Glutathione stimulates sulfated estrogen transport by multidrug resistance protein 1.

Multidrug resistance protein 1 (MRP1) is an ATP-binding cassette (ABC) transporter that transports a range of hydrophobic xenobiotics, as well as relatively hydrophilic organic anion conjugates. The protein is present at high levels in testicular Leydig and Sertoli cells. Studies with knockout mice suggest that MRP1 may protect germ cells from exposure to some cytotoxic xenobiotics, but potential endobiotic substrates in this organ have not been identified. Previously, we have shown certain D-ring, but not A-ring, estrogen glucuronides can act as competitive inhibitors of MRP1 mediated transport, suggesting that they are potential substrates for the protein. In the case of 17 beta-estradiol-17 beta-d-glucuronide, this has been confirmed by direct transport studies. The Leydig cell is the major site of estrogen conjugation in the testis. However, the principal products of conjugation are A-ring estrogen sulfates, which are then effluxed from the cell by an unknown transporter. To determine whether MRP1/mrp1 could fulfill this function, we used membrane vesicles from MRP1-transfected HeLa cells to assess this possibility. We found that estradiol and estrone 3-sulfate alone were poor competitors of MRP1-mediated transport of the cysteinyl leukotriene, leukotriene C(4). However, in the presence of reduced glutathione (GSH), their inhibitory potency was markedly increased. Direct transport studies using [(3)H]estrone 3-sulfate confirmed that the conjugated estrogen could be efficiently transported (K(m) = 0.73 microm, V(max) = 440 pmol mg(-)1 protein min(-)1), but only in the presence of either GSH or the nonreducing alkyl derivative, S-methyl GSH. In contrast to previous studies using vincristine as a substrate, we detected no reciprocal increase in MRP1-mediated GSH transport. These results provide the first example of GSH-stimulated, MRP1-mediated transport of a potential endogenous substrate and expand the range of MRP1 substrates whose transport is stimulated by GSH to include certain hydrophilic conjugated endobiotics, in addition to previously identified hydrophobic xenobiotics.

Human multidrug resistance protein (MRP) 1 1 is a member of the ATP-binding cassette superfamily of transmembrane transporters, which was originally discovered by virtue of its association with drug resistance in tumor cells (1). It is now also known to be a primary active transporter of many conjugated organic anions (2). The first substrate shown to be actively transported by MRP1 using inside-out membrane vesicles was the glutathione-conjugated leukotriene, LTC 4 (3,4). Since then, the spectrum of molecules transported by MRP1 has been extended to include many other GSH conjugates, as well as several glucuronate and sulfate conjugates (5,6). A number of unconjugated amphiphilic anions have also been demonstrated to be substrates for MRP1 (7). However, using an in vitro membrane vesicle system unmodified forms of the natural product drugs to which MRP1 confers resistance are not directly transported by this protein, although primary active transport of some of them has been observed in the presence of reduced glutathione (GSH) (8 -11). Potential endogenous substrates of MRP1 that show a similar dependence on the presence of GSH have not been identified, although it has been observed that the intracellular levels of GSH are reduced in some drug selected and transfected cells overexpressing MRP1 in the absence of an exogenous substrate (12)(13)(14). In addition, GSH levels are elevated in some tissues in mrp1 Ϫ/Ϫ mice (15). Whether this is attributable to the efflux of GSH in association with transport of endogenous compounds has not been established. MRP1/mrp1 is highly expressed in the sex hormone-producing Leydig cells of the human and mouse testis (16,17), as well as in Sertoli cells of the mouse (17). Both of these cell types have high levels of GSH S-transferase activity (18,19) and are thought to contribute to detoxification in the testis by the formation of GS conjugates, which may be substrates for MRP1 or related transporters. Studies of mrp1 Ϫ/Ϫ mice have already provided evidence that testicular mrp1 protects the local tissue against drug-induced damage (17). Thus, the testes of mrp1deficient mice treated with the anticancer drug, etoposide phosphate, showed aberrant spermatogenesis with no sign of meiotic divisions and an increased number of prematurely released round germ cells. In contrast, the same treatment of the wildtype mice produced only a partially distorted spermatogenesis and meiotic divisions still occurred.
In addition to a protective role with respect to xenobiotic exposure, the relatively high levels of MRP1 in Leydig cells may also serve to protect the testis from the potential feminiz-* This work was supported by grants from Eli Lilly and Co. and the National Cancer Institute of Canada with funds from the Terry Fox Run. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ing effects of endogenously produced estrogen conjugates. Estrogen is synthesized in the testis and is required for normal testicular function, as revealed by studies of mice in which the estrogen receptor ␣ or P450 aromatase genes have been disrupted (20 -22). MRP1 has been shown to transport certain estrogen glucuronides but the major metabolite in Leydig cells is estrogen sulfate produced by estrogen sulfotransferase (EST) (5,23). Using 3Ј-phosphoadenosine 5Ј-phosphosulfate as a sulfate donor, this sulfation takes place at the 3-hydroxyl group of the parent molecule, generating the more hydrophilic estrogen 3-sulfate. The relative hydrophilicity of estrogen 3-sulfate prevents its ready diffusion across the plasma membrane. Thus, it has been assumed that an export pump is involved in its efflux from the cell. Testicular expression of estrogen sulfotransferase is mainly localized in Leydig cells and is regulated by luteinizing hormone via modulation of cAMP levels (23,24). Interestingly, the testes of 9 -12-month-old est Ϫ/Ϫ mice displayed Leydig cell hypertrophy/hyperplasia, as well as seminiferous tubule disruption. 2 The colocalization of MRP1 and estrogen sulfotransferase in the testis prompted us to investigate whether estrogen sulfates were also substrates of MRP1. We found that estrogen sulfates alone were very poor substrates and competed poorly for transport and binding of the high affinity MRP1 substrate LTC 4  ) and fluorographic reagent Amplify ® were from Amersham Pharmacia Biotech (Oakville, Ontario, Canada). Estrogen 3-sulfates, nucleotides, GSH, verapamil, S-methyl GSH, glutathione disulfide (GSSG), 2-mercaptoethanol, and DTT were purchased from Sigma. Estrogen 3-sulfates were dissolved with H 2 O to prepare stock solutions at 10 mM and diluted with transport buffer. The MRP1-specific murine monoclonal antibodies (mAbs) QCRL-1, QCRL-2, QCRL-3, and QCRL-4 have been described previously (25,26).
Cell Culture-The HeLa T5 cells transfected with the pRcCMV vector containing the MRP1 coding sequence and the HeLa C1 cells transfected with empty pRcCMV vector have been described previously (27). Both T5 and C1 were cultured in RPMI 1640 medium with 5% defined bovine calf serum and maintained in 400 g ml Ϫ1 Geneticin (G418). HEK293 cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum.
Membrane Vesicle Preparation and Transport Studies-Plasma membrane vesicles were prepared as described (8). Briefly, cell pellets were covered with buffer containing 50 mM Tris-HCl, 250 mM sucrose, 0.25 mM CaCl 2 , and protease inhibitors and frozen at Ϫ70°C overnight. The cells were then thawed and disrupted by N 2 cavitation (10-min equilibrium at 200 p.s.i.). EDTA was added to 1 mM before centrifugation at 500 ϫ g for 15 min to remove cell debris. To increase the yield of membrane vesicles, the resulting pellet was washed once with 5 ml of transport buffer (50 mM Tris-HCl and 250 mM sucrose, pH 7.5) and centrifuged again. The supernatants were pooled, layered over 35% (w/w) sucrose in 10 mM Tris-HCl (pH 7.5), and centrifuged at 100,000 ϫ g for 1 h. The interface was collected in washing buffer (10 mM Tris-HCl and 25 mM sucrose) and followed by centrifugation at 100,000 ϫ g for 30 min. The membrane pellet resuspended in transport buffer and passed 10 times through a 27.5-gauge needle for vesicle formation. The membrane vesicles were then aliquoted and stored at Ϫ70°C until use.
ATP-dependent transport of [ 3 H]LTC 4 into the inside-out membrane vesicles was measured by a rapid filtration technique (8). In standard transport assays, 2.5 g of membrane vesicles were used in a 25-l reaction volume and incubated at 23°C in the presence of 50 nM [ 3 H]LTC 4 , 10 mM MgCl 2 , and 4 mM ATP or AMP in transport buffer. Where indicated, GSH was added to 1 mM unless otherwise stated. Uptake was stopped by rapid dilution in ice-cold transport buffer and followed by filtration through glass fiber (type A/E) filters (Gelman Sciences, Dorval, Quebec, Canada) that had been presoaked overnight at 4°C in transport buffer. All of the data had been corrected for the amount of [ 3 H]LTC 4 that remained bound to the filter in the absence of the membrane vesicles, which was usually less than 5-10% of the total radioactivity. For kinetic analysis of LTC 4 transport in the presence of estrogen sulfates and/or GSH, LTC 4 was included at concentrations ranging from 16 nM to 1 M and ATP-dependent [ 3 H]LTC 4 uptake was determined as above.
ATP-dependent uptake of [ 3 H]estrone 3-sulfate was measured by rapid filtration as above, except that the incubation temperature was 37°C and substrate concentration was 300 nM unless otherwise indicated. Uptake was stopped after 60 s or at the time indicated by rapid dilution in ice-cold buffer, and the reaction mixture was filtered through glass fiber filters. Where indicated, MRP1-specific mAbs were added to 10 g ml Ϫ1 and preincubated with membrane vesicles on ice for 1 h. All data were corrected by subtracting nonspecific binding of [ 3 H]estrone 3-sulfate to the filter, which was usually less than 5% of the total radioactivity. For kinetic analysis of GSH-enhanced estrone 3-sulfate transport in the absence or presence of LTC 4 , estrone 3-sulfate was included at concentrations ranging from 125 nM to 16 M and ATP-dependent [ 3 H]estrone 3-sulfate uptake was determined as above. Transport of [ 3 H] GSH into the membrane vesicles was also measured by rapid filtration as above. In a 50-l reaction volume, 22 g of membrane vesicle protein were incubated at 37°C for 20 min in the presence of 100 M [ 3 H]GSH (80 or 288 nCi/reaction), 10 mM DTT, 10 mM MgCl 2 , and 4 mM ATP or AMP in transport buffer. Estrone 3-sulfate or estradiol 3-sulfate was added to several concentrations ranging from 0.2 to 20 M. Verapamil was used a positive control for stimulation of [ 3 H]GSH transport (28) and was added to 100 M. All data were corrected by subtracting the amount of [ 3 H]GSH that remained bound to the filter in the presence of 4 mM AMP, which was usually less than 5% of the total radioactivity. Samples were alternately irradiated for 30 s at 312 nm in a Stratalinker, followed by snap-freezing in liquid nitrogen, for a total of 10 min. Radiolabeled vesicles were solubilized in Laemmli's buffer and analyzed on a 7.5% gel by SDS-polyacrylamide gel electrophoresis. The gel was fixed in isopropanol:water:acetic acid (25:65:10) for 30 min and then soaked in Amplify for 15-20 min. After drying under vacuum at 80°C, the gel was placed in close contact with x-ray film at Ϫ70°C for 2 weeks (8).

Photoaffinity Labeling of MRP1 with [ 3 H]LTC 4 and Inhibition of Labeling by Estrogen
[ 3 H]Estradiol Sulfate Accumulation in Intact HEK Cells-The expression vectors pCDNA3 containing mouse estrogen-sulfotransferase cDNA (pCDNA3-est) and pCEBV7 containing human MRP1 cDNA (pCEBV7-MRP1) were described previously (29,30), and transfection was performed with FuGENE 6 transfection reagent (Roche Molecular Biochemicals). HEK293 cells (2 ϫ 10 6 ) were seeded into each T75 flask and transfected on the following day with a mixture of 40 l of FuGENE reagent and 10 g of pCDNA3-est or together with 10 g of pCEBV7-MRP1. After 66 h, the cells were harvested. Half of the cells were then used for steroid accumulation experiments and the other half for an EST activity assay. [ 3 H]Estradiol sulfate accumulation was initiated by addition of [ 3 H]estradiol to a final concentration of 500 nM to a suspension of untransfected or transfected cells (5 ϫ 10 5 cells in 100 l of serum-free DMEM) at 37°C and stopped at 20 min by adding 1 ml of ice-cold phosphate-buffered saline. The cells were washed twice with ice-cold phosphate-buffered saline., solubilized in 200 l of 0.5% SDS, and extracted with 400 l of ethyl acetate. The aqueous phase was taken for liquid scintillation counting, and 3 H-labeled aqueous soluble steroid accumulation was expressed as dpm/10 6 cells. The procedures for preparation of cytosolic proteins and EST activity assay were the same as described previously (29,31).

Inhibition of MRP1-mediated LTC 4 Transport by Estrogen
Sulfates-To determine whether estrogen sulfates were potential substrates for MRP1, we initially tested the ability of estrone and estradiol 3-sulfate to compete for transport of the high affinity MRP1 substrate, LTC 4 . Neither estrogen sulfate alone was a potent inhibitor of [ 3 H]LTC 4 uptake by vesicles prepared from MRP1-transfected HeLa T5 (HeLa-MRP) cells ( Fig. 1, A and B). At lower concentrations of the conjugated estrogens (20 -200 nM), a reproducible 25-50% increase in the rate of LTC 4 transport was observed with significant inhibition of transport being observed when concentrations of the estrogen sulfates reached more than 20 M. Previously, GSH has been shown to increase the inhibitory potency of some hydrophobic xenobiotics, which are themselves poor competitive inhibitors of LTC 4 transport (8). However, no effect on the transport of conjugated endobiotics has been reported. We found that addition of 1 mM GSH markedly increased the inhibitory potency of the sulfated estrogens, resulting in a decrease in IC 50 values from 31 and 50 M to 0.2 and 0.3 M, for estrone sulfate and estradiol sulfate, respectively (Fig. 1, A and B).
To determine whether inhibition of LTC 4 transport by the estrogen sulfates was competitive, Eadie-Hofstee plots of LTC 4 transport in the absence or presence of 2 M estrogen sulfate were constructed. At this concentration, estrone sulfate alone inhibited [ 3 H]LTC 4 transport by ϳ10% while estradiol sulfate stimulated transport by ϳ25% (Fig. 1, A and B). However, in the presence of 1 mM GSH, both estrone and estradiol sulfate  estrone 3-sulfate was very low (5-6 pmol mg Ϫ1 protein min Ϫ1 ). However, an approximate 10-fold increase in the rate of transport was observed when 1 mM GSH was present (Fig. 3A). GSH-stimulated transport was linear for up to 60 s at a rate of ϳ60 pmol mg Ϫ1 protein min Ϫ1 with an initial concentration of 300 nM estrone 3-sulfate. In the presence of 4 mM AMP and 1 mM GSH, estrone 3-sulfate uptake by the vesicles was 2-4 pmol mg Ϫ1 protein min Ϫ1 . ATP-dependent and GSH-stimulated estrone 3-sulfate transport was not detected with membrane vesicles prepared from vector-transfected HeLa C1 control cells (Fig. 3B).
To confirm that vesicle-associated estrone sulfate was indicative of transport into the vesicle lumen, rather than surface or intramembrane binding, we determined the effect of increasing extravesicular osmolarity. As expected for true transport, the amount of vesicle-associated [ 3 H]estrone sulfate decreased with increasing concentrations of sucrose extrapolating to zero at infinite osmolarity (Fig. 4A).
To determine the kinetic parameters of estrone sulfate transport and the effect of GSH, rates of ATP-dependent uptake in the presence and absence of 1 mM GSH were determined at several concentrations of estrone 3-sulfate (125 nM to 16 M) to obtain K m and V max values (Fig. 3C). A nonlinear regression analysis of the data yielded an apparent K m of 0.73 Ϯ 0.17 M for estrone 3-sulfate and a V max of 440 Ϯ 27 pmol mg Ϫ1 protein min Ϫ1 in the presence of GSH, and an apparent K m of 4.2 Ϯ 1.1 M for estrone 3-sulfate and a V max of 107 Ϯ 10 pmol mg Ϫ1 protein min Ϫ1 in its absence. The inset shows an Eadie-Hofstee transformation of the data (Fig. 3C).
When the nonreducing S-methyl GSH was used in place of GSH, 1 mM S-methyl GSH stimulated estrone 3-sulfate uptake by T5 membrane vesicles, somewhat more effectively than the same concentration of GSH (Fig. 5A). In contrast, the sulfhydryl reducing agents DTT (10 mM) and 2-mercaptoethanol (5 mM) and oxidized glutathione GSSG (0.05 mM) had no effect on estrone 3-sulfate transport. Since S-methyl GSH appeared to be a more potent stimulator of estrone sulfate transport than GSH, rates of transport were determined as a function of GSH or S-methyl GSH concentration (Fig. 5B). Both compounds exhibited a similar concentration dependence with 50% of maximal stimulation being observed at a concentration of ϳ0.5 mM. However, the maximal rate of estrone sulfate transport obtained with S-methyl GSH was ϳ2.5-fold greater than with GSH itself.
Inhibition of Estrone 3-Sulfate Transport by LTC 4 -Since we found that estrogen sulfates competitively inhibited MRP1mediated [ 3 H]LTC 4 transport in the presence of GSH, we determined whether the reverse was also true. Thus, the ability of LTC 4 to inhibit [ 3 H]estrone 3-sulfate transport by MRP1 was examined in T5 vesicles. GSH-enhanced estrone 3-sulfate uptake was inhibited by LTC 4 in a dose-dependent manner (Fig.  6A). An Eadie-Hofstee plot of estrone 3-sulfate uptake in the presence of 0.2 M LTC 4 indicated that the inhibition was competitive, with an apparent K i (LTC 4 ) of 0.2 M (Fig. 6B).

Effect of Sulfated Estrogens on MRP1-mediated [ 3 H]GSH
Transport-Previous studies of the GSH stimulated transport of the unconjugated xenobiotic vincristine revealed a reciprocal stimulation of GSH transport by the drug, suggesting a cotransport mechanism (10). In addition, verapamil was shown to markedly stimulate GSH transport by MRP1, but in this case no transport of verapamil could be detected (28). Consequently, we examined the ability of the estrogen sulfates to stimulate GSH transport using verapamil as a positive control. As expected, verapamil at 100 M significantly enhanced GSH transport by about 3-fold. In the first set of experiments, 80 nCi of [ 3 H]GSH was used in each reaction and no stimulation of GSH transport was observed with either estrogen sulfate at 0.2 and 2 M. A modest stimulation (25%) was observed at 20 M but only with estrone sulfate (Fig. 7A). In a second set of experiments, the amount of [ 3 H]GSH was increased to 288 nCi/ reaction to obtain a higher sensitivity of the transport assay. The modest stimulation of GSH transport by estrone sulfate was also detected at the concentration of 100 M. However, addition of 100 M estradiol sulfate inhibited rather than enhanced the low basal level of ATP-dependent [ 3 H]GSH transport, which was ϳ25-30 pmol mg Ϫ1 min Ϫ1 at an initial concentration of 100 M (Fig. 7B). This rate is approximately equivalent to the rate of [ 3 H]estrone sulfate uptake in the presence of 100 M GSH. To determine whether the basal level of ATP-dependent GSH transport observed with vesicles from MRP1 transfected cells was MRP1-mediated, we also examined the rate of ATP-dependent GSH transport by vesicles from HeLa C1 cells transfected with an empty vector and the ability of the MRP1 mAb QCRL-3 to inhibit transport. The rate of [ 3 H]GSH transport by C1 vesicles was less than 5 pmol mg Ϫ1 min Ϫ1 , and basal GSH transport by T5 vesicles was completely inhibited by QCRL-3 at 10 g ml Ϫ1 .
[ 3 H]Estradiol Sulfate Accumulation in Intact HEK Cells Transiently Expressing est or est plus MRP1-To determine whether the MRP1-mediated estrogen sulfate transport observed in vitro with membrane vesicles could also be detected in intact cells. HEK cells were transfected with pCDNA3-est alone and with pCEBV7-MRP1 together. The production and accumulation of [ 3 H]estradiol sulfate in the est-transfected cells was then determined and compared with untransfected control cells, and with cells coexpressing est and MRP1. EST specifically catalyzes estrogen sulfation, and the identities of the products have been confirmed previously by high perform- ance liquid chromatography to be estrogen 3-sulfates (31). Thus, an increased accumulation of tritium in the est-transfected cells incubated with [ 3 H]estradiol reflects the accumulation of [ 3 H]estradiol 3-sulfate, which unlike estradiol does not readily diffuse across the plasma membrane. Cytosolic extracts were also prepared from control and transfected cells and assayed directly for est activity as described previously (29,31). The level of est activity obtained with extracts from cells transfected with either the est vector alone or cotransfected with the MRP1 vector were ϳ10-fold higher than in control cells. Cotransfection with the MRP1 vector had no detectable effect on the levels of EST activity (data not shown). As shown in Fig. 8, over a 20-min period, cells transfected with the EST vector accumulated ϳ2.5-fold more tritium than control cells and cotransfection with the MRP1 vector decreased this accumulation to a level that was only 50% higher than the controls. This decrease was statistically significant (p Ͻ 0.01, unpaired t test) (Fig. 8), consistent with an MRP1-mediated efflux of sulfated estrogen from the intact cells.

DISCUSSION
Our data derived from studies with MRP1-enriched membrane vesicles and intact cells demonstrate that estrone and estradiol 3-sulfates are potential endogenous substrates for MRP1. The in vitro experiments, in which ATP-dependent transport of [ 3 H]estrone sulfate was examined directly using plasma membrane vesicles from MRP1-transfected cells, confirmed that the conjugated estrogen is an MRP1 substrate and also revealed that a physiological concentration of GSH is required for its efficient transport. Direct demonstration of MRP1-mediated estrogen sulfate efflux from intact cells was precluded by the fact that the hydrophilic conjugates enter intact cells very poorly. Consequently, it was necessary to produce the estradiol 3-sulfate intracellularly by conjugation of [ 3 H]estradiol. This was accomplished by using short term transfection to express est in HEK cells in the presence and absence of MRP1. The results demonstrated that the presence of MRP1 significantly decreased accumulation of the watersoluble estrogen conjugate produced by est.
The K m value (0.73 M) for MRP1-mediated and GSH-enhanced estrone sulfate transport is comparable with that for organic anion transporter 3 (OAT3)-or organic anion transporter 4 (OAT4)-mediated transport of estrone sulfate (34,35). However, these latter transporters have a more restricted tissue distribution than MRP1/mrp1 and are involved in the uptake rather than efflux of organic anions. The relatively high coexpression of MRP1/mrp1 with EST in Leydig cells suggests that MRP1 is likely to be involved in estrogen sulfate efflux from the testis. Since estrone sulfate can be converted back into estrone by estrogen sulfatase, efficient removal of estrone sulfate by an export pump is expected to be important for maintenance of low estrogen levels in organs such as the testis. Studies of mrp1 Ϫ/Ϫ mice treated with etoposide phosphate strongly suggest that mrp1 may also protect the testis from exposure to cytotoxic xenobiotics (17). In addition, the recent report that EST is able to sulfate environmental estrogens, such as bis-, 4-octyl-, and p-nonylphenols, raises the possibility that MRP1 or related proteins could play a role in the cellular elimination of the conjugates of these estrogenic compounds (36).
The requirement of GSH for efficient transport of the sulfated estrogen provides the first example of GSH-enhanced transport of a conjugated endogenous substrate by MRP1. Previous studies have shown that GSH is required for the transport of some xenobiotics including unmodified chemotherapeutic agents such as vincristine (8 -10), daunorubicin (9, 11) and aflatoxin B 1 (32), and enhances transport of etoposide conjugated with glucuronate (37). GSH has been reported recently to stimulate transport of [ 3 H]luteolin 7-O-diglucuronyl-4Ј-O-glucuronide in plant leaves, but the transporter involved has not been characterized structurally. However, it appears functionally to be a membrane-potential sensitive member of the ABC superfamily (38).
We have previously proposed that MRP1 contains a bipartite site to which the hydrophobic and anionic moieties of its conjugated substrates bind (8). Studies of the GSH-stimulated transport of vincristine, and the influence of GSH on the ability of vincristine to compete for LTC 4 transport by MRP1, indicate that GSH not only increases the V max for the drug but also the affinity with which it interacts with the protein, as reflected by an approximate 20-fold increase in its inhibitory potency. These studies also indicated that vincristine reciprocally increases the affinity of MRP1 for GSH (10). The data suggest that initial low affinity binding of either GSH or drug by MRP1 induces a conformational change in the protein such that high affinity binding of the cotransported substrate can occur. In this respect, the model differs from the two-site model proposed by Borst et al. (39) in which the protein is envisaged to have two binding sites: one that has a relatively high affinity for drug and low affinity for GSH and another with high affinity for GSH and low affinity for drug (39). Although this model may explain the ability of drug and GSH to reciprocally stimulate cotransport, it is difficult to explain why GSH markedly increases the ability of drug to competitively inhibit transport of high affinity, conjugated substrates such as LTC 4 . Consistent with the existence of a low affinity site for GSH alone, our photoaffinity experiments with [ 3 H]LTC 4 indicate that GSH and S-methyl GSH decrease [ 3 H]LTC 4 labeling of MRP1 in a concentration-dependent manner in the millimolar range, suggesting that the binding sites for LTC 4 encompass a low affinity site to which GSH (or S-methyl GSH) can bind.
Recent studies of drug binding and transport by the bacterial, homodimeric ABC multidrug resistance protein, Lmra, suggest positive cooperativity between allosterically linked low and high affinity substrate binding sites (40). A "two-cylinder engine" model, which embodies the alternating catalytic sites model proposed by Senior et al. (41), has been proposed in which the transport of drug from a high affinity intracellular site to a low affinity extracellular site on the same subunit of the homodimer is driven by the alternating hydrolysis of ATP at one or the other nucleotide binding domain. Thus, the two subunits cycle with respect to exposure of high or low affinity sites (40).
The evidence for positive cooperativity is based in part on the observation that two substrates that compete for transport at high concentrations reciprocally stimulate transport of each other at lower concentrations. Unlike the bacterial homodimeric transporters and transporters such as P-glycoprotein, the nucleotide binding domains of the MRP-related proteins are relatively divergent. Evidence to date suggests that they are also not functionally equivalent and may not alternate catalytically (42). Similarly, there is no evidence of primary structure conservation between the NH 2 and COOH-proximal membrane-spanning domains of MRP1. However, we have consistently observed that low concentrations of the estrogen sulfates in the absence of GSH stimulate LTC 4 transport and binding, as evidenced by photocross-linking studies, while they compete at higher concentrations. Thus, the data are also consistent with the existence of interacting binding sites being present on MRP1. However, since both the estrogen sulfates and LTC 4 are relatively hydrophilic and the duration of transport assays is extremely short, the stimulation observed suggests that the sites are accessible from the cytoplasmic face of the membrane. In addition, the stimulation of binding occurs in the complete absence of nucleotide, indicating that if allosteric changes in structure occur following initial interaction with substrate, they do not require the binding and/or hydrolysis of ATP.
In some respects, the results we have obtained in the present study with estrone sulfate are similar to previous observations on the effect of GSH on rate of transport and apparent binding affinity of unmodified hydrophobic substrates such as vincristine, despite the fact that the compound is conjugated. However, in the case of vincristine, aflatoxin B 1 , and daunorubicin, it has not been possible to determine kinetic parameters of transport in the absence of GSH. Our data demonstrate that GSH decreases the K m for estrone sulfate from 4.2 to 0.73 M and increases V max from 107 to 440 pmol min Ϫ1 mg protein Ϫ1 . In addition, GSH also markedly enhanced the ability of estrone sulfate to inhibit photolabeling of MRP1 with LTC 4 . In these experiments, the GSH-enhanced inhibitory potency of estrone sulfate was obtained in the absence of ATP, indicating that any induced change in affinity for the conjugated estrogen occurs without a requirement for either nucleotide binding or hydrolysis.
Earlier studies with conjugated estrogens indicated that E 2 17␤G could compete for LTC 4 binding and transport by MRP1 while estrogens conjugated with glucuronide at the 3 position of the A-ring are very poor inhibitors (5). These studies also revealed that a change in site of glucuronidation from the 17␤ to 16␣ position on the D-ring markedly decreased the affinity for the protein, as judged by the difference between the K i values for E 3 17␤G (1.4 M) and E 3 16␣G (45 M) as inhibitors of E 2 17␤G transport. Taken together, the data indicate that strict structural requirements with respect to the site of glucuronidation of the steroid nucleus must be met to interact with MRP1 as a substrate or competitive inhibitor. Sulfation at the 3-position of the A-ring of E 2 17␤G had little or no effect on its ability to compete for transport, implying that the presence of the sulfate group neither precluded nor enhanced interaction with the protein. In contrast, the conjugated bile salt glycolithocholate 3-sulfate was an effective inhibitor of E 2 17␤G transport when compared with bile salts that were not conjugated at this position of the A-ring. This is consistent with the possibility that sulfation of the A-ring, in the absence of additional anionic conjugation, might enhance interaction with MRP1. A low level of transport of estrone sulfate could be detected in the absence of GSH, suggesting that sulfation resulted in the formation of a relatively low affinity, low capacity substrate in which the sulfate presumably does not prevent GSH from interacting with the protein, either because it binds to a different site or because it interacts only weakly with the GSH binding site. This latter possibility would be consistent with the inhibition of basal MRP1-mediated GSH transport by HeLa T5 vesicles observed at high concentrations of estradiol sulfate. However, the GSH has no effect on the inhibitory potency of other A-ring conjugates, such as E 2 3␤G or on the transport of E 2 17␤G, or estradiol itself (data not shown). Thus with respect to stimulation of transport by MRP1, it remains difficult to predict which conjugated or nonconjugated compounds might be affected by the presence of GSH.
It has been observed that GSH levels are increased in some tissues of mrp knockout mice (15) and decreased in drug-selected or transfected cells that overexpress MRP1 (12,13), suggesting that MRP1 might efflux GSH either alone or via cotransport with currently unidentified, endogenous substrates. The stimulation of estrogen sulfate transport by GSH suggested that these compounds might be candidates for such substrates. Thus far, vesicle transport studies have provided strong evidence for cotransport of GSH with some xenobiotics (15,10). However, in other cases, it has not been possible to detect a xenobiotic-dependent stimulation of GSH transport (10,11). In addition, we have shown that some compounds, such as verapamil, can markedly stimulate MRP1-mediated GSH transport with no detectable net transport of the compound itself (28).
Despite the readily demonstrable GSH stimulation of estrone sulfate transport, we have not been able to detect reciprocal stimulation of GSH transport by the conjugated estrogen. No significant increase in GSH uptake by membrane vesicles could be detected in the presence of either estradiol or estrone sulfate over a 100-fold range of conjugated estrogen concentrations (0.2-20 M). However, because of practical limitations, the highest concentration of [ 3 H]GSH used in these studies is ϳ100 M, and at this concentration the rate of estrone sulfate transport is ϳ25-30 pmol min Ϫ1 mg protein Ϫ1 . Using [ 3 H]GSH with a 7-fold higher specific activity than used in previous studies, we were able to detect a low but significant rate of ATP-dependent, MRP1-mediated GSH transport in the absence of a second substrate. This rate of basal GSH transport was approximately equivalent to the rate of estrone sulfate transport under the experimental conditions used. This leaves open the possibility that the "basal" rate of GSH transport by the protein is adequate under the experimental conditions to support cotransport of some substrates. However, this would imply that the interaction of substrates such as the estrogen sulfates, in contrast to compounds like verapamil, does not significantly affect the binding and transport of GSH. These studies expand the range of potential MRP1 substrates that require GSH for efficient transport to include certain hydrophilic anionic as well as previously identified hydrophobic, unconjugated compounds.