Vanadate trapping of nucleotide at the ATP-binding sites of human multidrug resistance P-glycoprotein exposes different residues to the drug-binding site

Abstract

The human multidrug resistance P-glycoprotein uses ATP to transport a wide variety of structurally unrelated cytotoxic compounds out of the cell. In this study, we used cysteine-scanning mutagenesis and cross-linking studies to identify residues that are exposed to the drug-binding site upon vanadate trapping. In the absence of nucleotides, C222(TM4) was cross-linked to C868(TM10) and C872(TM10); C306(TM5) was cross-linked to C868(TM10), C872(TM10), C945(TM11), C982(TM12), and C984(TM12); and C339(TM6) was cross-linked to C868(TM10), C872(TM10), C942(TM11), C982(TM12), and C985(TM12). These cysteines are in the middle of the predicted transmembrane (TM) segments and form the drug-binding site. Cross-linking between 332C(TM6) and cysteines introduced at the extracellular side of other TM segments was also done. In the absence of nucleotides, residues 332C and 856C on the extracellular side of TMs 6 and 10, respectively, were cross-linked with a 13-Å cross-linker (M8M, 3,6-dioxaoctane-1,8-diyl bismethanethiosulfonate). ATP plus vanadate inhibited cross-linking between 332C(TM6) and 856C(TM10) as well as those in the drug-binding site. Instead, vanadate trapping promoted cross-linking between 332C(TM6) and 976C(TM12) with a 10-Å cross-linker (M6M, 1,6-hexanediyl bismethanethiosulfonate). When ATP hydrolysis was allowed to proceed, then 332C(TM12) could form a disulfide bond with 975C(TM12). The cross-linking pattern of 332C(TM6) with residues in TM10 and TM12 indicates that the drug-binding site undergoes dynamic and relatively large conformational changes, and that different residues are exposed to the drug-binding site during the resting phase, upon vanadate trapping and at the completion of the catalytic cycle.

The human multidrug resistance P-glycoprotein (P-gp) is an ATP-dependent drug pump located at the plasma membrane that can transport a wide variety of structurally diverse compounds of different sizes (recently reviewed in ref. 1). P-gp is clinically important because overexpression of P-gp contributes to the phenomenon of multidrug resistance during treatment of AIDS and cancer. Many of these compounds in these treatment regimens are also substrates of P-gp (2–4). P-gp also plays an important role in mediating the bio-availability of oral drugs because of its relatively higher level of expression in the intestine, brain, liver, and kidney (5–10).

P-gp consists of 1,280 aa that are organized in two repeating units of 610 aa that are joined by a linker region of about 60 aa (11). Each repeat has six transmembrane (TM) segments and a hydrophilic domain containing an ATP-binding site (11–13).

An important goal in understanding the mechanism of drug transport by P-gp is to determine how ATP hydrolysis is coupled to drug efflux. Both halves of P-gp can hydrolyze ATP (14, 15), but drug-stimulated ATPase activity (14) and conferring of drug resistance (16) requires interaction between the two halves of P-gp. Both halves of P-gp are required for activity because drug binding requires interaction between the NH₂ and COOH-terminal TM domains (16) whereas both ATP binding sites are needed for ATPase activity (17–21) and the nucleotide-binding sites appear to function in an alternating mechanism (22).

It is not clear how ATP hydrolysis influences the structure of the drug-binding site of P-gp so that drug efflux can occur. An approach to studying the dynamic changes that occur during ATP-dependent drug efflux is to use cysteine-scanning mutagenesis with thiol-specific cross-linker substrates and vanadate trapping. Disulfide cross-linking studies have provided important information into the structure of membrane proteins (23–27). Recently, we identified residues in TMs 4, 5, 6, 10, 11, and 12 that contribute to the drug-binding site (28–31). Vanadate traps ADP at one of the two nucleotide-binding sites by mimicking the transition state of the γ-phosphate of ATP during ATP hydrolysis. Vanadate trapping at one site inhibits ATP hydrolysis at the second site (32).

In this study, we examined the pattern of cross-linking by thiol-specific cross-linkers in the presence of ATP and vanadate in an attempt to map the ATP-dependent conformational changes that occur within the drug-binding site during the catalytic cycle.

Materials and Methods

Construction of Mutants.

Wild-type P-gp has seven endogenous cysteines at positions 137, 431, 717, 956, 1074, 1125, and 1227. None of the cysteines are required for activity because mutation of all cysteines to alanine (Cys-less P-gp) resulted in an active molecule (12, 33). The Cys-less P-gp cDNA was also modified to code for 10 histidine residues at the COOH end of the molecule [Cys-less P-gp(His)₁₀]. The presence of the histidine tag facilitated purification of the Cys-less P-gp by nickel-chelate chromatography (34). Cysteine residues were then introduced into the Cys-less P-gp(His)₁₀ as described (30), and the integrity of the mutated cDNA was confirmed by sequencing the entire cDNA (35).

Expression and Disulfide Cross-Linking Analysis.

Expression of P-gp mutants has been described (34). Briefly, 10 10-cm diameter culture plates of human embryonic kidney (HEK) 293 cells were transfected with the mutant cDNA. After 24 h, the medium was replaced with fresh medium containing 10 μM cyclosporin A. Cyclosporin A is a substrate of P-gp and acts as a powerful chemical chaperone for rescuing misprocessed mutants of P-gp (36). After another 24 h, membranes were prepared from transfected cells and suspended in Tris-buffered saline (10 mM Tris⋅HCl, pH 7.4/150 mM NaCl) containing 10 mM MgCl₂. The membranes were then preincubated for 15 min at room temperature (22°C) in the presence of no nucleotides, 5 mM ATP, 5 mM ADP, 5 mM AMP⋅PNP, 5 mM ATP plus 0.2 mM sodium orthovanadate, or 5 mM ADP plus 0.2 mM sodium orthovanadate. Sodium orthovanadate was boiled for 2 min before use to break down polymeric species (37). The membranes were then incubated with 0.2 mM methanethiosulfonate (MTS) cross-linker (Toronto Research Chemicals, Downsview, ON, Canada) for various times at 4°C or 22°C. The following MTS cross-linkers were used: M2M, 1,2-ethanediyl bismethanethiosulfonate (5.2 Å); M3M, 1,3-propanediyl bismethanethiosulfonate (6.5 A); M4M, 1,4-butanediyl bismethanethiosulfonate (7.8 Å); M5M, 1,5-pentanediyl bismethanethiosulfonate (9.1 Å); M6M, 1,6-hexanediyl bismethanethiosulfonate (10.4 Å); M8M, 3,6-dioxaoctane-1,8-diyl bismethanethiosulfonate (13 Å); M11M, 3,6,9-trioxaundecane-1,11-diyl bismethanethiosulfonate (16.9 Å); M14M, 3,6,9,12-tetraoxatetradecane-1,14-diyl bismethanethiosulfonate (20.8 Å); and M17M, 3,6,9,12,15-pentaoxaheptadecane-1,17-diyl bismethanethiosulfonate (24.7 Å). The sizes of the cross-linker spacer arms (Å) were calculated previously (38). The MTS compounds were prepared as 100 mM stock solutions in DMSO. The reactions were stopped by addition of an equal volume of SDS sample buffer (0.125 mM Tris⋅HCl, pH 6.8), 4% (wt/vol) SDS, 10% (vol/vol) glycerol, and 10 mM N-ethylmaleimide. The samples were subjected to SDS/PAGE (39) on 7.5% acrylamide gels and immunoblot analysis with rabbit polyclonal antibody (40).

Results

P-gp can interact with drug substrates and be cross-linked in the absence of any nucleotide (16). For drug efflux to occur, however, ATP hydrolysis is required (41, 42). Therefore, the main function of ATP hydrolysis during drug transport must be to alter the structure of the drug-binding site such that the affinity for the bound substrate is changed. This would then lead to release of the bound substrate.

An approach to trapping P-gp in its transition state after ATP hydrolysis is to use vanadate inhibition of P-gp. Vanadate inhibits P-gp activity by trapping ADP at one of the two nucleotide-binding sites immediately after ATP hydrolysis because it mimics the structure of the terminal phosphate of ATP (32).

To test whether vanadate trapping of nucleotide affected the structure of the drug-binding site, we searched for changes in the cross-linking pattern of P-gp with MTS cross-linkers during the ATP cycle. Thiol-reactive MTS cross-linkers are useful compounds for monitoring dynamic changes within the drug-binding site of P-gp because they are substrates of P-gp and can stimulate the ATPase activity of P-gp up to 10-fold. This property and their ability to cross-link cysteine residues allowed them to be used to determine the dimensions of the drug-binding site (38). We previously reported that residues in TMs 4–6 and TMs 10–12 (Fig. 1) form the drug-binding site of P-gp (30, 31) and that cysteines in the middle of these TM segments can be linked by cross-linkers with different spacer arms. C222(TM4) was cross-linked to C868(TM10) and C872(TM10); C306(TM5) was cross-linked to C868(TM10), C872(TM10), C945(TM11), C982(TM12) and C984(TM12); and C339(TM6) was cross-linked to C868(TM10), C872(TM10), C942(TM11), C982(TM12), and C985(TM12) (31). Accordingly, membranes prepared from HEK 293 cells expressing double cysteine P-gp mutants were first incubated for 15 min at room temperature with or without 5 mM ATP and 0.2 mM vanadate. Under these conditions, the ATPase activity of P-gp is inhibited by more than 90% (43) because of trapping of ADP at one of the two nucleotide-binding sites. The membranes were then incubated with MTS cross-linkers of various sizes. We reasoned that because the MTS cross-linkers could cross-link cysteine residues in the drug-binding site in the absence of nucleotides, then ATP hydrolysis and vanadate trapping should inhibit cross-linking of some residues. Similarly, new cross-links should be generated if there is a conformational change in the drug-binding site. Table 1 shows that in all cases vanadate trapping caused a reduction in the amount of cross-linked product. An example of the results obtained with mutant L339C(TM6)/A985C(TM12) is shown in Fig. 2. In the absence of ATP and vanadate, there was extensive cross-linking with M14M (3,6,9,12-tetraoxatetradecane-1,14-diyl bismethanethiosulfonate) and M17M (3,6,9,12,15-pentaoxaheptadecane-1,17-diyl bismethanethiosulfonate), as the major product migrated with lower mobility than mature (170 kDa) P-gp in SDS gels (Fig. 2A). In previous studies, it was shown that cross-linking between residues in the NH₂ and COOH halves of P-gp resulted in a product that migrated with lower mobility in SDS gels (44, 45). After treatment with ATP plus vanadate, however, cross-linking with both compounds was inhibited (Fig. 2B). These results show that the structure of the drug-binding site undergoes a large conformational change during hydrolysis of the first ATP.

Figure 1.

Schematic diagram of P-gp. The full-length P-gp is shown where the 12 TMs are represented as numbered rectangles, the zigzag line represents the linker region, the branched lines represent glycosylation sites, and ATP represents the location of the nucleotide-binding sites. Residues that were identified as being within the drug-binding site and that could be cross-linked with MTS cross-linkers are shown (○). The locations of cysteine residues that show ATP-dependent cross-linking are shown (●).

Table 1.

Effect of nucleotides on cross-linking by MTS compounds

Mutant	No nucleotide	ATP/VO4	AMP⋅PNP
S222C(TM4)/I868C(TM10)	5, 6, 8, 17*	↓†	—‡
S222C(TM4)/G872C(TM10)	5, 6, 17	↓	—
I306C(TM5)/I868C(TM10)	8, 17	↓	—
I306C(TM5)/G872C(TM10)	8, 17	↓	—
I306C(TM5)/T945C(TM11)	8, 11, 17	↓	—
I306C(TM5)/V982C(TM12)	8, 11, 14, 17	↓	—
I306C(TM5)/G984C(TM12)	8, 11, 14, 17	↓	—
L339C(TM6)/I868C(TM10)	8, 11, 14, 17	↓	—
L339C(TM6)/G872C(TM10)	8, 11, 14, 17	↓	—
L339C(TM6)/F942C(TM11)	17	↓	—
L339C(TM6)/T945C(TM11)	14, 17	↓	—
L339C(TM6)/V982C(TM12)	11, 14, 17	↓	—
L339C(TM6)/A985C(TM12)	14, 17	↓	—

^*Cross-linked product detected in SDS/PAGE by using MTS cross-linkers. The number represents the length of the spacer arm of the MTS cross-linker (38). 

^† Cross-linking by the MTS cross-linkers was inhibited. 

^‡ No effect on the cross-linking pattern. 

Figure 2.

Effect of ATP plus vanadate or AMP⋅PNP on cross-linking of mutant L339C/A985C. Membranes prepared from HEK 293 cells expressing mutant L339C/A985C were preincubated with no nucleotide (A), ATP plus vanadate (B), or AMP⋅PNP (C) for 15 min at room temperature. The membranes were then incubated with various MTS (0.2 mM) cross-linkers for 30 min at 4°C. The reactions were then stopped by addition of an equal volume of SDS sample buffer containing 10 mM N-ethylmaleimide. The samples were subjected to SDS/PAGE followed by immunoblot analysis. The numbers represent the type of MTS compound used. The positions of the cross-linked (X-link) and mature (170 kDa) P-gp are indicated.

It has been reported that binding of nonhydrolyzable ATP analogs can alter drug binding (46). Therefore, we tested whether the nonhydrolyzable ATP analog, AMP⋅PNP, affected the MTS cross-linking pattern of L339C/A985C. Fig. 2C shows that the presence of AMP⋅PNP did not significantly affect the cross-linking pattern of this mutant. Similarly, the cross-linking pattern of the other mutants was not affected by the presence of AMP⋅PNP (Table 1). It appears that reduction in cross-linking with ATP and vanadate is caused by the TM segments having undergone a significant rearrangement as a result of ATP hydrolysis. This does not occur during nucleotide binding. If the TM segments are indeed undergoing a rearrangement, then it should be possible to detect enhanced cross-linking at new sites.

A potentially useful site that could be used to detect enhanced cross-linking after vanadate trapping is at residue L332C(TM6). This TM segment is directly connected to the first nucleotide-binding site (Fig. 1), and any conformational change in the first nucleotide-binding domain would directly affect this TM segment. In previous studies, it was shown that ATP hydrolysis induces this residue to come close to L975C(TM12) such that cross-linking occurs in the presence of a zero-length cross-linker (copper phenanthroline) (47). In this instance, cross-linking between L332C(TM6) and L975C(TM12) occurred late in the reaction cycle because no cross-linked product was detected in the presence of nonhydrolyzable ATP analogues or when in the presence of ATP plus vanadate. Therefore, it may be possible to follow the dynamic movement of residue L332C(TM6) if we could differentiate between the cross-linking patterns in the absence of nucleotides and during the transition state immediately after ATP hydrolysis at one nucleotide-binding site. Accordingly, we constructed mutant P-gps containing L332C(TM6) and another cysteine residue close to the extracellular side of TMs 7–12 (Table 2). Membranes prepared from transfected HEK 293 cells expressing these mutant P-gps were treated for 15 min at room temperature with or without ATP plus vanadate, followed by incubation with various MTS cross-linkers or copper phenanthroline (zero-length cross-linker) at 4°C. The reactions were done at 4°C to reduce molecular motion in the protein. Copper phenanthroline did not cross-link any of the mutants (Table 2). In the absence of ATP, only mutant L332C(TM6)/Q856C(TM10) was cross-linked by M6M, M8M, and M11M (3,6,9-trioxaundecane-1,11-diyl bismethanethiosulfonate) (Table 2; Fig. 3A). In the transition vanadate-trapped state, however, cross-linking of mutant L332C(TM6)/Q856C(TM10) was decreased (Fig. 3B). By contrast, cross-linking by M5M (1,5-pentanediyl bismethanethiosulfonate), M6M, and M8M could be detected in the mutant L332C(TM6)/L976C(TM12) (Fig. 3D). No cross-linked product was detected in this mutant in the absence of nucleotide (Fig. 3C). These results indicate the presence of ATP plus vanadate promoted a reorientation of residue L332C(TM6) away from Q856C(TM10) toward L976C(TM12). Cross-linking was specific for mutant L332C(TM6)/L976C(TM12) because none of the other mutants were cross-linked in the presence of ATP plus vanadate. P-gp mutants with one cysteine residue were not cross-linked under any of the conditions shown in Fig. 3 (data not shown).

Table 2.

Effect of nucleotides on cross-linking

Residue 332C(TM6) and	Copper phenanthroline*	No nucleotide (+ MTS cross-linker)	ATP/VO4 (+ MTS cross-linker)
A727C(TM7)	—†	—	—
F728C	—	—	—
A729C	—	—	—
I730C	—	—	—
I731C	—	—	—
F732C	—	—	—
F759C(TM8)	—	—	—
L760C	—	—	—
A761C	—	—	—
L762C	—	—	—
G763C	—	—	—
I764C	—	—	—
I847C(TM9)	—	—	—
I848C	—	—	—
I849C	—	—	—
S850C	—	—	—
F851C	—	—	—
I852C	—	—	—
Q856C(TM10)	—	6, 8, 11‡	—
L857C	—	—	—
T858C	—	—	—
L859C	—	—	—
L860C	—	—	—
L861C	—	—	—
S952C(TM11)	—	—	—
Y953C	—	—	—
A954C	—	—	—
G955C	—	—	—
C956C	—	—	—
F957C	—	—	—
L975C(TM12)	—	—	—
L976C	—	—	5, 6, 8‡
V977C	—	—	—
F978C	—	—	—
S979C	—	—	—
A980C	—	—	—

^*Cross-linking was done at 4°C in the absence of nucleotides. 

^† No cross-linked product detected on SDS/PAGE. 

^‡ Cross-linked product was detected with these MTS cross-linkers. The numbers refer to the length of the spacer arm in the MTS cross-linker (38). 

Figure 3.

Effect of ATP plus vanadate on cross-linking of mutants L332C/Q856C and L332C/L976C. Membranes prepared from HEK 293 cells expressing the mutant L332C/Q856C (A and B) or mutant L332C/L976C (C and D) were preincubated with no nucleotide (A and C) or with ATP plus vanadate (B and D) and then cross-linked with various MTS (0.2 mM) cross-linkers. The reaction was stopped by addition of an equal volume of SDS sample buffer containing 10 mM N-ethlymaleimide, and the sample was subjected to SDS/PAGE and immunoblot analysis. The numbers represent the type of MTS compound used. The positions of the cross-linked (X-link) and mature (170 kDa) P-gp are indicated.

To determine the conditions that promoted cross-linking of mutant L332C(TM6)/L976C(TM12), cross-linking with M6M was carried out in the presence of ATP, ATP plus vanadate, AMP⋅PNP, ADP, ADP plus vanadate, or vanadate alone. Cross-linking was then carried out with M6M cross-linker because it was the best cross-linker for mutant L332C(TM6)/L976C(TM12) in the presence of ATP plus vanadate (see Fig. 3D). Fig. 4 shows that cross-linking of mutant L332C(TM6)/L976C(TM12) occurred only in the presence of ATP plus vanadate. These results indicate that cross-linking was caused by trapping of the transition state during the reaction cycle rather than binding of ATP, ADP, or vanadate. A potential problem in the cross-linking experiments is that the membrane preparations may already contain bound nucleotides. To test for the presence of bound nucleotide, mutant L332C(TM6)/L976C(TM12) membranes were incubated with vanadate at 37°C because at this temperature a transition-state intermediate can be generated with ATP or ADP plus vanadate (48). The mutant was then chilled on ice and treated with M6M cross-linker, and the samples were subjected to SDS/PAGE and immunoblot analysis. No cross-linked product was observed (data not shown), indicating that the mutant did not contain bound nucleotides. Also, little or no transition-state intermediate is generated with ADP plus vanadate at 4°C (48), which may explain the absence of cross-linking in Fig. 4F.

Figure 4.

Effect of nucleotides and vanadate on cross-linking of mutant L332C/976C. Membranes prepared from HEK 293 cells expressing mutant L332C/L976C were incubated for 15 min at room temperature in the presence of no nucleotide (A), ATP (B), ATP plus vanadate (C), AMP⋅PNP (D), ADP (E), ADP plus vanadate (F), or vanadate (G). The samples were then placed on ice and M6M cross-linker was added to a final concentration of 0.2 mM. At the indicated times, samples were removed and the reaction was stopped by addition of an equal volume of SDS sample buffer containing 10 mM N-ethlymaleimide, and the sample was subjected to SDS/PAGE and immunoblot analysis. The numbers represent the type of MTS compound used. The positions of the cross-linked (X-link) and mature (170 kDa) P-gp are indicated.

Discussion

Our results show that the TM domain can exist in at least three different conformational states during the catalytic cycle. In the resting state (absence of ATP), we had shown that the cysteine residues located in TMs 4–6 and TMs 10–12 form the drug-binding site. These residues, 222(TM4), 306(TM5), 339(TM6), 868(TM10), 872(TM10), 942C(TM11), 945(TM11), 982(TM12), 984(TM12), and 985(TM12) must line a common drug-binding site and were cross-linked by thiol-specific cross-linkers of various lengths (38).

The next step in the catalytic cycle is the hydrolysis of the first ATP molecule. Vanadate trapping allowed us to characterize the first hydrolysis step because P-gp is stably inhibited upon trapping of ADP at a single catalytic site (32). Although both nucleotide-binding sites are catalytically active, locking of one site by vanadate blocks ATP hydrolysis at the second site (15). In this transition state, the pattern of cross-linking of P-gp by MTS as seen in the resting state, was altered. This inhibition of cross-linking required ATP hydrolysis and not just nucleotide binding because no detectable change in the cross-linking pattern was observed in the presence of the nonhydrolyzable ATP analog, AMP⋅PNP.

Although vanadate trapping of nucleotide inhibited the cross-linking pattern observed in the resting state, it also promoted cross-linking of different residues. Residues 332C(TM6) and 976C(TM12) are predicted to be close to the extracellular side of the membrane. Cross-linking of these residues by M6M depended on the formation of the vanadate-trapped transition state because little or no detectable cross-linking was detected at 4°C in the presence of ATP, AMP⋅PNP, ADP, ADP plus vanadate, or vanadate alone.

Based on the results from a previous study, it appears that the drug-binding site can exist in another conformational state after ATP hydrolysis at the first site (47). In this study, oxidative cross-linking with copper phenanthroline (zero-length) between 332C(TM6) and 975C(TM12) occurred only during ATP hydrolysis. No cross-linking was observed in the presence of ATP plus vanadate or in the presence or absence of the nonhydrolyzable analog of ATP, AMP⋅PNP. Based on the results in the present paper, an explanation for the observed cross-linking of 332C(TM6)/975C(TM12) (47) is that P-gp must undergo further conformation changes when ATP hydrolysis occurs at the second nucleotide binding site (i.e., complete ATP hydrolysis), resulting in both residues being very close together.

The results in this article could best be explained in a model that incorporates results from this and previous studies (43, 47). Recently, it was shown that the drug-binding site is “funnel-shaped,” narrow at the cytoplasmic side and wide at the extracellular surface (43). Fig. 5A shows that binding of drug substrates to P-gp does not require the presence of nucleotides or even the presence of the two nucleotide-binding sites because a P-gp deletion mutant lacking both nucleotide-binding sites still bound drug substrates (16). The drug substrates likely enter P-gp from the lipid bilayer (49, 50). The second step in the transport cycle is ATP hydrolysis at one nucleotide-binding site. Although both nucleotide-binding sites are catalytically active, hydrolysis of the first ATP molecule occurs nonselectively at the NH₂- or COOH-terminal nucleotide binding site (15). Hydrolysis of the first ATP caused a significant and detectable change in the structure of the drug-binding site. Cross-linking between residues in the NH₂ and COOH terminal in the resting phase is not detected (Table 2). This alteration in structure of the drug-binding site affects the affinity of P-gp for the MTS cross-linkers (Fig. 5B). This finding would be consistent with the report that ATP hydrolysis at one nucleotide-binding site lowered the affinity of P-gp for [¹²⁵I]-iodoarylazidoprazosin by more than 30-fold (51). Another explanation for inhibition of cross-linking within the drug-binding site in the vanadate-trapped state is that movement of TM segments alters the accessibility of the cysteines to MTS cross-linkers. Our results are also consistent with those of Sauna and Ambudkar (51) in that binding of nucleotides was not sufficient to affect interaction of P-gp with drug substrates.

Figure 5.

Model for the drug transport. (A) Drug substrate (⧫) in the extracellular fluid diffuses first into the lipid bilayer and then into the drug-binding site of P-gp. (B) Hydrolysis of the first ATP causes conformational changes in the TM domain, such as rotation (43) or lateral movement (47) of one or more of the TM segments, resulting in exposure of different residues to the drug-binding site. (C) Hydrolysis of a second ATP molecule results in another conformational change in the TM domain that causes 332C and 975C to come very close together and occlusion of the drug-binding site and simultaneous release of substrate. The cross-linking properties of L332C(TM6) during the transport cycle are indicated. In A, 332C(TM6) can be cross-linked to C856(TM10) with M8M; in B, 332C(TM6) can be cross-linked to C976C(TM12) by using M6M; and in C, 332C(TM6) comes close enough to 975C(TM12) to form a disulfide bond in the presence of an oxidant.

Recently, Rosenberg et al. (52) reported that P-gp binding to AMP⋅PNP altered the structure of the TM domains. In contrast, we could not detect any effect of AMP⋅PNP on cross-linking (see Table 1 and Figs. 2 and 4). A possible explanation is that binding of nucleotide causes conformation changes in other TMs that were not tested for cross-linking. Further cross-linking studies will be required to address this issue. In addition, the use of hydrophilic cross-linkers may detect the presence of other conformational states as would be predicted from the funnel-shaped nature of the drug-binding site (38), where a more aqueous environment would exist at the wider end of the funnel.

A third step would be hydrolysis at the second nucleotide-binding site to complete the catalytic cycle. The requirement for two ATP hydrolysis steps during each transport cycle would be consistent with the measured stoichiometry for P-gp where two ATP molecules are required per reaction cycle (53, 54). At this step, residues 332C and 975C must be very close together (47), resulting in occlusion of the drug-binding site and simultaneous release of substrate from the binding site (Fig. 5C) and resetting of P-gp back to its original conformation (51).

Abbreviations

MTS

methanethiosulfonate

P-gp

P-glycoprotein

TM

transmembrane

M6M

1,6-hexanediyl bismethanethiosulfonate

M8M

3,6-dioxaoctane-1,8-diyl bismethanethiosulfonate

HEK

human embryonic kidney