Expression of Multidrug Resistance Protein/GS-X Pump and γ-Glutamylcysteine Synthetase Genes Is Regulated by Oxidative Stress*

Expression of the MRP1 gene encoding the GS-X pump and of the γ-GCSh gene encoding the heavy (catalytic) subunit of the γ-glutamylcysteine synthetase is frequently elevated in many drug-resistant cell lines and can be co-induced by many cytotoxic agents. However, mechanisms that regulate the expression of these genes remain to be elucidated. We report here that like γ-GCSh, the expression of MRP1 can be induced in cultured cells treated with pro-oxidants such astert-butylhydroquinone, 2,3-dimethoxy-1,4-naphthoquinone, and menadione. Intracellular reactive oxygen intermediate (ROI) levels were increased in hepatoma cells treated withtert-butylhydroquinone for 2 h as measured by flow cytometry using an ROI-specific probe, dihydrorhodamine 123. Elevated GSH levels in stably γ-GCSh-transfected cell lines down-regulated endogenous MRP1 and γ-GCShexpression. ROI levels in these transfected cells were lower than those in the untransfected control. In the cell lines in which depleting cellular GSH pools did not affect the expression of theMRP1 and γ-GCSh genes, only minor increased intracellular levels of ROIs were observed. These results suggest that intracellular ROI levels play an important role in the regulation ofMRP1 and γ-GCSh expression. Our data also suggest that elevated intracellular GSH levels not only facilitate substrate transport by the MRP1/GS-X pump as previously demonstrated, but also suppress MRP1 and γ-GCShexpression.

Human MRP1 (multidrug resistance protein) encoded by MRP1 was first isolated by molecular cloning from doxorubicinselected multidrug-resistant lung cancer cells (Ref. 1; reviewed in Ref. 2). Studies using plasma membrane vesicles prepared from MRP1-overproducing cell lines demonstrated increased ATP-dependent, high-affinity transport activities of cysteinyl leukotrienes (e.g. LTC 4 ) 1 (3,4). Deletion of homologous MRP1 alleles in mice results in impaired response to inflammatory stimulus in these animals because LTC 4 is a potent mediator of the inflammation reaction (5). These findings suggest that MRP1 encodes the previously described GS-X (ATP-dependent glutathione S-conjugate export) pump (6). In addition to transporting LTC 4 and its related glutathione S-conjugates, naturally occurring organic conjugates, including 17␤-estradiol (17␤-D-glucuronide), and bile salt conjugates, including 6␣-glucuronosylhydrodeoxychlorate and 3␣-sulfatolithocholytaurine, are also good substrates for the MRP1/GS-X pump (7)(8)(9). There are also reports suggesting that GSH may serve as a cofactor in MRP1/GS-X pump-mediated drug transport (8,11). In addition, the MRP1/GS-X pump is responsible for the release of GSSG from cells. This active export of GSSG is considered to be an important mechanism to maintain the reduced status of intracellular thiols under oxidative stress (12,13). These observations underscore the importance of GSH for the function of MRP1.
Biosynthesis of GSH is controlled by multiple enzyme systems (reviewed in Refs. 14 and 15). The first step of GSH biosynthesis, catalyzed by ␥-glutamylcysteine synthetase (EC 6.3.2.2), is the rate-limiting step. The mammalian ␥-glutamylcysteine synthetase holoenzyme is a heterodimer consisting of a 73-kDa heavy subunit (␥-GCSh) (16,17) and a 28-kDa light subunit (18,19). Although the heavy subunit contains the entire catalytic activity, its activity can be modulated by the association with the light subunit, the regulatory subunit. To facilitate the MRP1/GS-X pump-mediated transport, it is speculated that activities, particularly those of ␥-GCSh, could be increased to furnish intracellular GSH. Indeed, we recently demonstrated the frequent coexpression patterns of MRP1 and ␥-GCSh mRNAs in many drug-resistant cell lines (20 -22) as well as in human colorectal cancers (23).
The frequent coexpression pattern between MRP1 and ␥-GCSh mRNAs suggests that these genes may be coordinately regulated. However, the underlying mechanisms that regulate the expression of these genes are not known. In this study, we demonstrate that like ␥-GCSh, the expression of MRP1 could be induced by many pro-oxidants. Moreover, elevated levels of the physiological antioxidant GSH, conferred by ectopic expression of ␥-GCSh, down-regulated MRP1 and ␥-GCSh expression.
We also measured intracellular levels of reactive oxygen intermediates (ROIs) in those cells and demonstrated a potential link between ROI levels and MRP1 and ␥-GCSh expression. Our results are consistent with the idea that expression of MRP1 and ␥-GCSh is sensitive to the intracellular oxidationreduction (redox) status.
Cell Cultures-The rat hepatoma cell line H-4-II-E and the human hepatoma cell line HepG2 were purchased from the American Type Culture Collection Center (Rockville, MD). Human small cell lung cancer cell lines (SCLC and SR3A) were a generous gift from Dr. Niramol Savaraj (University of Miami, Miami, FL). SR3A was a doxorubicinresistant cell population established from SCLC as described previously (20). SR3A cells contained 3-and 1.5-fold higher MRP1 and ␥-GCSh mRNAs, respectively, compared with SCLC cells (22). The cells were maintained in Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% fetal calf serum (Life Technologies, Inc.), 1 mM glutamine, and 50 g/ml neomycin in a humidified incubator containing 5% CO 2 . SR3A and its transfectants were fed twice weekly with medium containing 0.17 mM doxorubicin.
Transfection of ␥-GCSh cDNA-The calcium phosphate precipitation method was used to transfect cells with DNA. pCMV/GCSh, which contains a full-length human ␥-GCSh cDNA under the transcriptional control of the cytomegalovirus promoter and the neomycin resistance marker for G418 selection, was used for transfection (24). G418-resistant cell lines, either from pooled colonies or from individual clones, were established. Each pool contained at least 20 independent colonies.
RNA Isolation and RNase Protection Assay-The procedures used for isolation of RNA, preparations of human MRP1 and ␥-GCSh probes, and RNase protection assays have been described previously (20,21). For the RNase protection assay of rat MRP1 mRNA, an antisense probe was synthesized from a recombinant plasmid DNA template containing the polymerase chain reaction product generated using the forward primer 5Ј-GCTGGGAAATCATCCCTCAC and the backward primer 5Ј-GGATCCTGTGGAATGATG derived from the two conserved regions in Sequences I and II of rat MRP1 cDNA (25), respectively. This probe produced a 140-nucleotide protected fragment in the RNase protection assay. For the rat ␥-GCSh probe, a 990-nucleotide fragment of cDNA was synthesized using the forward primer 5Ј-GGAGGAGGAGGGGG-CGG and the backward primer 5Ј-TCTTCAGGGGCTCCAGTCC. (These primer sequences were selected because they are conserved in the human and rat homologues.) The polymerase chain reaction product was subcloned into pSPT18 vector and sequenced. The plasmid was linearized by digestion at the internal PstI site, and the probe was synthesized with T7 polymerase to generate a 183-nucleotide fragment in the RNase protection assay.
Quantitative analyses of mRNA levels were carried out by densitometric scanning of the autoradiographs using an SI Personal Densitometer (Molecular Dynamics, Inc., Sunnyvale, CA). The autoradiographic signals corresponding to each mRNA species were converted into digitized images using computer software provided by the vendor. Statistical analyses were performed using the Statistica program (StatSoft, Tulsa, OK) and the Macintosh Excel software.
Measurement of ROIs by Flow Cytometry-DHR123 was prepared as a 5 mM stock in dimethyl sulfoxide (Me 2 SO) and used at a final concentration of 1 M. Cells (5 ϫ 10 5 ) were plated on a 65-mm plate for 24 h. Cells were treated with t-BHQ or BSO according to the specifications described, followed by DHR123. Cells were trypsinized into a single-cell suspension. R123 fluorescence intensity resulting from DHR123 oxidation was measured by a FACScan flow cytometer (Becton Dickinson, Mountain View, CA) with excitation at 488 nm and was detected between 515 and 550 mm according to the procedure described by Goossens et al. (26). Cell debris and dead cells (Ͻ5%) were calculated and eliminated with forward light scatter. Data analysis was performed using LYSYSII software (Becton Dickinson), which provided linearized values for the logarithmic fluorescent histograms.
Other Methods-Measurements of total cellular glutathione (GSH ϩ 2ϫGSSG) and transport of LTC 4 using membrane vesicles followed the methods described previously (20). Shi et al. (27) demonstrated that expression of ␥-GCSh in cultured rat cells could be induced by NQ and menadione. Mulcahy et al. (28) reported that exposing human hepatoma HepG2 cells to ␤-naphthoflavone resulted in an increase in ␥-GCSh mRNA levels. These results demonstrated that expression of ␥-GCSh could be up-regulated by pro-oxidants. Because we previously demonstrated coordinated expression of MRP1 and ␥-GCSh mRNAs in many experimental systems (20 -23), we examined whether the expression of MRP1 could also be induced by the pro-oxidants. Fig. 1A shows an RNase protection assay of MRP1 mRNA levels in rat hepatoma H-4-II-E cells treated with t-BHQ, NQ, and menadione for time periods from 6 to 48 h. In all cases, increased MRP1 mRNA levels could be seen 6 h after the treatment and declined 48 h after the treatment, except for those treated with NQ. The time course rise and fall of MRP1 mRNA varied among the different pro-oxidants used. These results demonstrate that expression of MRP1, like that of ␥-GCSh (27,28), is regulated by pro-oxidants. Co-induction of MRP1 and ␥-GCSh by t-BHQ was also observed in human HepG2 cells (Fig. 1B). Maximal levels of induction by these pro-oxidants ranged from 2.5-to 4.2-fold. These pro-oxidants exert oxidative stress on the cells by forming highly reactive, short half-life ROIs such as superoxide and hydrogen peroxide (29 -31).

Increased LTC 4 Transport Activities in Membrane Vesicles Prepared from t-BHQ-treated Cells-Previous studies have
demonstrated that membrane vesicles prepared from MRP1overexpressing cells exhibit increased ATP-dependent transport of LTC 4 (3,4). To investigate whether the pro-oxidanttreated cells exhibit enhancement of such activities, we prepared membrane vesicles from H-4-II-E cells treated with 100 M t-BHQ for 24 h and from those that were untreated. LTC 4 uptake activities were measured in an incubation mixture with or without ATP. Elevated ATP-dependent LTC 4 transport activity was observed (Fig. 2). Similar results were observed in H-4-II-E cells treated with 50 M NQ for 24 h (data not shown). These results demonstrate that the functional MRP1/GS-X pump was induced by these pro-oxidants.
Measurement of t-BHQ-induced ROI Formation by Flow Cytometry-A previous study demonstrated, using EPR spectroscopy, that the formation of ROIs in t-BHQ-treated cells is concentration-and time-dependent (26). Production of ROIs apparently leveled off within 1 h. In this study, we used a fluorescent probe, DHR123, to monitor the formation of ROIs. DHR123, an uncharged and nonfluorescent dye, passively diffuses across most cell membranes and reacts with ROIs, resulting in the formation of cationic fluorescent R123 that can be measured by flow cytometry. DHR123 has been used as a molecular probe for the measurement of ROI formation in cultured endothelial cells (32) and in murine fibrosarcoma cells treated with tumor necrosis factor (26).
H-4-II-E cells were treated with t-BHQ for different time intervals. Control cells were treated with vehicle (Me 2 SO) only. Two h prior to cell harvest, DHR123 was added. The accumulation of R123 in the untreated and t-BHQ-treated cells (due to basal oxidative metabolism) was measured by flow cytometry. Fig. 3 shows that the mean value of R123 fluorescence intensity in cells treated with t-BHQ for 2 h increased by 150% as compared with that in the untreated control. Increasing the t-BHQ treatment time to 4 h reduced the mean fluorescence intensity to a level that was ϳ35% greater than the control level. Prolonged t-BHQ treatments resulted in further reduc-tions of the mean fluorescence intensity. These results suggest that ROIs were produced in the initial phase of t-BHQ treatments. The reduced accumulation of R123 in the prolonged t-BHQ-treated cells may reflect the short half-life of the prooxidant or the elimination/sequestration of the produced R123 by a yet to be elucidated transporter. Although indirect evidence suggests that R123 may be a substrate for the MRP1/ GS-X pump, 2 the reduction of the R123 fluorescence in the 4-h treated cells was not likely due to the enhanced expression of MRP1 because an increase in MRP1 mRNA levels was not observed 6 h after the treatment (Fig. 1A).

Suppression of MRP1 and ␥-GCSh Expression by Elevated
Intracellular GSH Levels-To further substantiate the effects of oxidative stress on MRP1 and ␥-GCSh expression, we analyzed the expression of these genes in cells expressing elevated levels of the important physiological antioxidant GSH. GSH levels can be elevated by overexpressing ␥-GCSh (24,33). A recombinant plasmid encoding human ␥-GCSh was used to transfect cell line SCLC and its doxorubicin-resistant line, SR3A. This plasmid DNA also contains a neomycin-resistant marker for selection. Positive transfectants were established from the pooled colonies or from single-clone isolation. The use of pooled colonies in the analyses could minimize clonal variations among individual clones. Eight ␥-GCSh-positive SR3A cell lines have been obtained: five from pools of Neo R colonies (pools 1-5) and three from individual colonies (clones 13-15). One ␥-GCSh-transfected  (Table  I). RNase protection assays were performed to determine the steady-state levels of MRP1 and ␥-GCSh mRNAs (Fig. 4), and results from three independently prepared RNA samples were densitometrically analyzed. The transfected cell lines exhibited elevated ␥-GCSh mRNA levels, ranging from 1.8-to 10-fold (Table I), but only the values for cell lines 1, 4, 13, and 14 were statistically significant, despite the fact that all the cell lines displayed statistically significant increases in GSH levels (increases ranging from 1.93-to 4.97-fold). These results may reflect the differences in sensitivities of detecting methods: for the biochemical determination of GSH levels, only 1.9-fold increases were required to be statistically significant, whereas 3-fold increases in mRNA levels were required for the RNase protection assay. A similar explanation may also apply to the transfected SCLC cell line, which displayed ϳ2-fold increases in both ␥-GCSh mRNA and GSH levels. Northern hybridization was also used to determine MRP1 and ␥-GCSh mRNA levels in these cell lines. However, we could detect hybridization signals only in transfectants 1 and 14, the two most ␥-GCSh mRNA-abundant cell lines (data not shown). These results suggest that the RNase protection assay employed here was more sensitive than the Northern hybridization in the analyses of these mRNAs. Our results are consistent with previous findings showing that transfection of ␥-GCSh cDNA results in elevated intracellular GSH levels (24,33).
RNase protection assays also revealed that, although many transfected cell lines had reduced endogenous MRP1 mRNA levels (Table I) Table I). The inability to detect a significant reduction of MRP1 mRNA in other transfected cell lines may reflect the technical limitation of RNase protection. Alternatively, these results may suggest that ␥-GCSh mRNA levels must reach a sufficient threshold to exert suppressive effects on MRP1 expression. Whether elevated expression of ␥-GCSh down-regulates endogenous ␥-GCSh expression could not be concluded from these experiments because human ␥-GCSh cDNA was used to transfect the human cells. To address this issue, we performed similar experiments by transfecting human ␥-GCSh cDNA into rat hepatoma H-4-II-E cells. Two cell lines, designated H9 and H17, were obtained from individual colonies. Expression of the transfected human ␥-GCSh cDNA in the rat cells was measured by RNase protection assay using human antisense ␥-GCSh as a probe, and the results were compared with those obtained in human HepG2 cells. Levels of human ␥-GCSh mRNA in H9 cells were comparable to those in HepG2 cells, whereas H17 cells contained about one-third of the amount (Fig. 5B). Levels of endogenous MRP1 and ␥-GCSh mRNAs in the transfected cells were then determined using the rat probes that gave rise to protection fragments of 140 and 183 nucleo- tides, respectively (Fig. 5A, lanes 1-3). These two protected fragments were not detected in the same analysis when total RNA from human hepatoma HepG2 cells was used (Fig. 5A,  lane 4), although a strong protection signal corresponding to 165 nucleotides, possibly due to the cross-hybridization to the human ␥-GCSh mRNA, was present. These results suggest that these probes could adequately measure endogenous MRP1 and ␥-GCSh mRNA levels in the H-4-II-E cell line and its transfected cells.
H9 and H17 cells exhibited 1.48-and 1.2-fold increases, respectively, in GSH levels (Table I). RNase protection assays revealed that endogenous MRP1 and ␥-GCSh mRNA levels were reduced in these transfectants, but only the results for the H9 line were statistically significant (Fig. 5A and Table I). These results demonstrate that overexpression of ␥-GCSh down-regulated both MRP1 and ␥-GCSh expression. Furthermore, it appears that smaller GSH increases were sufficient to down-regulate MRP1 in the transfected rat H-4-II-E cells compared with the transfected SR3A cells (Table I).
Reduction of ROI Formation in GSH-overexpressing Cell Lines-The SR3A cell line and its ␥-GCSh-transfected cell lines 13-15 were treated with DHR123 or vehicle only for 6 h to measure the intracellular ROI levels. Cells were harvested, and the fluorescence intensities in the treated cells and in the vehicle-treated cells (measuring autofluorescence) were analyzed by flow cytometry. Results from duplicate experiments showed that clones 13-15 exhibited 19, 25, and 22% reduction, respectively, of the mean fluorescence intensities compared with untransfected cells (Fig. 6A). Similar analyses revealed that the ␥-GCSh-transfected H-4-II-E cell lines H9 and H17 exhibited 22 and 23% reduction, respectively, of the mean fluorescence intensities in comparison with the untransfected control (Fig. 6B). These results suggest that overexpression of ␥-GCSh is associated with reduced intracellular ROI levels.  To investigate whether overexpression of ␥-GCSh resulted in reduced MRP1/GS-X pump activities, we prepared membrane vesicles from SR3A and H-4-II-E cells and their corresponding ␥-GCSh-transfected cell lines, clones 14 and H9. Reduced ATPdependent LTC 4 transport activities were observed in the samples obtained from the transfected cells compared with those from the corresponding untreated cells (Fig. 7). These results demonstrate that overproduction of ␥-GCSh resulted in reduction of MRP1/GS-X pump function.
No Alteration of MRP1 and ␥-GCSh Expression in GSHdepleted Cells-Having established that increased GSH levels down-regulated MRP1 and ␥-GCSh expression, we then investigated whether depleting intracellular GSH levels could affect the expression of MRP1 and ␥-GCSh. We used BSO, which is a highly selective inhibitor of ␥-GCSh that binds to the active site of the enzyme. Administration of BSO to cultured cells and animals effectively turns off cellular GSH synthesis (34 -37). SCLC and SR3A cells were treated with various concentrations of BSO for 24 h. Intracellular GSH levels in the treated cells were determined. Treating these cells with 5 M BSO resulted in depleting 71 and 65% of GSH levels in SR3A and SCLC cells, respectively (Fig. 8A). Increasing BSO concentrations failed to produce further reduction of GSH levels in SCLC cells, although levels of GSH in SR3A cells may be reduced somewhat. These results are consistent with the previous findings suggesting a BSO-resistant intracellular GSH pool (34 -36). Results from three independent experiments showed that no alterations of MRP1 and ␥-GCSh mRNA levels were observed in these BSO-treated cells. A representative result is shown in Fig. 8B. We conclude that reducing GSH levels has no effect on the expression of MRP1 and ␥-GCSh in these cells. Concomitantly, flow cytometric measurements revealed only minor (Ͻ15%) increases in ROI levels in these cells treated with BSO ( Fig. 9). Such low levels of ROI increase may not exert sufficient stress to alter MRP1 and ␥-GCSh expression in these cells.

DISCUSSION
The roles of MRP1 in conferring multidrug resistance in cultured cells have been conclusively demonstrated by transfection experiments (39,40). The identification of the function of MRP1 as transporter of GSH-containing substrates underscores the important roles of GSH in MRP1-mediated drug transport. In this study, we present evidence showing that expression of MRP1, like that of ␥-GCSh, can be induced in cultured cells treated with the pro-oxidants t-BHQ, NQ, and menadione ( Fig. 1A) (27, 28). These results further substantiate the nature of the coordinate expression pattern between MRP1 and ␥-GCSh (20 -23). Moreover, these agents are known to introduce intracellular oxidation-reduction labile conditions by virtue of their capacities to undergo 1-and/or 2-electron valency changes, leading to an alteration of intracellular oxy radicals and generation of oxidative stress to the cells (29 -31). Our present results suggest that expression of MRP1 and ␥-GCSh is redox-sensitive.  Table I. nt, nucleotides.

FIG. 5. Measurement of endogenous MRP1 and ␥-GCSh mRNA levels in H-4-II-E cells transfected with human ␥-GCSh cDNA plasmids.
RNAs prepared from transfected cells (H9 and H17) and untransfected controls were subjected to RNase protection assays using the rat MRP1 and ␥-GCSh probes (denoted by mrp and ␥-gcsh, respectively) (A) and human ␥-GCSh probes (B). In both cases, RNA from human HepG2 cells was used as a reference. Note that the signals detected by the rat probes in the transfected cells are not present in the HepG2 lane (A). The dots in A represent signals of cross-hybridization to unknown RNA that could be used as a reference for equal loading of samples. nt, nucleotides. ␥-GCSh expression. GSH is a well known antioxidant that directly reacts with free radicals and serves as a cofactor for GSH peroxidase to reduce hydrogen peroxide, resulting in reduction of the oxidative cellular environment (38). Using the fluorescent probe DHR123, we demonstrated that reduced ROI levels indeed occurred in several independently established cell lines containing elevated GSH levels (Fig. 6). These results, together with the finding of no alteration of MRP1 and ␥-GCSh mRNA levels in the BSO-treated cells in which only marginal increases of ROIs were found, are consistent with the hypoth-esis that expression of MRP1 and ␥-GCSh is regulated by intracellular redox status. To the best of our knowledge, this is the first report describing that the redox regulatory mechanism is associated with the expression of the MRP1 gene.
The reasons why depleting Ͼ65% of intracellular GSH levels by BSO failed to induce appreciable increases in ROI and MRP1 and ␥-GCSh mRNA levels are unknown. GSH is among the most abundant thiol-containing small peptide in living organisms (1-10 mM). Such abundant levels may preserve a capacity that can afford substantial loss without tipping the balance of intracellular redox conditions. Previous studies (34 -36) and the present results (Fig. 8) suggest that there is a BSO-resistant intracellular GSH pool (34 -36). Although the nature of this residual GSH pool has not been fully elucidated, it has been suggested that this may represent a compartmentalized GSH pool, located primarily in mitochondria and nuclei (Fig. 8). The mitochondrial GSH pool involved in the regulation of intramitochondrial redox status has been demonstrated in many cell types (35,36). Although this mitochondrial GSH is sensitive to depletion by many antitumor agents, it is relatively insensitive to BSO depletion. Likewise, many reports have suggested that GSH pools in nuclei are more resistant than those in the cytoplasm to BSO depletion (42)(43)(44)(45)(46). However, since our knowledge is rather limited at this point, it is speculated that these compartmentalized GSH levels may play a dominant role in controlling intracellular ROIs and gene regulation compared with those in the cytoplasmic compartment. This hypothesis requires further experimental demonstration.
A previous study (34) demonstrated moderate levels (1.7-2.5-fold) of increases in ␥-GCSh transcript in BSO-treated HBT5 and HBT28 brain tumor cells under conditions in which 95% of intracellular GSH was depleted. These results differ from those presented in this study. Although the basis for this discrepancy is unclear, it may reflect the differential response of different cell lines to BSO. For example, treating HBT5 and HBT28 cells with 100 M BSO depleted 95% of the GSH content in these cells, whereas under the same conditions, only 85 and 65% depletions were seen in SR3A and SCLC cells, respectively. Significant heterogeneity in the effects of BSO on ␥-GCSh expression has indeed been reported in these HBT lines (34). Likewise, in BSO-treated newborn rats, although Ͼ90% reduction of GSH levels was found in many organs, including liver, kidney, heart, and skeletal muscle, no major changes in ␥-glutamylcysteine synthetase activities were seen in these organs, except kidney (38), supporting the differential effects of BSO on ␥-glutamylcysteine synthetase expression in animals.
The observation that under conditions where no appreciable alterations in ROI levels show no alterations in MRP1 and ␥-GCSh expression in BSO-treated SCLC and SR3A cells further strengthens the role of intracellular redox conditions in MRP1 and ␥-GCSh regulation. These results are also consistent with the hypothesis proposed by Meister and co-workers (38). These investigators postulated that since BSO does not react with GSH and since there is no evidence that the sulfoximine moiety itself exerts toxicity (38), the BSO-induced GSH deficiency-related stress, if any, may represent that in normal physiological metabolism and may differ significantly from that produced by many known cytotoxic compounds.
The mechanism(s) underlying redox-mediated regulation of MRP1 and ␥-GCSh expression are unknown at present. An oxidative stress-responsive element (ORE) located distal to the ␥-GCSh promoter has recently been identified (28). Several ORE-like sequences located upstream from the promoter of MRP1 have been noted (22,47); however, whether these sites can function as authentic OREs remains to be demonstrated. It is possible that these putative cis-acting elements, by interacting with redox-sensitive transcription factors, may confer oxidative stress-induced MRP1 and ␥-GCSh expression. This model would be consistent with the redox-regulated mechanisms of gene expression demonstrated in Escherichia coli (48). Because nucleotide sequences in many OREs share striking similarities with those in AP-1-binding sites, it has been proposed that AP-1 transcription factors may be involved in the regulation of oxidative stress-induced gene expression (29). However, there are also reports suggesting that, in the physiological context, the major ORE-binding and -activating protein is not AP-1 (49, 50). Elucidation of transduction signaling in response to oxidative stress leading to gene activation is necessary.
We previously reported that expression of MRP1 and ␥-GCSh can be transiently induced by many cytotoxic agents, including antitumor agents (cisplatin and alkylating agents) and heavy metals (20,21). In many cases, co-induction of MRP1 and ␥-GCSh was observed. Our present results show that overexpression of ␥-GCSh mRNA suppresses the expression of MRP1, indicating that increasing ␥-GCSh mRNA levels per se cannot up-regulate MRP1 expression. Thus, the coexpression of MRP1 and ␥-GCSh induced by these agents is most likely due to the induced cytotoxic stress, rather than cross-talk between these two genes.
The present results, as well as those described previously (21), suggest that a dynamic GSH homeostasis may be associated with MRP1 and ␥-GCSh expression. The stress-induced MRP1 and ␥-GCSh expression by cytotoxic agents may transiently enhance GSH levels and GS-X pump activity (20,21). Many cytotoxic agents that are known to transiently induce MRP1 and ␥-GCSh expression can potentially be conjugated by GSH (20,21), and the resulting glutathione S-conjugates may be extruded by MRP1-mediated transport, resulting in the consumption of intracellular GSH. Likewise, abrogation of MRP1 expression in MRP1 knockout mice is associated with accumulation of GSH (51). Elevated GSH levels have been previously demonstrated to feedback-suppress ␥-glutamylcysteine synthetase enzymatic activity (17,41). The present study, showing that elevated GSH levels also down-regulate ␥-GCSh expression, reveals a previously unreported feedback inhibitory mechanism of gene expression. Furthermore, because GSH levels are involved in many important cellular signaling pathways (52)(53)(54), the present findings may have a broad implication beyond the presently described MRP1 and ␥-GCSh expression. Finally, our present study may have clinical relevance when modulators of MRP1 and GSH function are considered in combination cancer chemotherapy (10).