtration (Centri.Spin 20 columns, Princeton Separations, Inc., Adelphia, NJ). The columns were pre-equilibrated with nickel column elution buffer containing 5 mg/ml sheep brain phosphatidylethanolamine and 1 mM verapamil. The verapamil-stimulated ATPase activity in the flowthrough fraction was measured as described above.
In the protection experiments, the P-gp:lipid samples were treated with 2 mM verapamil for 5 min at 22°C before addition of 50 M MTS-verapamil. This is the saturating concentration of verapamil for stimulation of ATPase activity, whereas 50 M is above the K m (25 M) for stimulation by MTS-verapamil in Cys-less P-gp. The samples were then incubated for 10 min at 22°C, and unreacted MTS-verapamil was removed by gel filtration (Centri.Spin 20 columns). Verapamil-stimulated ATPase activity was then measured (final verapamil concentration, 1 mM). Fig. 1A. To convert verapamil to a thiol-reactive compound, a methanethiosulfonate group was attached to the methylamine group in verapamil by an ethyl linker arm (Fig. 1B). Alkylthiosulfonates react selectively with cysteines in a protein resulting in a disulfide attachment of the R group and release of a sulfinic acid byproduct (40,41). They generally react more rapidly with cysteines than with other thiol-specific compounds such as maleimides or iodoacetates (40).

Effect of MTS-verapamil on ATPase activity of the Cys-less P-gp-The structure of verapamil is shown in
To test whether MTS-verapamil was still a substrate of P-gp, we assayed for stimulation of ATPase activity. It has been shown that substrate-stimulated ATPase activity correlates with transport activity, because the turnover numbers are quite similar (42). Measurement of MTS-verapamil-stimulated ATPase activity was done on Cys-less P-gp in which all endogenous cysteines were changed to alanine (16). The Cys-less P-gp retains the ability to confer drug resistance and exhibits drug-stimulated ATPase activity (12). Wild-type P-gp could not be used for measuring MTS-verapamil-stimulated ATPase activity because of two highly reactive cysteine residues in the Walker A consensus sites. These cysteine residues are readily modified by N-ethylmaleimide resulting in an inactive enzyme (16).
Cys-less P-gp (His) 10 was expressed in HEK 293 cells and isolated by nickel-chelate chromatography. The isolated P-gp was reconstituted with lipid, and ATPase activity was measured in the presence of varying concentrations of verapamil or MTS-verapamil. Both compounds showed maximal stimulation of ATPase activity, which occurred at about 600 M, and had similar K m values (about 25 M) for stimulation (Fig. 2). MTSverapamil, however, showed lower maximal stimulation of ATPase activity than verapamil (9.6-and 13.3-fold for MTSverapamil and verapamil, respectively). This indicated that Cys-less P-gp had a similar affinity for verapamil and MTSverapamil and that they may bind at the same binding site.
Inhibition of Activity of Single Cys P-gp Mutants by MTSverapamil-A cysteine residue was introduced at each position in the predicted TM segments of Cys-less P-gp (His) 10 . A total of 252 single-cysteine mutants were constructed (28 -30). Each mutant was expressed in HEK 293 cells, and the mutant pro- In TM1 (Fig. 3A), three mutants, V52C, G54C, and G62C, were not assayed because of very low yield of P-gp protein. The remaining 18 mutants showed little or no inhibition by MTSverapamil, because they all retained Ͼ80% of their activity relative to the mock-treated sample.
In TM2, one mutant, G122C, was not tested because of low expression. In this TM segment, two mutants, Y118C and V125C, were inhibited more than 50% by MTS-verapamil (Fig.  3B). The concentrations of MTS-verapamil required to inhibit 50% of the ATPase activity (I 50 ) for mutants Y118C and V125C were 40 and 63 M, respectively.
The TM3 Cys mutants were relatively insensitive to inhibition by MTS-verapamil. All of these mutants retained more than 75% of the activity of the mock-treated sample (Fig. 3C).
One mutant in TM4, S222C, was inhibited by 78% by MTSverapamil (Fig. 3D). This mutant had an I 50 of 40 M for MTS-verapamil. By contrast, the ATPase activity of mutant S228C was inhibited by 43%, whereas the remaining mutants retained more than 75% of their activity.
Mutants within predicted TM5 showed little inhibition by MTS-verapamil (Fig. 3E). All of these mutants retained more than 75% of their activity after treatment with MTS-verapamil. The verapamil-stimulated ATPase activity of mutant II306C, however, was stimulated 2.3-fold after reaction with MTS-verapamil. This is consistent with a previous observation that I306C was stimulated when reacted with dBBn (30).
The ATPase activity of two mutants, L339C and A342C, in TM6 were quite sensitive to inhibition by MTS-verapamil (Fig.  3F). Mutants L339C and A342C retained 13 and 10% of the activity of the mock-treated samples and had I 50 values of 29 and 14 M, respectively. Inhibition of ATPase activity by MTSverapamil of three mutants, G341C, S344C, and G346C, could not be measured because of very low expression levels or low verapamil-stimulated ATPase activity. Full-length P-gp is not seen in mutant G341C, because it is cleaved by proteases within the first extracellular loop (43). Mutant S344C is difficult to express in large quantities because of low expression levels (28). Mutant G346C was expressed at levels comparable to the Cys-less P-gp but possessed very low verapamil-stimulated ATPase activity (10% of the Cys-less activity). The remaining mutants were relatively insensitive to inhibition by MTS-verapamil.
The mutants in TM8 were relatively insensitive to inhibition by MTS-verapamil, because none of the mutants were inhibited by more than 25% (Fig. 3H). Mutant G763C could not be tested for inhibition by MTS-verapamil, because it had very low verapamil-stimulated ATPase activity (11% of Cys-less P-gp) (30).
The ATPase activities of mutants A841C and N842C in TM9 were inhibited about 60% by MTS-verapamil and had I 50 values of 61 and 100 M, respectively. The remaining mutants retained more than 75% of their activity after treatment with MTS-verapamil (Fig. 3I).
There were several mutants in TMs 10, 11, and 12 that were very sensitive to inhibition by MTS-verapamil. In TM10, the ATPase activities of mutants I868C and A871C were inhibited by 70 -75% by MTS-verapamil (Fig. 3J). These mutants had I 50 values for MTS-verapamil of 25 and 14 M, respectively. Inhibition of the ATPase activity of mutant P866C by MTS-verapamil could not be tested because of very low verapamil-stimulated ATPase activity. We had previously reported that mutations to P866 caused large changes in substrate specificity of the transporter (25) but had very low verapamil-stimulated ATPase activity (6% of Cys-less P-gp) (30).
The activities of two mutants, F942C and T945C, in TM11 were inhibited by 76 and 87% by MTS-verapamil and had I 50 values of 20 and 8 M, respectively (Fig. 3K). Moderate inhibition by MTS-verapamil was observed in mutants M948C (52%) and F957C (57%). In a previous study, we reported that the mutant F957C (TM11) was susceptible to proteolytic degradation (29). We have found that enough mutant enzyme could be isolated for measurement of activity when the transfected cells were grown for longer periods (2 days) in the presence of cyclosporin A.
Three mutants in TM12 (V982C, G984C, and A985C) were inhibited by more than 80% by treatment with MTS-verapamil (Fig. 3L). These mutants had I 50 values of 12, 17, and 38 M for MTS-verapamil, respectively. One mutant, G989C, could not be tested for inhibition because of very low verapamil-stimulated ATPase activity. Although this mutant was expressed at levels comparable to the Cys-less parent, the verapamil-stimulated ATPase activity remained very low (10% of Cys-less P-gp) (28). Protection from Inhibition by Verapamil-Although MTSverapamil can modify cysteines within the drug-binding domain of P-gp, the highly reactive nature of the MTS group can allow it to react with any accessible cysteine in other parts of the molecule. To be able to differentiate cysteines that are within or close to the drug-binding site from those that are not, we relied on verapamil protection assays. The rationale is that because verapamil and MTS-verapamil had similar K m values for Cys-less P-gp (Fig. 2), then preincubation of mutants with verapamil should protect the mutant from inactivation by MTS-verapamil if the reactive cysteine is within or close to the binding site.

DISCUSSION
Cysteine-scanning mutagenesis and modification with sulfhydryl-specific agents was first used to probe the acetylcholine receptor channel (44). Modifications of this method have since been used to study the structure and function of many membrane proteins (reviewed in Ref. 45). This technique has been extensively used in the study of the bacterial lac permease protein (46). P-gp has been a very good eukaryotic protein for cysteine-scanning mutagenesis studies, because the Cys-less P-gp remains functional (reviewed in Ref. 47). Experiments that provide insight into the structure and mechanism of P-gp are possible because of the ability to synthesize thiol-reactive analogs of P-gp substrates such as MTS-verapamil.
Fifteen of the 242 Cys mutants tested for inhibition by MTSverapamil showed significant inhibition by MTS-verapamil. In general, the reactive mutants in TMs 4, 6, 10, 11, and 12 had lower I 50 values than those in TMs 2, 7, and 9 (Table I). This may indicate that TMs 4, 6, 10, 11, and 12 are within the drug-binding domain or that this region of P-gp is more accessible to MTS-verapamil. Verapamil significantly protected residues S222C (TM4), L339 (TM6), A342 (TM6), and G984 (TM12) from inactivation by MTS-verapamil. This indicates that these residues either line or are close to the verapamilbinding site. To picture the potential drug-binding domain of P-gp, we constructed a model in which the residues in each TM segments are arranged in ␣-helical wheels (Fig. 5). The arrangement of the TM segments are based on the results of disulfide cross-linking studies that show TM 6 is close to TMs 10, 11, and 12 and that TM12 is close to TMs 4, 5, and 6 (48,49). In another ABC transporter, cystic fibrosis transmembrane conductance regulator, it has been shown that a salt bridge may exist between TM6 and TM8 (50). TMs 4, 5, 6, 10, 11, and 12 have been postulated to line the drugbinding domain, because cross-linking between residues in these segments is affected by the presence of drug substrates. The location of the drug-binding domain within the TMs would be consistent with the idea that P-gp removes substrates that are embedded in the lipid bilayer (51,52). It is thought that substrates diffuse into the lipid bilayer and are then extracted from the bilayer leaflets by P-gp.
Studies with another thiol-reactive substrate, dBBn, also indicate that TMs 4, 5, 6, 10, 11, and 12 are close to the drug-binding domain (30). A comparison of the inhibition by dBBn and MTS-verapamil is shown in Table I. There is some overlap in the cysteine residues that can react with either dibromobimane or MTS-verapamil. For example, S222C (TM4) and L339C (TM6) are inhibited by dBBn and MTS-verapamil. Both residues were protected from inhibition by the presence of verapamil. In contrast, residues Y118C (TM2), V125C (TM2), and V982C (TM12) were inhibited by dBBn and MTS-verapamil but were less protected from inhibition by verapamil. The difference in inhibition by dBBn and MTS-verapamil may be explained if we assume that there is a single drug-binding domain in P-gp that is flexible enough to accommodate structurally diverse substrates. This "substrate-induced fit" model of substrate binding to P-gp is consistent with that proposed for the soluble BmrR transcription factor that can also bind a wide variety of compounds. The crystal structure of BmrR indicates the presence of a single drug-binding pocket (53). Indeed, disulfide-cross-linking studies have shown that regions in P-gp are quite mobile (49,54). In some instances, cross-linking between residues in the TMs or between residues in the nucleotide-binding domains occurs at 37°C and not at lower temperatures. The relative mobility of the TM segments could allow P-gp to bind compounds of diverse structures. A substrate could induce a conformational change in the TM segments such that different residues in the TMs would contribute to its binding. In this scenario, it is possible that different substrates could have overlapping residues contributing to its  binding. This may be the case with the binding of dBBn and MTS-verapamil. Having many combinations of residues in the drug-binding domain that can contribute to binding of different substrates may explain why P-gp can bind such a wide range of structurally diverse compounds. It may also explain why P-gp has a different affinity for each substrate and may account for the reports of multiple drug-binding sites (31)(32)(33)(34)(35). Fig. 5 may also explain how mutations throughout the TM segments can directly or indirectly affect the affinity of P-gp for a particular substrate (23,25,26,(55)(56)(57)(58). Mutations to residues in the TMs may change the faces of the helices that are exposed toward the drug-binding site such that different residues are available for drug binding, thereby changing the affinity of P-gp for a particular substrate. MTS-verapamil should be a useful compound for future stud-ies on the mechanism of P-gp. Many P-gp substrates and modulators can be used to test if they will prevent modification of P-gp by MTS-verapamil, and the results will reveal if there is overlap in the binding sites. It will also be interesting to test whether membrane fluidity or lipid composition affects modification of P-gp by MTS-verapamil. P-gp activity is highly sensitive to its lipid environment (59,60). MTS-verapamil may be useful for characterizing other proteins. Verapamil and other phenylalkylamines are inhibitors of the L-type calcium channels (61-63) and Kv1.3 potassium channels (64). Blocking of the ion channels by verapamil is thought to occur by occlusion of the ion-conducting pore. Therefore, MTS-verapamil may be useful for mapping the pore region of these channel proteins. FIG. 5. Cysteine residues in the predicted TMs of P-gp that are inhibited by MTS-verapamil and protected by verapamil. Residues within the predicted TMs (1-12) (arranged as ␣-helical wheels) of P-gp are shown and viewed from the cytoplasmic side of the membrane. Circled white residues on black background, residues that when mutated to cysteine are inhibited by MTS-verapamil and highly protected by verapamil. Circled residues on stippled background, residues that when mutated to cysteine are inhibited by MTS-verapamil and are less protected by verapamil. Residues Y118C (TM2), V125C (TM2), A729C (TM7), A841C (TM9), N842C (TM9), A871C (TM10), V982C (TM12), and A985C (TM12) are also inhibited by MTS-verapamil, when they are changed to cysteine, but show little or no protection by verapamil.