Drug Binding in Human P-glycoprotein Causes Conformational Changes in Both Nucleotide-binding Domains

The human multidrug resistance P-glycoprotein (P-gp, ABCB1) uses ATP to transport many structurally diverse compounds out of the cell. It is an ABC transporter with two nucleotide-binding domains (NBDs) and two transmembrane domains (TMDs). Recently, we showed that the "LSGGQ" motif in one NBD ((531)LSGGQ(535) in NBD1; (1176)LSGGQ(1180) in NBD2) is adjacent to the "Walker A" sequence ((1070)GSSGCGKS(1077) in NBD2; (427)GNSGCGKS(434) in NBD1) in the other NBD (Loo, T. W., Bartlett, M. C., and Clarke, D. M. (2002) J. Biol. Chem. 277, 41303-41306). Drug substrates can stimulate or inhibit the ATPase activity of P-gp. Here, we report the effect of drug binding on cross-linking between the LSGGQ signature and Walker A sites (Cys(431)(NBD1)/C1176C(NBD2) and Cys(1074)(NBD2)/L531C(NBD1), respectively). Seven drug substrates (calcein-AM, demecolcine, cis(Z)-flupentixol, verapamil, cyclosporin A, Hoechst 33342, and trans(E)-flupentixol) were tested for their effect on oxidative cross-linking. Substrates that stimulated the ATPase activity of P-gp (calcein-AM, demecolcine, cis(Z)-flupentixol, and verapamil) increased the rate of cross-linking between Cys(431)(NBD1-Walker A)/C1176C(NBD2-LSGGQ) and between Cys(1074)(NBD2-Walker A)/L531C(NBD1-LSGGQ) when compared with cross-linking in the absence of drug substrate. By contrast, substrates that inhibited ATPase activity (cyclosporin A, Hoechst 33342, and trans(E)-flupentixol) decreased the rate of cross-linking. These results indicate that interaction between the LSGGQ motifs and Walker A sites must be essential for coupling drug binding to ATP hydrolysis. Drug binding in the transmembrane domains can induce long range conformational changes in the NBDs, such that compounds that stimulate or inhibit ATPase activity must decrease and increase, respectively, the distance between the Walker A and LSGGQ sequences.


Introduction
P-glycoprotein (P-gp) is an ATP-dependent drug pump that transports numerous structurally diverse compounds of different sizes out of the cell (recently reviewed in (1,2)). Therefore, P-gp can complicate cancer and AIDS chemotherapy because many therapeutic compounds are substrates of P-gp (3,4). P-gp is a single polypeptide of 1280 amino acids. It is organized as two repeating units of 610 amino acids that are joined by a linker region of about 60 amino acids (5). Each repeat has six transmembrane (TM) segments and a hydrophilic domain containing an ATP-binding site (6,7). P-gp functions as a monomer (8), but the two halves of the molecule do not have to be covalently linked for function (9,10). The transmembrane domains alone are sufficient to mediate drug-binding (10), but both ATP-binding sites must be functional for drug efflux activity (11)(12)(13)(14).
An important aspect in understanding the mechanism of P-gp is how drug transport is coupled to ATP hydrolysis. The observations that drug binding to P-gp can either stimulate or inhibit ATP hydrolysis suggests that drug-binding and ATP hydrolysis 5 In this study, we examined the effect of drug substrates on cross-linking between the LSGGQ motifs and the Walker A sites. by guest on March 24, 2020 http://www.jbc.org/

Materials and Methods
Construction of Mutants -A histidine-tagged Cys-less P-gp was constructed and then used for making mutants containing pairs of cysteines (6,19,20). Two mutants that contained a cysteine in the LSGGQ site and another in the Walker A site were constructed (18). One mutant (L531C/C1074) contained a cysteine in the NH 2  Expression, Disulfide Cross-linking Analysis and Purification -The mutant cDNAs were expressed in HEK 293 cells in the presence of cyclosporin A to promote maturation of P-gp (21,22). Membranes were prepared as described previously (19,23).
For disulfide cross-linking analysis, aliquots of membranes were added to equal volumes of TBS (10 mM Tris-HCl, pH 7.4, 150 mM NaCl) containing 1 mM Cu 2+ (phenanthroline) 3 . The samples were incubated at 21 o C or 4 o C for various intervals and the reactions stopped by addition of SDS sample buffer (125 mM Tris-HCl, pH 6.8,

7
To test the effect of nucleotide or vanadate on cross-linking, the membranes were incubated with an equal volume of TBS containing the following: 1) 12 mM ATP, 24 mM MgCl 2 and 0.6 mM sodium orthovanadate; 2) 12 mM ATP and 24 mM MgCl 2 ; 3) 12 mM ATP; 4) 24 mM MgCl 2 ; 5) 0.6 mM sodium orthovanadate; 6) 12 mM ADP, or 7) 12 mM AMP-PNP. Sodium orthovanadate was prepared from Na 3 VO 4 , pH 10 (24) and boiled for 2 min to break down polymeric species (25). The samples were incubated for 10 min at 37 o C and then cooled in an ice-bath before treatment with oxidant at 21 o C.
At this temperature, there is almost complete cross-linking in both mutants (18).
To test the effect of drug substrates on cross-linking, the mutant P-gps were preincubated with drug substrate for 10 min at 21 o C, then chilled at 4 o C for 10 min and then treated with oxidant. At 4 o C, the rate of cross-linking is also slowed and this allowed us to detect changes in cross-linking.
Purification of histidine-tagged P-gp mutants and assay of drug-stimulated ATPase activities were done as described previously (23,26) except that the isolated samples were mixed with Escherichia coli lipid rather than sheep brain phosphatidylethanolamine. Escherichia coli lipids were used because basal P-gp ATPase activity is higher in these lipids than with sheep brain phosphatidylethanolamine. This made measurement of inhibition of P-gp ATPase activity much easier. Also drug-stimulated ATPase activity of P-gp reconstituted with Escherichia coli lipids is similar to that measured in isolated mammalian plasma membranes that are enriched in P-gp (15).

Results
We previously showed by disulfide cross-linking analysis that the contact between the NBDs of Pgp could occur between the LSGGQ signature sequence in one NBD and the Walker A site in the other NBD (18). Mutants in which the leucine residue in the LSGGQ site is replaced with cysteine can be oxidatively cross-linked with the endogenous cysteine in the opposing Walker A sequence ((L531C(NBD1- To determine whether cross-linking between the NBDs could be disrupted, the mutants were pre-treated with nucleotide or subjected to vanadate trapping. P-gp traps nucleotide in the presence of vanadate plus Mg.ATP and results in a transition state (27,28). Vanadate traps ADP at either NBD by occupying the position of the γ-phosphate adjacent to ADP. Vanadate trapping at one site then inhibits ATP hydrolysis at the second ATP-binding site (27). Fig. 1 shows that inhibition of cross-linking in mutants L531C/C1074 and C431/L1176C was observed only after treatment with vanadate plus Mg.ATP. Inhibition of cross-linking was not observed when the mutants were pretreated with ATP, MgCl 2 , vanadate, Mg.ATP, ADP or with the non-hydrolyzable ATP analog AMP.PNP. It is unlikely that vanadate trapping of nucleotide denatures the protein since trapping of nucleotide is reversible (27).
We had proposed that the "LSGGQ' motifs might participate in transmitting conformational changes from the transmembrane domains to the NBDs (18). Mutations in the LSGGQ motifs do not prevent ATP binding or vanadate-trapping of nucleotide (29,30). One way of inducing different conformational changes in the TMDs is to use drug substrates with different structures. P-gp is interesting in that some drug substrates stimulate while others inhibit the ATPase activity of P-gp. Therefore, it is possible that the conformational changes in the TMDs may be monitored by changes in the crosslinking patterns in mutants L531C/C1074 and C431/L1176C.  (Fig. 2). The membranes were then treated with oxidant at 4 o C for various intervals. The rationale for doing the cross-linking at 4 o C was that thermal motion in the protein would be reduced and that subtle changes caused by drug substrate binding may be detected. Fig. 3 shows the effect of substrate on cross-linking of mutant L531C/C1074. In the absence of drug substrate, about 50% of the mutant protein was cross-linked by 16 min. In the presence of compounds (calcein-AM, demecolcine, cis(Z)-flupentixol and verapamil) that stimulate the ATPase activity of P-gp, however, the rate of cross-linking was significantly increased so that 50% cross-linking occurred by 2 min. The presence of the inhibitory compounds (cyclosporin A, Hoechst 33342 and trans(E)-flupentixol) had the opposite effect. In the presence of these inhibitors, less than 50% cross-linking was observed at 32 min (Fig. 3).
We then tested the effect of drug substrates on cross-linking of mutant C431/L1176C. In the absence of drug substrates, 50 % of the mutant protein was crosslinked with oxidant after 16 min (Fig. 4). In the presence of the stimulatory drug substrates (calcein-AM, demecolcine, cis(Z)-flupentixol and verapamil), the rate of cross-linking was increased since 50% of the mutant protein was cross-linked by 2-4 min. Drug substrates that inhibited ATPase activity (cyclosporin A, Hoechst 33342 and trans(E)-flupentixol) of the mutant P-gp also inhibited the cross-linking of the mutant protein. Fig. 4 shows that in the presence of these compounds, less than 50% crosslinking occurred at 32 min.
The effect of substrates on mutants L531C/C1074 and C431/L1176C were very similar. Compounds that stimulated the ATPase activity also stimulated the rate of cross-linking of the mutants, while those that inhibited ATPase activity also inhibited the rate of cross-linking. It is unlikely that drug substrates are binding directly to the ATPbinding sites, since it has been shown that drug binding does not alter the affinity for ATP (15,16,32).

Discussion
Disulfide cross-linking between adjacent cysteines in the Walker A sequence of one NBD and the "LSGGQ" site in the other NBD is a useful approach for monitoring changes in the NBDs. A condition that dramatically affects cross-linking between these two sites occurs after vanadate trapping of nucleotide (Fig. 1). Hydrolysis of ATP in the presence of vanadate was essential for inhibition of cross-linking because cross-linking was not inhibited in the presence of the non-hydrolyzable ATP analog, AMP.PNP.
Similarly, cross-linking was not inhibited when the mutants were pre-incubated with only, ATP, MgCl 2 , ADP, vanadate or Mg.ATP. These results indicate that the trapped vanadate either occupies the space between the cross-linkable cysteines in the LSGGQ and Walker A sites in the two NBDs or that its presence causes the two NBDs to move apart.
The drug-binding site in the TMDs (33,34) and the ATP binding sites in P-gp must be quite far apart. Fluorescence resonance energy transfer studies indicate that the ATP-binding sites are about 40 Angstrom from the drug-binding site (35). Our results show that binding of drug substrates must induce conformational changes in the drugbinding site that are transmitted distally to the NBDs. Similarly, conformational changes in the NBDs are also transmitted to the drug-binding site in the TMDs (25,28).
Therefore, there must be continuous "cross-talk" among the domains of P-gp.
The LSGGQ sequence can increase or decrease the rate of ATP hydrolysis depending on its distance from the Walker A site as shown in Fig. 5. Inhibitory substrates may cause both of the Walker A and LSGGQ sites to move farther apart and/or reduce the rate of ATP hydrolysis. In this study, the inhibitory substrates reduced the rate of cross-linking and this may occur by moving the two NBDs apart (Figs. 3 and   4). Drug substrates that stimulate ATPase activity of P-gp must bring the Walker A and LSGGQ sites closer together so that hydrolysis of ATP occurs at a faster rate.
There is no detailed crystal structure information about eukarytic ABC transporters. Recent crystal structure studies on other ABC transporters, Rad50cd (36), BtuCD (37) and MJ0796 (38), however, show that the LSGGQ sequences in these proteins are located adjacent to the γ-phosphate of ATP. It is likely that as the LSGGQ motif moves closer to the Walker A site, the rate of ATP hydrolysis increases. Our results support this idea and may explain why a substrate is a stimulator or inhibitor of P-gp. These results could form the basis for the development of better inhibitors of Pgp.