Coordinated Action of Glutathione S-Transferases (GSTs) and Multidrug Resistance Protein 1 (MRP1) in Antineoplastic Drug Detoxification

To examine the role of multidrug resistance protein 1 (MRP1) and glutathione S-transferases (GSTs) in cellular resistance to antineoplastic drugs, derivatives of MCF7 breast carcinoma cells were developed that express MRP1 in combination with one of three human cytosolic isozymes of GST. Expression of MRP1 alone confers resistance to several drugs representing the multidrug resistance phenotype, drugs including doxorubicin, vincristine, etoposide, and mitoxantrone. However, co-expression with MRP1 of any of the human GST isozymes A1-1, M1-1, or P1-1 failed to augment MRP1-associated resistance to these drugs. In contrast, combined expression of MRP1 and GST A1-1 conferred ∼4-fold resistance to the anticancer drug chlorambucil. Expression of MRP1 alone failed to confer resistance to chlorambucil, showing that the observed protection from chlorambucil cytotoxicity was absolutely dependent upon GST A1-1 protein. Moreover, using inhibitors of GST (dicumarol) or MRP1 (sulfinpyrazone), it was shown that in MCF7 cells resistance to chlorambucil requires both intact MRP1-dependent efflux pump activity and, for full protection, GST A1-1 catalytic activity. These results are the first demonstration that GST A1-1 and MRP1 can act in synergy to protect cells from the cytotoxicity of a nitrogen mustard, chlorambucil.

trophilic alkylating agents, can form conjugates with glutathione both spontaneously and in GST-catalyzed reactions (7)(8)(9)(10)(11)(12)(13)(14). Despite these catalytic activities, the role of GSTs in the protection of cells from the cytotoxicities of these cancer drugs remains equivocal due to the inconsistent results obtained by different laboratories (5,(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25). Although some investigators have found associations between cellular resistance to some anticancer drugs and expression of particular isozymes of GST, other investigators have found no such associations in other cell lines. In MCF7 breast carcinoma cells, increased expression of Alpha, Mu, or Pi class GSTs failed to confer any consistent, significant resistance to a variety of anticancer drugs, including drugs known to be substrates of GSTs (19,20,24). We recently proposed that conjugation of some of these drugs and toxins with glutathione may represent only partial detoxification. In this view, export of the glutathione conjugate is required to fully potentiate the GST-mediated protection. The identification of MRP1 as an important glutathione-conjugate efflux pump (26 -30) raises the possibility that MRP1 and GST may act in synergy to confer cellular resistance to some of these compounds (31). This hypothesis was recently validated for the model carcinogen, 4-nitroquinoline-1-oxide (31). These studies showed that GST P1-1-associated protection from 4-nitroquinoline-1-oxide cytotoxicity was dependent upon concomitant expression of MRP1. Additionally, GST P1-1-associated protection from 4-nitroquinoline-1-oxide-mediated DNA adduct formation was greatly enhanced by co-expression of MRP1.
Previous studies failed to demonstrate an association between increased MRP1 expression and resistance to alkylating agents in MCF7/VP cells (32). In another study examining paired cell lines that differed in the levels of MRP1 expressed, increased MRP1 was associated with chlorambucil resistance in one but not the other two paired cell lines (33). These inconsistent relationships between increased MRP1 and alkylating agent sensitivity indicate that other factors, such as GST or glutathione levels, may be important in determining whether or not MRP1 will mediate protection from the cytotoxicities of some alkylating cancer drugs. Additionally, properties of some of the bifunctional alkylating agents and their metabolites make them particularly interesting candidates for GST/ MRP1-mediated detoxification. Even after formation of monoglutathionyl derivatives, some of these bifunctional alkylating agents retain significant reactivity at the unmodified alkylating group and may therefore remain cytotoxic. Moreover, glutathione conjugates of some of these compounds, the nitrogen mustards melphalan and chlorambucil, are known to be transported by MRP1-containing membrane vesicles in vitro (27,33).
We have developed cellular models to examine the role of GST/MRP1 synergy in the emergence of anticancer drug resistance. Although cell-free, in vitro analyses of toxin conjugation and transport are informative, cellular models of MRP1 and GST function are essential to determine the cytoprotective consequences of coordinated MRP1 and GST expression and to determine the precise mechanisms of cellular detoxifications. Accordingly, from MCF7 cells that express extremely low levels of MRP1 and GST, we have developed derivative sublines that express MRP1 alone or in combination with representatives of three major classes of human cytosolic GSTs, GST A1-1, M1-1, and P1-1. Results show that GST A1-1 operates in synergy with MRP1 to confer resistance to the antineoplastic nitrogen mustard chlorambucil. The studies indicate the mechanism of synergy involves GST catalytic activity as well as MRP1-mediated efflux of glutathione conjugates.
In some experiments, cells were preincubated for 15 min in the presence of vehicle or inhibitor (0.1 mM dicumarol (in HBSS) or 2 mM sulfinpyrazone (in DMEM plus 1% fetal bovine serum)) followed by co-incubation of chlorambucil with vehicle or inhibitor (0.1 mM dicumarol (in HBSS) or 2 mM sulfinpyrazone (DMEM/1% fetal calf serum)) for 1 h.
Preparation of Affinity-purified, Recombinant Human GST A1-1-The human GSTA1 cDNA (36) was ligated into the EcoRI site of the pPROK-1 plasmid expression vector (CLONTECH). The protein was expressed following transformation into E. coli DH5a. The culture was grown overnight and then chilled, pelleted, and resuspended in 0.1 volume of 10 mM KPO4 (pH 8) plus 5 mM EDTA. The suspension was warmed to 20°C, incubated in lysozyme (0.2 mg/ml) for 10 min, chilled on ice, sonicated, and then centrifuged at 10,000 ϫ g for 20 min. The supernatant was adjusted to 0.2 M NaCl and loaded onto a glutathione affinity column (Sigma G-4510) equilibrated with 50 mM Tris buffer (pH 7.4) plus 0.2 M NaCl. The column was washed with 10 volumes of equilibration buffer, and then GSTA1-1 was eluted with buffer containing 50 mM Tris (pH 9.6), 0.2 M NaCl, and 5 mM glutathione. The eluate was neutralized with dilute HCl, dialyzed against 10 mM KPO 4 (pH 7.4) plus 50% glycerol, and stored at Ϫ20°C.
Biochemical and in Vitro GST Enzyme Analyses-Glutathione levels were measured by the enzymatic recycling method (39). GST activities were determined using 1-chloro-2,4-dinitrobenzene (CDNB) as substrate (40). Enzyme kinetic and in vitro inhibition analyses were accomplished using affinity-purified, recombinant human GST A1-1. For these assays, CDNB served as the varied substrate, and glutathione concentration was fixed at 2 mM. Activity was monitored spectrophotometrically at 25°C and pH 6.5 as described by Habig et al. (40).
Northern blots of total cellular RNA were done as described (41) using probes derived from the entire human GST A1, GST M1, or GST P1 cDNAs (36,37,42) or the 5Ј 2629 base pairs of the MRP1 cDNA (43). The cDNA probes were labeled with [␣-32 P]dCTP by random priming (44). Western blot analyses of GST isoforms were accomplished as described previously using affinity-purified rabbit polyclonal antibodies directed against human GST A1, GST M1, or GST P1 subunits (19). Western blotting of membrane protein preparations for MRP1 was done with the QCRL-1 antibody kindly provided by Dr. S. P. C. Cole (45).
In Vivo (Intact Cells) Analyses of the Effects of Inhibitors on MRP1mediated DNP-SG Efflux and GST Activity-For the analysis of DNP-SG efflux, MCF7/VP␣ cells were plated to a density of 0.6 ϫ 10 6 cells/well in 6-well plates. 24 h later, cells were preincubated for 2 min in HBSS plus vehicle or 2 mM sulfinpyrazone. Following preincubations, cells were co-incubated at 25°C with 1 M CDNB plus vehicle or 2 mM sulfinpyrazone. At the indicated times, medium was removed and acidified to 10% perchloric acid. Cells were lysed in 10% perchloric acid. Medium and lysate samples were prepared for reverse phase high performance liquid chromatography analysis as described (31). Details of DNP-SG chromatography will be described elsewhere. 2 Briefly, acidsoluble samples were eluted isocratically from a C18 reverse phase column in 20% methanol ϩ 0.1% trifluoroacetic acid. Chromatographs were monitored spectrophotometrically at 340 nm, and DNP-SG levels were determined from the areas of peaks eluting at the position of authentic DNP-SG by comparison with DNP-SG standards. Concentrations of DNP-SG reference standards were determined using ⑀ 340 ϭ 10.2 cm Ϫ1 mM Ϫ1 .
The effect of sulfinpyrazone or dicumarol on GST activity in vivo was determined as follows. MCF7/VP␣ cells (0.5 ϫ 10 6 /well) were seeded in 6-well plates. 24 h later, cells were incubated in DMEM plus 1% fetal calf serum Ϯ 2 mM sulfinpyrazone or in HBSS Ϯ 0.1 mM dicumarol at 37°C in 5% CO 2 . At indicated times, cells were washed and incubated in HBSS equilibrated to 25°C containing inhibitor (2 mM sulfinpyrazone or 0.1 mM dicumarol) or vehicle plus 10 M CDNB. Relative in vivo GST activities were determined as the rate of total DNP-SG formation (DNP-SG in medium plus cell lysates) for 2 and 5 min at 25°C. DNP-SG levels were measured chromatographically as described above. The levels of DNP-SG formation increased linearly with time throughout the 5 min incubation periods.

RESULTS
Characterization of Cell Lines-MCF7 breast carcinoma cells were chosen to study the effects of expression of MRP1 alone or in combination with isozymes of GST. These cell lines constitute a good model system for such studies because parental MCF7/WT cells express extremely low levels of cytosolic GSTs and MRP1. Additionally, a MDR derivative of MCF7/WT, MCF7/VP, is available that has as its primary genetic and phenotypic change the overexpression of high levels of MRP1 (32). Like MCF7/WT, MCF7/VP cells have very low endogenous cytosolic GST levels. Neither MCF7/WT nor MCF7/VP contain detectable MDR1 (P-glycoprotein) or the MRP2 (cMOAT or cMRP) isoform (32). 3 Parental MCF7/VP cells were stably transfected with expression vectors containing cDNAs encoding the human GST A1-1 (MCF7/VP␣), GST M1-1 (MCF7/VP), or GST P1-1 (MCF7/ VP). Additionally, control cell lines were generated by stable transfection of parental MCF7/WT and MCF7/VP cells with empty expression vectors encoding antibiotic resistance to hygromycin (hyg) or geneticin (neo) but not GST. Expression of GST isozyme-specific protein and mRNA was detected only in cell lines transfected with GST expression vectors (Fig. 1). In these cells (MCF7/VP␣, MCF7/VP, and MCF7/VP), relatively high levels of GST activity were confirmed by enzyme assays (Table I). GST activities remained uniformly low in the parental and control cell lines (Table I). High levels of MRP1 expression (membrane protein and mRNA) were seen only in parental MCF7/VP cells and their transfected derivatives. Coexpression of GST isozymes had no effect on the levels of MRP1 (Fig. 2). Finally, glutathione levels were comparable in all of the cell lines tested (Table I).
Transfection of control vectors devoid of GST cDNA inserts into parental MCF7/WT and MCF7/VP cells had no effect on GST levels (MCF7/WT-neo and -hyg; MCF7/VP-neo and -hyg), glutathione levels (MCF7/WT-neo and MCF7/VP-neo), or MRP1 expression (Table I). Moreover, cytotoxicity profiles of parental cells MCF7/WT and MCF7/VP were similar to the corresponding cells transfected with control vectors (MCF7/ WT-neo and -hyg; MCF7/VP-neo and -hyg). Thus, data reported in the text below will not differentiate between the parental cells and their corresponding derivatives transfected with control vectors.
Effect of GSTs on the Sensitivities to Drug Cytotoxicities of MRP1-expressing MCF7 Cells-The effect of MRP1 expression, alone or in combination with three different isozymes of cytosolic GST, on cellular sensitivities to various cytotoxic anticancer drugs was tested. Many of these data are summarized in Table II. A comparison of the relative resistance of MCF7/WT versus MCF7/VP cells to four drugs of the MDR phenotype, VP-16, doxorubicin, vincristine, and mitoxantrone, confirmed previous findings (32) that increased expression of MRP1 alone confers resistance to these drugs. Although VP-16, doxorubicin, and vincristine are not known to form conjugates with glutathione, we wondered whether these drugs or their metabolic derivatives might be unrecognized, toxic substrates of any of the three cytosolic GST isozymes tested and, if so, whether the GST isozymes would augment MRP1-mediated resistance to these drugs. Although mitoxantrone can form glutathione conjugates, it is not known whether any of the cytosolic GSTs tested can catalyze these reactions. However, as shown in Table II, expression of relatively high levels of GST A1-1 (MCF7/VP␣ cells), GST M1-1 (MCF7/VP cells), or GST P1-1 (MCF7/VP cells) had no significant effect on the level of MRP1-associated resistance to these MDR-related drugs.
In contrast to the MDR phenotype-associated drugs, a variety of alkylating anticancer drugs are known to form glutathione conjugates both spontaneously and in GST-catalyzed reac-tions (7)(8)(9)(10)(11)(12)(13)(14). However, MRP1 has not consistently been associated with resistance to these drugs. Nevertheless, a number of glutathione conjugates, including monoglutathionyl derivatives of chlorambucil and melphalan, are known to be substrates of MRP1-dependent transport in isolated membrane vesicles (27,33). Therefore, we investigated whether MRP1 could support resistance to some of these alkylating agents if co-expressed with the GST isozyme appropriate to accelerate conjugate formation. There are reports that implicate GST catalysis in detoxification reactions involving all of the alkylating agents listed in Table II (13, 15-18, 21-23, 25, 46). However, neither MRP1 nor the isozymes of GST tested had any significant effect on relative resistance to melphalan, 1,3-bis(2chloroethyl)-1-nitrosourea, thiotepa, or hepsulfan. In contrast, co-expression of GST A1-1 and MRP1 conferred significant resistance to chlorambucil, a nitrogen mustard closely related to melphalan (Fig. 3). Indeed, cumulative results from 12 independent experiments showed that GST A1-1 expression is associated with ϳ4-fold (3.74 Ϯ 0.19 S.E.) resistance to chlorambucil in MCF7/VP␣ cells compared with MCF7/VP cells (p Ͻ 0.0001). Resistance to chlorambucil was GST A1-1 isozyme-specific because co-expression of MRP1 with GST M1-1 or GST P1-1 had no effect on cellular sensitivity to chlorambucil (not shown).

Expression of GST A1-1 Protein and Catalytic Activity Are
Required for MCF7/VP␣ Resistance to Chlorambucil-Inspection of cytotoxicity profiles of MCF7/WT versus MCF7/VP reveals that MRP1 alone does not confer resistance to chlorambucil (Fig. 3). Indeed, as shown here and reported previously (32), expression of MRP1 alone is associated with modest sensitization to chlorambucil cytotoxicity. This indicates that the resistance to chlorambucil in MRP1-positive cells is absolutely dependent upon GST A1-1 protein expression. To confirm this relationship and to eliminate the possibility that MCF7/VP␣ is an exceptional MCF7/VP clone, in which the resistance to chlorambucil is independent of GST A1-1, we tested whether inhibition of GST catalytic activity in vivo would reverse resistance in MCF7/VP␣ cells.
Dicumarol was chosen as a candidate inhibitor because it has been reported to inhibit a rodent Alpha class GST and was expected to inhibit the human ortholog, GST A1-1 (47). Additionally, dicumarol is lipophilic and therefore should freely traverse the cell membrane of intact cells. To determine whether the inhibitory properties and potency of dicumarol would likely be sufficient to inhibit cellular GST A1-1 at subcytotoxic levels, we first examined the kinetics of dicumarol inhibition of the conversion of CDNB to DNP-SG by purified, recombinant GST A1-1 in vitro. These results (Fig. 4A) show that dicumarol inhibition is of the mixed type and is quite potent for this GST A1-1-catalyzed reaction (K ic ϭ 3.6 M; K iu ϭ 4.4 M) (48). The effect of exogenously added dicumarol on GST A1-1 activity in vivo is examined in Fig. 4B. These data show that 0.1 mM dicumarol inhibited GST activity in MCF7/ VP␣ cells by at least 80%, that inhibition was maximal within 15 min, and that inhibition remained at this level throughout the 1-h incubation. Treatment of intact cells with 0.1 mM di-cumarol had no effect on the kinetics of MRP1-dependent DNP-SG efflux (data not shown). Based upon these preliminary results, the effect of GST A1-1 inhibition by dicumarol on MCF7/VP␣ cell sensitivity to chlorambucil was examined.
Treatment of MCF7/VP␣ cells with 0.1 mM dicumarol sufficient to inhibit GST activity by Ն 80% resulted in nearly complete reversal of GST A1-1-associated resistance (Fig. 5A). Moreover, dicumarol-mediated sensitization was selective for GST A1-1-expressing MCF7/VP␣ cells because dicumarol had no significant effect on the sensitivities of MCF7/WT and MCF7/VP cells to chlorambucil cytotoxicity (Fig. 5, A and B).  These results demonstrate that GST A1-1 catalytic activity is indeed required for chlorambucil resistance in MCF7/VP␣ cells. However, these data leave undefined the role of MRP1 in this resistance to chlorambucil.

MRP1-dependent Glutathione Conjugate Efflux Activity Is Required for Chlorambucil Resistance in MCF7/VP␣ Cells-To
evaluate the requirement for MRP1 in the observed chlorambucil resistant phenotype, we used sulfinpyrazone to inhibit MRP1-mediated efflux of glutathione conjugates in vivo. The concentration of sulfinpyrazone added (2 mM) was without significant cytotoxicity. Moreover, addition of 2 mM sulfinpyrazone had very little effect on in vivo GST A1-1 activity over the entire 1-h incubation period (Fig. 6A). However, under these same conditions, sulfinpyrazone effectively inhibited MRP1-dependent efflux of the glutathione conjugate, DNP-SG, (Fig. 6B,  top panel) to efflux rates comparable to those seen when MRP1dependent efflux is inhibited by ATP depletion (31). 4 Additionally, inhibition of efflux by sulfinpyrazone resulted in a profound increase in intracellular accumulation of DNP-SG (Fig.  6B, bottom panel). The effect of similar treatments with sulfinpyrazone on chlorambucil cytotoxicity is reported in Fig. 7. Sulfinpyrazone completely reverses MCF7/VP␣ resistance to chlorambucil but has little effect on MCF7/WT or MCF7/VP cells (Fig. 7, A and B). Thus, sulfinpyrazone sensitization to chlorambucil toxicity is selective for GST A1-1-expressing cells.
We conclude that MRP1-dependent glutathione conjugate efflux activity is absolutely required to potentiate GST A1-1mediated resistance to chlorambucil in MCF7 cells under the conditions used. DISCUSSION Our results establish that MRP1 and GST A1-1 act in synergy to confer resistance to chlorambucil in MCF7 cells. The co-dependence of resistance upon both GST A1-1 and MRP1 is remarkable. First, that GST A1-1 activity alone does not confer protection indicates that the glutathione conjugate(s), or other metabolites, of chlorambucil, the formation of which is favored in the presence of GST A1-1, may be toxic to the cells. Because the conjugate(s) and other derivatives of chlorambucil formed are more water-soluble than the parent compound and, therefore, less permeable to the plasma membrane, they may accumulate to high intracellular levels in the absence of MRP1 or another suitable efflux mechanism. This may explain why GST A1-1, at the levels achieved, is insufficient to confer protection from chlorambucil cytotoxicity in the absence of MRP1-dependent efflux activity. It is perhaps even more remarkable that MRP1 alone fails to confer protection from chlorambucil cytotoxicity. This finding is particularly significant because glutathione conjugation of chlorambucil can occur non-enzymatically (10, 49) and because glutathione conjugates of chlorambucil, especially the monoglutathionyl derivative, are 4 S. Diah, unpublished data. reportedly substrates of MRP1-dependent transport in isolated membrane vesicles (33). Thus, it is not immediately obvious why MRP1 alone does not afford some protection from chlorambucil toxicity. These issues have important implications for understanding the dynamics of drug detoxification and the relative importance of the components of drug detoxification studied-components that include phase II drug conjugation (GST/glutathione) and phase III drug/conjugate efflux (MRP1).
GST A1-1 is known to catalyze the formation of the monoglutathionyl, but not the diglutathionyl, derivative of chlorambucil from glutathione and chlorambucil (49). However, the catalytic rate enhancement is relatively modest (10,11,49). Indeed, Meyer et al. (49) suggest that, rather than increasing the overall extent of conjugation with glutathione, the major effect of the human GST alpha class isoforms, A1-1, A1-2, and A2-2 on chlorambucil metabolism may be to increase the proportion of monoglutathionyl chlorambucil relative to diglutathionyl chlorambucil and other derivatives (including hydroxyand phosphate-substituted metabolites) (49). This altered profile is believed to be the combined consequence of 1) GST Alpha class catalysis of monoglutathionyl chlorambucil formation, and 2) the ability of these GSTs to sequester monoglutathionyl chlorambucil at the enzyme active site with high affinity (49).
Because the monoglutathionyl chlorambucil retains one of the reactive chloroethyl groups, this metabolite of chlorambucil is only partially detoxified. Although GST A1-1 may sequester the monoglutathionyl derivative and thereby mitigate some toxicity, with continued exposure to chlorambucil, this GST: monoglutathionyl chlorambucil interaction will become saturated and will inhibit further catalysis (49,50). Thus, in the absence of MRP1-dependent efflux of the monoglutathionyl derivative, the detoxification capacity of GST A1-1 may be quickly exceeded. It is possible that the other derivatives may also have some toxicities at high intracellular levels.
Our transfection data show that GST A1-1 protein is required for MRP1-associated resistance to chlorambucil in MCF7 cells. Moreover, results from in vivo inhibition of GST A1-1 with dicumarol indicate that GST catalytic activity is also an important requirement for maximum resistance. We have recently confirmed that purified GST A1-1 does bind radiolabeled monoglutathionyl chlorambucil with considerable avidity. 3 Hence, both catalytic activity and chlorambucil conjugate binding could be important mechanisms of GST A1-1-associated cytoprotection. In our view, GST A1-1 may serve to catalyze the substitution of glutathione to one of the chloroethyl groups of chlorambucil. These monoglutathionyl derivatives, formed both enzymatically and non-enzymatically, can then be sequestered as relatively benign complexes with GST A1-1 until they can be delivered to MRP1 for export. In the absence of MRP1, as monoglutathionyl chlorambucil accumulates intracellularly, the catalytic activity GST A1-1 is compromised by product inhibition and the binding capacity of the enzyme for monoglutathionyl chlorambucil is exceeded. Consequently, the levels of chlorambucil remain high, and the levels of reactive, free monoglutathionyl chlorambucil and its other derivatives accumulate intracellularly resulting in increased cytotoxicity. In the absence of GST A1-1, not only is the monoglutathionyl derivative free to react with cellular macromolecules, but the distribution of chlorambucil metabolites is shifted away from the monoglutathionyl forms to other derivatives that are significantly poorer substrates for MRP1-mediated efflux. The potential importance of such GST A1-1-dependent changes in the profile of chlorambucil glutathione conjugates and derivatives is underscored by the recent findings of Barnouin et al. (33). These investigators show that the monoglutathionyl chlorambucil is by far the best chlorambucil derivative for MRP1-dependent transport in isolated membrane vesicles in vitro.
The three isozymes of GST tested confer no protection to MRP1-expressing MCF7/VP cells against the cytotoxicities of the four other alkylating agents examined. The reason for this is unknown. The GSTs examined may have little impact in vivo on the metabolism of thiotepa, hepsulfan, or 1,3-bis(2-chloroethyl)-1-nitrosourea. Additionally, it is not known whether MRP1 can support the efflux of glutathione conjugates of thiotepa or hepsulfan. Some GST isozymes are reported to catalyze the denitrosation of 1,3-bis(2-chloroethyl)-1-nitrosourea, but these studies have not identified stable glutathione conjugate intermediates (15,22). Thus, there may be no stable glutathione conjugate of 1,3-bis(2-chloroethyl)-1-nitrosourea to serve as a potential MRP1 substrate. Particularly interesting is the failure of GST A1-1 and MRP1 to confer resistance to melphalan, a nitrogen mustard closely related to chlorambucil. The explanation for the difference in melphalan and chlorambucil resistance is unknown. Structural differences in these drugs may result in GST A1-1 having distinctly different effects on the metabolic profiles of melphalan and chlorambucil or their glutathione conjugates in vivo. Alternatively, glutathione conjugates of melphalan may be less efficiently exported by MRP1. Regardless of the explanation, our results show that MRP1/ GST resistance synergy is both specific for a particular GST isozyme:drug pair and highly drug-selective, even among structurally related drugs.
The magnitude of MRP1-mediated resistance to four drugs of the MDR phenotype was not significantly augmented by coexpression of GST A1-1, GST M1-1, or GST P1-1. This result was not surprising for the drugs doxorubicin, VP-16, and vincristine, which are not known to form stable conjugates with glutathione. However, it was important to examine these drugs because it was possible that unrecognized metabolites or derivatives of these compounds might form glutathione conjugates that could influence their export by MRP1. In contrast, mitoxantrone reacts with glutathione to form conjugates (51). However, mitoxantone is not known to be a substrate of cytosolic GSTs, although it is reported to be a substrate of microsomal GST (52). The failure of the cytosolic GSTs tested to augment MRP1-mediated resistance to mitoxantrone is consistent with these data.
In summary, we demonstrate that GST A1-1 can act in synergy with MRP1 to confer resistance to the alkylating agent chlorambucil. The concept of coordinated action of phase II GST-dependent and phase III MRP1-dependent (and other membrane transport proteins) processes will very likely prove to be of general importance for detoxification of a variety of xenobiotic compounds, electrophiles that include genotoxic carcinogens (31) and anticancer drugs. The mechanism of GST A1-1/MRP1 resistance to chlorambucil depends upon both GST A1-1 catalytic activity and MRP1-dependent efflux activity. Moreover, because MRP1 is required to potentiate GST A1-1associated resistance to chlorambucil, this suggests the possibility that some of the glutathione conjugates or other metabolites may themselves be important cellular toxins. Additionally, the removal of these conjugates by MRP1 may be required to relieve product inhibition and thereby maintain continued GST catalysis of drug conjugation. Finally, GST/ MRP1 resistance synergy is highly specific for the particular xenobiotic: GST isozyme pair. There are multiple known and putative membrane-associated transport proteins related to MRP1 (53). It is possible that the complete detoxification of xenobiotic-glutathione conjugates will prove to be similarly dependent upon the specific membrane-associated transport protein expressed.