Oxidative and electrophilic stress induces multidrug resistance–associated protein transporters via the nuclear factor-E2–related factor-2 transcriptional pathway


  • Potential conflict of interest: Nothing to report.


Multidrug resistance–associated proteins (Mrps) are adenosine triphosphate–dependent transporters that efflux chemicals out of cells. In the liver, Mrp2 transports bilirubin-glucuronide, glutathione (GSH), and drug conjugates into bile, whereas Mrp3 and Mrp4 efflux these entities into blood. The purpose of this study was to determine whether oxidative conditions (that is, the disruption of hepatic GSH synthesis) or the administration of nuclear factor-E2–related factor-2 (Nrf2) activators (oltipraz and butylated hydroxyanisole) can induce hepatic Mrp transporters and whether that induction is through the Nrf2 transcriptional pathway. Livers from hepatocyte-specific glutamate-cysteine ligase catalytic subunit–null mice had increased nuclear Nrf2 levels, marked gene and protein induction of the Nrf2 target gene NAD(P)H:quinone oxidoreductase 1, as well as Mrp2, Mrp3, and Mrp4 expression. The treatment of wild-type and Nrf2-null mice with oltipraz and butylated hydroxyanisole demonstrated that the induction of Mrp2, Mrp3, and Mrp4 is Nrf2-dependent. In Hepa1c1c7 cells treated with the Nrf2 activator tert-butyl hydroquinone, chromatin immunoprecipitation with Nrf2 antibodies revealed the binding of Nrf2 to antioxidant response elements in the promoter regions of mouse Mrp2 [−185 base pairs (bp)], Mrp3 (−9919 bp), and Mrp4 (−3767 bp). Conclusion: The activation of the Nrf2 regulatory pathway stimulates the coordinated induction of hepatic Mrps. (HEPATOLOGY 2007.)

Multidrug resistance–associated proteins (Mrps) are adenosine triphosphate (ATP)–dependent transporters involved in the efflux of a vast number of endogenous and exogenous chemicals. Mrps play a key role in cellular protection by removing xenobiotics, metabolites, and endogenous substrates that can accumulate in tissues and lead to toxicity. In the liver, Mrp2 is a canalicular transporter involved in the biliary excretion of many endogenous and exogenous metabolites, including leukotrienes and bilirubin, and many drugs, such as protease inhibitors.1–3 Mrp3 and Mrp4 are high-capacity, low-affinity basolateral transporters that are highly inducible. These hepatoprotective transporters export conjugated bile acids, bilirubin, and other substrates from hepatocytes back into the bloodstream for renal excretion.4–6

Mrps might be considered a double-edged sword; although their constitutive expression is important in normal cellular homeostasis, their expression can be increased in some tumors, and this can allow the tumor to exploit the protective properties of Mrps. The concept of multidrug resistance is actually the combination of several protective strategies that can occur in a tumor cell, including increased efflux of chemotherapeutics due to the induction of transporters7, 8 and increases in cytoprotective and antioxidative enzymes to protect against chemotherapy-induced oxidative stress.9 For example, tumor cells often have increased levels of glutathione (GSH) synthesis because of high expression of glutamate-cysteine ligase (Gcl) and in parallel have increases in the efflux of chemotherapeutics because of the induction of transporters. The role of Mrps in transporting GSH-conjugated chemotherapeutics out of cells10–12 and in maintaining GSH homeostasis has been established13–15 and includes the cotransport of GSH with Mrp substrates such as bile acids.16 Such an example of coordinated protection can be observed with leukemic cells, which have induced expression of the glutamate-cysteine ligase catalytic subunit (Gclc) and Mrp1, which correlates with marked increases in GSH levels.17 Such multifaceted protection could be due to concerted regulation via a single mechanism.

GSH, which is present in millimolar concentrations, is one of the liver's main defenses against oxidative stress because of the conjugating, scavenging, and reducing capabilities of this molecule.18 GSH synthesis involves 2 major steps: the first and rate-limiting step is the formation of γ-glutamylcysteine, and it is catalyzed by Gcl. The Gcl holoenzyme is composed of catalytic (Gclc) and modifier [glutamate-cysteine ligase modifier unit (Gclm)] subunits, and the global ablation of Gclc by conventional gene targeting techniques in mice is incompatible with life.19 However, hepatocyte-specific Gclc-null mice can survive for weeks, but they eventually die because of hepatic failure.20 In Gclc-null mice, the progressive depletion of GSH was observed: 37% of wild-type (WT) levels by postnatal day 14, 7% by postnatal day 21, and maximal deletion by postnatal day 28, with only 4.5% GSH remaining. The deletion of the Gclc gene is progressive with age and follows the expression of the albumin-cre recombinase (ALB-CRE) transgene, thus impacting GSH levels over time. Gclc messenger RNA (mRNA) and protein are very low by postnatal day 14, and recombination at the Gclc locus is complete by postnatal day 21. Oxidant stress–responsive genes, including heme oxygenase 1 (Ho-1) and metallothionein 1, are elevated by postnatal day 14, at which time the GSH level in the whole liver is 37% of the WT counterparts. In this model, it is important to note that (1) the recombination of Gclc is restricted to hepatocytes, (2) not all hepatocytes lose Gclc simultaneously, and (3) nonparenchymal cells in the liver retain Gclc expression. Because GSH is essential to cellular defense against oxidative stress, hepatocyte-specific deletion of Gclc seemingly offers an excellent model for the study of oxidative stress–related mechanisms.

One transcription factor that serves as a cellular sensor for oxidative stress is nuclear factor E2–related factor-2 (Nrf2). Nrf2 belongs to the cap'n'collar family of basic leucine zipper proteins, such as small musculo-aponeurotic fibrosarcoma (Maf) proteins, c-fos, c-jun, BTB and CNC homology 1 basic leucine zipper transcription factor 1 (Bach1), and nuclear factor-E2–related factor-1.21 Nrf2 is sequestered in the cytosol by kelch-like ECH-associated protein 1 (Keap1). During oxidative challenge, the modification of Keap1 sulfhydryl groups results in the release and nuclear translocation of Nrf2.22 Nrf2 can transcriptionally activate several enzymes involved in cellular protection, including NAD(P)H:quinone oxidoreductase 1 (Nqo1), Ho-1, Gclc, and Gclm,23 as well as phase II conjugation enzymes, such as glutathione S-transferase (GST) A2 and uridine 5′-diphospho-glucuronosyltransferase (UGT) 1A6.23, 24 The activation of this gene battery serves to decrease the oxidative burden of the cell by increasing GSH concentrations, reducing reactive intermediates such as quinones, and increasing general detoxification of chemicals via phase II conjugation reactions.

Because oxidative insult to hepatocytes could be expected to induce efflux pathways along with phase II detoxification enzymes, and because many phase II pathways are inducible through the action of the transcription factor Nrf2, we hypothesized that Mrp2, Mrp3, and Mrp4 are inducible by Nrf2. Previous work from this laboratory25 has suggested that Nrf2 activation can lead to Mrp up-regulation, and there is further evidence in vitro that Mrp2 may be regulated by Nrf2.26 The purpose of this study was to determine whether Mrp2, Mrp3, and Mrp4 are induced (1) in the absence of GSH or (2) after a treatment with Nrf2-activating chemicals in 2 mouse models of Nrf2 activation and (3) whether Mrp induction under these conditions is mediated via the Nrf2-transcriptional pathway. Therefore, the induction of Mrps was compared in WT and Nrf2-null mice, and an analysis for potential Nrf2-responsive elements in Mrp2, Mrp3, and Mrp4 5′-flanking regions was conducted.


ARE, antioxidant response element; ATP, adenosine triphosphate; Bach1, BTB and CNC homology 1 basic leucine zipper transcription factor 1; BHA, butylated hydroxyanisole; bp, base pairs; BSO, buthionine sulfoximine; CAT, catechol; CDNB, 1-chloro-2,4-dinitrobenzene; ChIP, chromatin immunoprecipitation; DEM, diethyl maleate; DMSO, dimethyl sulfoxide; EXQ, ethoxyquin; Gcl, glutamate-cysteine ligase; Gclc, glutamate-cysteine ligase catalytic subunit; Gclm, glutamate-cysteine ligase modifier unit; GSH, glutathione; GST, glutathione S-transferase; Hepa-1, Hepa1c1c7; Ho-1, heme oxygenase 1; IgG, immunoglobulin G; kb, kilobases; Keap1, kelch-like ECH-associated protein 1; LDH, lactate dehydrogenase; Maf, musculo-aponeurotic fibrosarcoma; mRNA, messenger RNA; Mrp, multidrug resistance–associated protein; Nqo1, NAD(P)H:quinone oxidoreductase 1; Nrf2, nuclear factor-E2–related factor-2; NT, no template; OPZ, oltipraz; PBS-T, phosphate-buffered saline with 0.2% Triton X-100; PBS-Tx, phosphate-buffered saline with 0.1% Triton X; PCR, polymerase chain reaction; RLU, relative light units; SULF, sulforaphane; tBHQ, tert-butyl hydroquinone; WT, wild-type.

Materials and Methods


Oltipraz (OPZ) was generously provided by Dr. Steven Safe (Texas A&M University), and the remaining chemicals were purchased from Sigma-Aldrich (St Louis, Mo). The Nrf2 antibody (H-300, sc-13032) was purchased from Santa Cruz Biotechnologies (Santa Cruz, CA). Nqo1 (ab2346) and beta-actin (ab8227) antibodies were purchased from Abcam (Cambridge, MA). The Gclc antibody was obtained from Terry Kavanagh (University of Washington). The Mrp2 antibody for the immunofluorescence analysis was provided by Bruno Steiger (University Hospital, Zurich, Switzerland). Mrp2, Mrp3, and Mrp4 antibodies (M2III-5, M3II-2, and M4I-10) for western blot analysis were provided by George Scheffer (Vrije Universiteit Medical Center, Amsterdam, The Netherlands).


Pair-matched littermate WT and Gcl hepatocyte-specific Gclc-null mice on a C57BL/6J and 129S6/SvJ background were engineered and bred at the University of Cincinnati. All studies with hepatocyte-specific Gclc-null mice were approved by the University of Cincinnati Medical Center Institutional Animal Care and Use Committee. The mice were between 14 and 28 days of age because of mortality beyond 1 month. WT and Nrf2-null mice27 on a C57BL/6 background were bred and housed in an American Animal Associations Laboratory Animal Care–accredited facility at the University of Kansas Medical Center, and all procedures were preapproved in accordance with Institutional Animal Care and Use Committee guidelines. WT and Nrf2-null mice were 8-10 weeks old during the experiments. Mice were allowed water and rodent chow ad libitum (Teklad; Harlan, Indianapolis, IN). All mice were maintained on 12-hour dark/light cycles.


WT and Nrf2-null mice were gavaged once daily with 150 mg/kg OPZ or injected with 350 mg/kg butylated hydroxyanisole (BHA) intraperitoneally for 4 consecutive days. Twenty-four hours after the last dose, the livers were removed, snap-frozen in liquid nitrogen, and stored at −80°C.

Cell Culture and Treatments.

Mouse Hepa1c1c7 (Hepa-1) cells (American Type Culture Collection, Manassas, VA) were cultured in Dulbecco's modified Eagle's medium (Invitrogen, Carlsbad, CA) with 5% charcoal-stripped fetal bovine serum (Hyclone, Logan, UT) in a humidified incubator at 37°C and maintained there at 5% CO2. Cells were cultured in 100-mm dishes and allowed to grow to 90%–95% confluence. Three dishes per group were treated with dimethyl sulfoxide (DMSO) or the following Nrf2 activators: 50 μM OPZ, 50 μM ethoxyquin (EXQ), 3 μM sulforaphane (SULF), 75 μM tert-butyl hydroquinone (tBHQ), 50 μM catechol (CAT), and 16 μM 1-chloro-2,4-dinitrobenzene (CDNB). Three dishes per treatment group were scraped, pooled into a single sample, and pelleted by centrifugation. Pellets were snap-frozen in liquid nitrogen and stored at −80°C until RNA extraction. The reported data represent the most effective concentration of each chemical, with the concentration derived from concentration-response studies spanning a concentration range of at least 3 orders of magnitude. Each chemical treatment was replicated 3 times, and the data represent the averages of those 3 experiments.

Branched DNA Assay.

Nqo1, Mrp2, Mrp3, and Mrp4 oligonucleotide probes28 were diluted in a tris-hydroxymethylaminomethane–ethylene diamine tetraacetic acid buffer (pH 8.0) according to instructions provided with the Quantigene branched DNA signal amplification kit (Panomics, Inc., Fremont, CA). Total RNA (1 μg/μL; 10μL) was added to each well of a 96-well plate containing 50 μL of capture hybridization buffer and 50 μL of each diluted probe set. The total RNA was allowed to hybridize overnight at 53°C in a hybridization oven. Subsequent hybridization steps were carried out according to the manufacturer's protocol, and luminescence was measured with a luminometer.

Western Blotting.

Liver cytosol and plasma membrane preparations were performed as described previously.29 Nuclear extracts were prepared with the NE-PER nuclear extraction kit according to the manufacturer's directions (Pierce Biotechnology, Rockford, IL). Protein concentrations were determined with Bio-Rad protein assay reagents (Bio-Rad Laboratories, Hercules, CA). Liver nuclear (for Nrf2 expression), cytosol (for Nqo1 and Gclc expression), and membrane preparations (for Mrp2, Mrp3, and Mrp4 expression) were loaded and separated on 8% (Mrp2, Mrp3, and Mrp4) or 10% (Nqo1, Gclc, and Nrf2) sodium dodecyl sulfate–polyacrylamide gels. Proteins were transferred overnight at 4°C to poly(vinylidene difluoride) membranes. Poly(vinylidene difluoride) membranes were blocked for 2 hours in a blocking buffer (1% nonfat dry milk with 0.5% Tween 20). All primary and secondary antibodies were diluted in the blocking buffer. The primary antibody dilutions were as follows: Nrf2, 1:1000; Gclc, 1:20,000; Mrp2, 1:600; Mrp3, 1:2000; and Mrp4, 1:2000. The blots were subsequently incubated with a species-appropriate horseradish peroxidase–conjugated secondary antibody for 1 hour. The membranes were stripped and reprobed with a dilution of 1:2500 anti–beta-actin rabbit polyclonal antibody to confirm an equal protein loading. Protein-antibody complexes were detected with an enhanced chemiluminescence kit (Amersham Life Science, Arlington Heights, IL) and exposed to Fuji Medical X-ray film (Fisher Scientific, Springfield, NJ). The intensity of the protein bands was quantified with Quantity One software (Bio-Rad Laboratories).

Indirect Immunofluorescence Analysis.

Staining for Mrp2, Mrp3, and Mrp4 was performed as previously described.30 For all liver analyses, the sections were 5 μm thick. All sections were both stained and imaged under uniform conditions for each antibody. Two to three tissues were analyzed and found to be similar for each treatment group, and 1 representative image is presented. Negative controls without antibody were also included in the analysis (data not shown). Briefly, liver cryosections (5 μm) were blocked with 5% serum/phosphate-buffered saline with 0.1% Triton X (PBS-Tx) for 1 hour and then incubated with a primary antibody diluted 1:100 in 5% serum/PBS-Tx for 2 hours at room temperature. The sections were subsequently washed 3 times in PBS-Tx and incubated for 1 hour with goat anti-rat Alexa 488 immunoglobulin G (IgG; Invitrogen) for Mrp3 and Mrp4 detection and with fluorescein isothiocyanate–labeled goat anti-rabbit IgG (Jackson Immunoresearch Laboratories, West Grove, PA) for Mrp2 detection. Secondary antibodies were diluted 1:200 in 5% goat serum/PBS-Tx.

Nrf2 was detected as previously described.29 Briefly, cryosections were air-dried at room temperature for 5 minutes and fixed with 4% paraformaldehyde. The sections were blocked at room temperature for 30 minutes with 5% donkey serum/phosphate-buffered saline with 0.2% Triton X-100 (PBS-T) and then incubated overnight with anti-Nrf2 (H-300; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) diluted 1:50 in 5% donkey serum/PBS-T. A fluorescein isothiocyanate–labeled secondary antibody (Jackson Immunoresearch Laboratories, West Grove, PA), diluted 1:200, was used, along with a rhodamine-labeled phalloidin antibody (Invitrogen) diluted 1:200 in 5% donkey serum/PBS-T. 4′-6-Diamidino-2-phenylindole staining was performed to define nuclear regions.

Promoter Analysis.

To identify potential response elements in the 5′-flanking regions of Mrp2, Mrp3, and Mrp4, the European Molecular Biology Laboratory mouse genome database (www.ensembl.org) was used to access Mrp sequences. DS Gene V1.5 software (Accelrys, San Diego, CA) was used to analyze these genomic sequences for putative response elements (RTGACNNNGC), the core antioxidant response elements (AREs) for Nrf2 binding.

Chromatin Immunoprecipitation (ChIP) Assays.

The mouse Hepa-1 cell line was treated with DMSO (vehicle) or 75 μM tBHQ and assayed for Nrf2 binding to AREs. ChIP assays were performed as previously described.31 Immunoprecipitations were performed with either rabbit IgG or anti-Nrf2 antibodies (Santa Cruz Biotechnology). A 1:20 dilution of the nonimmunoprecipitated DNA (input) was analyzed along with IgG or anti-Nrf2 precipitated DNA. A polymerase chain reaction (PCR) was performed with 30 cycles, and Nqo1 and lactose dehydrogenase genes were analyzed as positive and negative controls, respectively;31 the primers for ARE amplification are listed in Table 1.

Table 1. Primers for Chromatin Precipitation Assays

Statistical Analysis.

The error bars represent the standard errors of the mean. The data were analyzed with a 2-tailed Student t test. Asterisks represent statistical differences (P ≤ 0.05) in mRNA and protein levels between control and treated mice. When multiple analyses were performed, a one-way analysis of variance test was used, followed by Duncan's post hoc test.


Increased Nuclear Nrf2 Protein in Response to the Loss of Gclc.

The absence of the Gclc protein in 21-day-old Gclc-null mice was confirmed by western blotting (Fig. 1). Livers from Gclc-null mice demonstrated 2.5-fold increased nuclear Nrf2 protein in comparison with WT mice. Double immunofluorescence staining was performed on liver sections from 21-day-old WT and Gclc-null mice to determine the subcellular localization of Nrf2 (Fig. 1). In the WT mice, Nrf2 staining (green) was minimal and diffusely localized to the periphery of hepatocytes, with some nuclear expression observed. The actin cytoskeleton was stained in red. Nrf2 staining in liver sections from Gclc-null mice was strong and localized to the nucleus (blue). Increased nuclear Nrf2 protein, as determined by western blotting and immunofluorescence, indicated the translocation of Nrf2 after Gclc ablation.

Figure 1.

Western blot and immunofluorescent staining of Nrf2 proteins in livers from WT and Gclc-null mice. The top part shows western blots from cytosol (Gclc) and nuclear extracts (Nrf2) from livers of WT and Gclc-null mice. The bottom part shows immunofluorescent staining of Nrf2 in WT and Gclc-null mice. Double-labeling indirect immunofluorescence against Nrf2 (green) and actin (red) was performed on liver cryosections. Nuclei are stained blue. The bar represents 50 μm.

Increased Mrp mRNA and Protein Expression in Response to the Loss of Gclc.

The loss of Gclc in the liver is associated with large-scale changes in the oxidative conditions in the hepatic environment. In response to such stresses, mRNA expression of the marker gene for Nrf2 activation, Nqo1, and Mrp2, Mrp3, and Mrp4 was quantified in WT and Gclc hepatocyte-specific null mice at days 14, 21, 25, and 28 after birth (Fig. 2). Approximately an 8-fold increase in the mRNA expression of Nqo1 in Gclc-null mice was observed in all age groups. The maximal induction of Mrp2, Mrp3, and Mrp4 mRNA was approximately 3-fold, 2-fold, and 9-fold higher, respectively, than that of WT mice. Corresponding increases in Nqo1 and Mrp2, Mrp3, and Mrp4 proteins were observed in Gclc-null mice at 21 days of age as determined by western blotting (Fig. 2). The Nqo1 protein was 4.5-fold higher in Gclc-null mice in comparison with their WT counterparts. Similarly, the levels of Mrp2, Mrp3, and Mrp4 proteins were increased between 4-fold and 5-fold in Gclc-null mice.

Figure 2.

mRNA and protein expression of Nqo1, Mrp2, Mrp3, and Mrp4 from representative assays of WT and Gclc-null mice livers (n = 2 per group). The left part of the figure shows the mRNA expression of Nqo1, Mrp2, Mrp3, and Mrp4 in WT and Gclc-null mice from 14-28 days after birth. The data are presented as the mean relative light units ± the standard error of the mean. Asterisks represent a statistically significant difference between WT and Gclc mice (P ≤ 0.05). The top right shows western blotting from cytosol (Nqo1) or membrane preparations (Mrp2, Mrp3, and Mrp4) from livers of WT and Gclc-null mice. The bottom right shows quantitation of the optical density from the corresponding western blots, normalized to beta-actin.

The induction of Mrp2, Mrp3, and Mrp4 proteins was confirmed by immunofluorescent staining (Fig. 3). Mrp2 is highly expressed on the apical membrane of mouse hepatocytes. Liver sections from Gclc-null mice demonstrated enhanced canalicular Mrp2 staining throughout the liver lobule in comparison with WT mice. Mouse Mrp3 and Mrp4 proteins are typically localized to the basolateral membrane of hepatocytes, as evidenced by a distinct honeycomb pattern. Little Mrp3 or Mrp4 staining was observed in liver sections from 21-day-old WT mice. It is important to note that Mrp3 mRNA in the mouse liver does not increase to adult levels until 30 days of age.28 Increased staining of Mrp3 in Gclc-null mice was prominent on the membranes of hepatocytes throughout the liver, with greater intensity in periportal cells. In contrast, strong Mrp4 staining in Gclc-null mice was preferentially observed in centrilobular hepatocytes.

Figure 3.

Immunofluorescent staining of Mrp2, Mrp3, and Mrp4 proteins in livers from WT and Gclc-null mice. Indirect immunofluorescence against Mrp2, Mrp3, and Mrp4 (green) was conducted on liver cryosections. Representative regions are shown. The bar represents 75 μm.

Increased Nuclear Nrf2 Protein in Response to OPZ Treatment.

Similarly to Fig. 1, double immunofluorescence staining was performed on liver sections from WT and Nrf2-null mice to determine the effects of OPZ on Nrf2 localization (Fig. 4). In vehicle-treated WT mice, Nrf2 staining (green) was diffusely localized to the periphery of hepatocytes, with little nuclear expression observed. The actin cytoskeleton was stained in red. Nrf2 staining in liver sections from OPZ-treated WT mice was localized to the nucleus (blue; Fig. 4). No Nrf2 staining was detected in vehicle or OPZ-treated Nrf2-null mice.

Figure 4.

Immunofluorescent staining of Nrf2 proteins after a treatment with OPZ in WT and Nrf2-null mice. Double indirect immunofluorescence against Nrf2 (green) and actin (red) was performed on liver cryosections. Nuclei are stained blue. The bar represents 50 μm.

Hepatic Mrps Induced in an Nrf2-Specific Manner in Response to OPZ and BHA Administration.

After 4 days of the administration of either OPZ or BHA to WT and Nrf2-null mice, the mRNA expression of Nqo1, Mrp2, Mrp3, and Mrp4 was quantified (Fig. 5). OPZ and BHA induced Nqo1 more than 3-fold and 6-fold, respectively, in WT mice, but this induction was almost completely attenuated in Nrf2-null mice; this indicated that the treatments were inducing via the Nrf2 pathway, similar to previous observations.25

Figure 5.

mRNA expression of Nqo1, Mrp2, Mrp3, and Mrp4 in WT and Nrf2-null mice after exposure to OPZ and BHA. The data are presented as the mean relative light units (RLU) ± the standard error of the mean (left) and fold over control (right). Asterisks represent a statistically significant difference between treated and control mice (P < 0.05). Daggers represent statistically significant repression in Nrf2-null mice in comparison with matching treatments in WT mice.

OPZ and BHA produced modest increases in Mrp2 mRNA expression, and although the expression tended to be attenuated in Nrf2-null mice, the decrease was not statistically significant (Fig. 5). However, when the mRNA levels were expressed as fold over control, statistically significant inhibition of Mrp2 induction by OPZ and BHA was observed in Nrf2-null mice. The basal expression of the Mrp3 mRNA was approximately 33% lower in Nrf2-null mice than WT mice. When OPZ and BHA were administered to WT mice, a 2-fold induction of Mrp3 was observed. In Nrf2-null mice, the induction of Mrp3 by OPZ and BHA was reduced but not ablated. However, when the data were expressed as fold over control, because of differences in the Mrp3 basal expression, the induction of Mrp3 with OPZ remained similar between WT and Nrf2-null mice. After BHA administration, Mrp3 induction was significantly, but not completely, attenuated in Nrf2-null mice. Thus, Nrf2 clearly determines the basal regulation of Mrp3 and at least some aspects of Mrp3-inducible expression. The constitutive expression of Mrp4 was low, but no differences in the basal levels were observed between WT and Nrf2-null mice. The OPZ and BHA treatment induced Mrp4 approximately 3-fold, and this induction was abolished in Nrf2-null mice.

Because the induction of Mrp2, Mrp3, and Mrp4 mRNA following OPZ and BHA was generally similar, protein levels were quantified only in OPZ-treated WT and Nrf2-null mice, with matching controls, and normalized to beta actin (Fig. 6). The basal expression of Nqo1 was markedly lower in Nrf2-null mice than WT mice. Furthermore, OPZ administration increased Nqo1 protein levels more than 2-fold, and this induction was virtually abolished in Nrf2-null mice. The Mrp2 protein was induced approximately 3-fold by OPZ, which was blocked in Nrf2-null mice (Fig. 6). Similar to the mRNA data, the expression of the Mrp3 protein in Nrf2-null mice was less than half that observed in WT mice. Similarly, a 3-fold induction of Mrp3 was observed after OPZ administration, which was markedly attenuated in Nrf2-null mice. Little expression of Mrp4 was observed in the control liver, but expression was induced more than 5-fold after OPZ administration yet markedly reduced in Nrf2-null mice.

Figure 6.

Western blot of cytosolic Nqo1 along with membrane Mrp2, Mrp3, and Mrp4 expression in WT and Nrf2-null mice. The bottom figures represent the quantification of the optical density of the blots above, normalized to beta-actin controls. The data are presented as the mean relative protein expression ± the standard error of the mean. Asterisks represent a statistically significant difference between treated and control mice (P < 0.05). Daggers represent statistically significant repression in Nrf2-null mice in comparison with matching treatments in WT mice.

Immunofluorescent Staining of Hepatic Mrps in Response to OPZ Administration.

Mrp2, Mrp3, and Mrp4 immunofluorescence staining was performed on frozen liver sections from WT and Nrf2-null mice treated with either OPZ or vehicle (Fig. 7). In comparison with the vehicle-treated counterparts, the OPZ treatment increased Mrp2 staining in livers from WT and Nrf2-null mice, although enhanced staining was greater in WT liver cryosections.

Figure 7.

Immunofluorescent staining of Mrp2, Mrp3, and Mrp4 proteins after a treatment with OPZ in WT and Nrf2-null mice. Indirect immunofluorescence against Mrp2, Mrp3, and Mrp4 (green) was conducted on liver cryosections. Representative centrilobular regions are shown. The bar represents 75 μm.

Mrp3 staining in vehicle-treated WT livers was greatest in centrilobular hepatocytes, with a gradual decrease in the intensity toward the portal areas (Fig. 7). Compared with that of vehicle-treated WT mice, the Mrp3 staining was markedly reduced in vehicle-treated Nrf2-null liver sections, and this is indicative of lower basal expression in null mice. Enhanced Mrp3 staining was observed in OPZ-treated WT livers, but not in Nrf2-null livers.

Similar to western blot analysis, minimal immunofluorescent staining of Mrp4 was seen in the livers of vehicle-treated WT and Nrf2-null mice (Fig. 7). Strong basolateral Mrp4 staining was observed in WT livers after OPZ administration. The labeling of Mrp4 consisted of a single or double layer of cells surrounding the central vein. Minimal Mrp4 staining was detected in OPZ-treated Nrf2-null livers.

In Vitro Induction of Mrps in Mouse Cell Lines.

To determine whether Mrps are responsive to a broad range of chemicals known to activate Nrf2, Mrp induction was examined in Hepa-1. Dose responses with 6 compounds known to activate Nrf2, namely, OPZ, EXQ, SULF, tBHQ, CAT, and CDNB, were obtained from a mouse Hepa-1 cell line with chemical concentrations over 3 orders of magnitude. The data presented in Fig. 8 illustrate the results with the optimal concentration of each of the 6 chemicals for the maximal level of mRNA induction. Mrp2 was induced by 3 of the 6 chemicals, OPZ, EXQ, and tBHQ, and all 6 chemicals increased Nqo1, Mrp3, and Mrp4 expression between 1-fold and 4-fold.

Figure 8.

Expression of Nqo1, Mrp2, Mrp3, and Mrp4 mRNA in Hepa-1 cells after exposure to Nrf2-activating chemicals. The concentrations were 50 μM OPZ, 50 μM EXQ, 3 μM SULF, 75 μM tBHQ, 50 μM CAT, and 16 μM CDNB. The data are presented as mean relative light units (RLU) ± the standard error of the mean. Asterisks represent a statistically significant difference between DMSO and chemical treatment groups (P ≤ 0.05).

Analysis of Mrp2, Mrp3, and Mrp4 Promoters for Nrf2-Responsive Elements.

Multiple putative elements with antioxidant cores exist in the 5′-flanking regions of Mrp2, Mrp3, and Mrp4 genes (Fig. 9). For Mrp2, 2 AREs exist within 2 kilobases (kb) of the translational start site; the first ARE core is −185 base pairs (bp), and the second is −1478 bp upstream. For Mrp3, putative elements are located much farther upstream than for Mrp2, with an ARE at −9919 bp and an ARE-like sequence at −10,588 bp. For Mrp4, 4 putative elements exist within the first 11 kb of the 5′-promoter; the most proximal putative ARE can be found at −3295 bp, with an everted tandem of an ARE and ARE-like sequence located within 4 bp of each other at −3753 and −3767 bp. Two more distal AREs can be found at −9104 and −10,462 bp as well.

Figure 9.

Multiple putative AREs exist in the 5′-flanking sequences of the Mrp2, Mrp3, and Mrp4 genes.

Analysis of Nrf2 Binding to AREs.

Using nonimmunoprecipitated DNA (input), IgG precipitated DNA (IgG), or anti-Nrf2 antibody precipitated DNA (Nrf2) from Hepa-1 cells treated with either vehicle or tBHQ, PCR was conducted to determine whether Nrf2 binds AREs in the reporter-enhancer regions in the Nqo1, lactate dehydrogenase (LDH), Mrp2, Mrp3, and Mrp4 genes. Nqo1 (positive control) has a known ARE in the 5′-flanking sequence that can bind Nrf2, whereas LDH (negative control) does not contain an ARE in the promoter sequence. As expected, Nrf2 binding was detected in the Nqo1 promoter, weakly in the control-treated lane and much more strongly after the tBHQ treatment (Fig. 10), whereas Nrf2 binding was not observed with the LDH promoter fragment; this suggested Nrf2-specific immunoprecipitation of ARE regions. For Mrp2, the binding of Nrf2 to the proximal ARE (ARE #1) was observed with both vehicle and tBHQ-treated cells, and this result is in agreement with previously published observations.26 Little or no binding was observed in the more distal Mrp2 ARE (ARE #2), which may serve a minor role in Mrp2 regulation. For Mrp3, ARE #1 at 9919 bp upstream bound Nrf2, and this indicated modest binding in vehicle-treated cells that increased after tBHQ administration. Mrp4 has many potential AREs, so it was surprising that binding was observed only with a tandem of AREs (ARE #2 and ARE-like #3; 3753–3767 bp).

Figure 10.

Nrf2 can bind to the promoter regions of Mrp2, Mrp3, and Mrp4. Hepa-1 cells were treated with (−) DMSO or (+) tBHQ. DNA was crosslinked and sonicated (input), and the remainder of the DNA was immunoprecipitated with IgG (IgG) or Nrf2 antibody (Nrf2). No template (NT) represents the PCR reaction without any DNA included in the reaction. PCR primers specific for the promoter regions are listed in the Materials and Methods section.


Several studies have demonstrated that Mrp-dependent transport is inhibited when GSH is depleted,16, 32–34 but no studies have examined whether Mrp expression is altered after GSH loss. Thus, this study is the first to examine the effects of GSH deficiency on the expression of Mrp transporters. Furthermore, this study demonstrates that in response to oxidative and antioxidative activation, Nrf2 plays an important role in regulating hepatic Mrp expression.

Liver tissue has a high capacity for synthesizing and storing GSH. Buthionine sulfoximine (BSO), a GCL inhibitor, can deplete hepatic GSH 60% or more yet does not cause Nrf2 nuclear translocation in vivo.35 However, diethyl maleate (DEM), which depletes GSH by functioning as a Michael acceptor, increases the nuclear translocation of Nrf2.35 Although both chemicals have been used extensively to cause experimental GSH depletion in the liver, these chemicals may have alternative effects on the cell that are not currently understood. For example, DEM has been shown to directly inhibit Nrf2 proteosomal degradation and markedly increase the half-life of Nrf2; these processes are not generally attributed to depleted GSH levels.36 Gclc-null mice are devoid of such complexities and exhibit GSH depletion beyond the levels observed with these chemicals. In Gclc-null mice, maximal depletion of ∼96% of GSH has been observed,20 and this is markedly higher than what has been published in reports on DEM or BSO. Thus, this study represents perhaps a purer model of GSH loss on Nrf2 activation than previous experiments.

Initial modeling of the Nrf2/Keap1 system suggested that the loss of GSH changed the cellular redox environment and allowed for electrophilic accumulation. Thus, Keap1 served essentially as a sensor for electrophilic stress, releasing Nrf2 during oxidative conditions and allowing Nrf2 to translocate to the nucleus.37 Although Keap1 prevents Nrf2 translocation, it also plays a critical role in the turnover of Nrf2, serving as a Cul3-based E3 ligase substrate adaptor protein that aids in proteosomal degradation of Nrf2.38 Modifications of reactive cysteines on Keap1 are known to alter the Nrf2/Keap1 interface, and this process is now believed to be an important, if not critical, aspect of Nrf2 activation.22, 39 Although GSH depletion and Nrf2 activation are known to be linked, the exact mechanism of how this process occurs remains unclear at present. A recent report by Wang et al.40 demonstrated that the depletion of GSH with BSO was not sufficient to drive the expression of an ARE promoter/reporter construct, yet BSO combined with other Nrf2 activators led to the synergistic activation of this construct. This may suggest that although GSH depletion does not cause Nrf2 activation directly, the attenuation of the GSH buffering system may make the Nrf2/Keap1 pathway more susceptible to chemical activation by strong electrophiles. Furthermore, if GSH is depleted beyond a certain threshold as in the case of the Gclc model, even weak endogenous electrophiles may be able to activate Nrf2 because the buffer system is no longer effective.

In hepatocyte-specific Gclc-null mice, the GSH synthetic capacity is progressively eliminated in hepatocytes. Because of the γ-glutamyl linkage, the GSH tripeptide resists intracellular catabolism. It therefore must be exported—MRPs are a major route of export—before it can be broken down. GSH is catabolized at the cell's extracellular surface by γ-glutamyl transpeptidase, and no efficient import mechanism for intact GSH exists. Thus, GSH must be synthesized intracellularly. Without Gclc, synthesis cannot proceed, and hepatocytes lose virtually all GSH; this leads to an inability to maintain the oxidative homeostasis in the cell. This oxidative environment liberates Nrf2 from Keap1, allowing for Nrf2-mediated induction of genes such as Nqo1, along with marked increases in hepatic Mrps (Fig. 2). Gclc-null mice have increased levels of conjugated bilirubin in the blood, with only minor increases in unconjugated bilirubin (data not shown); this is similar to what is observed in TR rats, which harbor a spontaneous mutation in the Mrp2 gene. This is consistent with Mrp3 and Mrp4 induction, which can efflux conjugated bilirubin when up-regulated.41, 42 Furthermore, Gclc-null mice suffer from liver steatosis, inflammation and hepatocyte death, all conditions which cause compensation in transporter expression and influence the overall liver condition. At the cellular level, the accumulation of mitochondrial damage in Gclc-null mice also decreases ATP levels, and this can also limit turnover rates by transporters that are ATP-dependent.

Various natural and synthetic compounds protect against carcinogenesis by inducing phase II detoxification enzymes. Monofunctional phase II enzyme inducers, such as BHA and OPZ, tend to specifically induce gene transcription of phase II enzymes through ARE/electrophile response element sequences found in these genes43, 44 but have minimal effects on cytochrome P450s, which generally lack such sequences.44 The enhanced expression of cytoprotective enzymes such as Nqo1 and Gsts and antioxidant enzymes such as Ho-1, Gclc, and thioredoxin occurs through AREs in the promoter region of these genes.45–49

Several ARE core elements can similarly be found in the promoter regions of Mrp2, Mrp3, and Mrp4 (Fig. 9). Mrp2 has 2 ARE sequences located 185 and 1478 bp upstream of the translational start site, and a recent study has demonstrated that these AREs are capable of binding Nrf2 in electrophoretic mobility shift assays, although the more proximal ARE binds Nrf2 with much higher affinity.26 The results from the present ChIP assay mirror those observations, with strong binding observed with the proximal ARE and with only slight binding observed with the second ARE (Fig. 10). Furthermore, the more distal ARE probably has a propensity for Maf homodimerization according to a sequence analysis of this element,50 which suggests that this ARE could potentially negatively regulate Mrp2.51 For Mrp3, only the ARE that was 9919 bp upstream of the translational start site showed Nrf2 binding. The distance from the start site and the moderate binding may suggest other AREs located in intron sequences or 3′ to the gene or indirect regulation through other transcription factors that are important for Mrp3 regulation. Mrp4 has 5 total putative AREs, with 1 cluster of ARE and ARE-like elements at 3295, 3767, and 3753 bp upstream and another more distal cluster at −9104 and −10,462 bp. Two of these elements at 3.7 kb form an everted ARE repeat sequence that is similar to the SX2 fragment in the HO-1 promoter.52 This element was the only 1 of the Mrp4 AREs that showed Nrf2 binding in ChIP assays. Thus, the presence of these ARE/electrophile response elements gives further evidence that Nrf2 can directly interact with cis-acting elements in the promoter regions of Mrp2, Mrp3, and Mrp4, as some of these elements show positive Nrf2 binding in in vitro assays.

The inducible expression of phase II enzymes, such as GSTπ, by Nrf2 activators has been shown previously to be attenuated in Nrf2-null mice.53, 54 Clear regulation of downstream target genes is not uniform, and other factors, including the recruitment of coactivator and corepressor proteins, may ultimately be critical in transcriptional expression. For example, Ho-1 is induced in an Nrf2-specific manner in many experiments, yet the loss of Keap1 does not induce Ho-1.55 However, multilayered regulation of Ho-1 by corepressors such as Bach1 may ultimately contribute to transcriptional regulation. Furthermore, although the Nqo1 promoter contains an ARE consensus sequence in the proximal promoter, Ho-1 contains distal response elements that are responsive to a variety of chemical or physiological stresses. Although Nrf2 induces Ho-1 in many models, additional factors besides Nrf2 may come into play.56, 57

Nrf2 is important not only for the induction of Mrps in liver but also for their constitutive levels. In Nrf2-null mice, the constitutive expression of Mrp3 mRNA expression is approximately 33% lower than in WT mice, and Mrp3 protein expression in Nrf2-null mice is barely detectable (Figs. 5–7). Thus, the regulation of Mrp3 mRNA and protein seems similar to the regulation observed with some GST isoforms, in which mainly constitutive expression is altered.58 In contrast, hepatic Mrp4 basal expression appears to be Nrf2-independent, yet after OPZ and BHA administration, Mrp4-inducible expression is completely Nrf2-dependent (Fig. 5). Thus, as demonstrated for phase II genes, Nrf2 regulation in response to oxidative insult and antioxidant activation can alter basal, inducible, or basal and inducible expression of a target gene.

Several in vitro studies have shown that Nrf2 may regulate Mrps. In mouse embryonic fibroblasts, it has been previously shown that Mrp1 is Nrf2-dependent.59 In this study, the basal expression of Mrp1 in the liver is very low, and the Nrf2 dependence could not be determined (data not shown). Previous experiments using the KEAP1-overexpressing HepG2 cell line showed increases in the cellular content of the anionic dye C-369, suggesting that when increased KEAP1 sequesters and prevents Nrf2 transactivation, a decreased cellular efflux function can be observed.60 Furthermore, recent experiments demonstrating that the AREs in the Mrp2 promoter are responsive to Nrf2 activation,26 in combination with the Nrf2-null mouse experiments and ChIP assays performed in this study, strongly support the conclusion that Mrp2 is an Nrf2-target gene. ChIP assays performed in this study furthermore show a role for Nrf2 in Mrp3 and Mrp4 regulation as well, although they do not fully discount the contributions of other factors or other AREs located further upstream, in intronic sequences, or in 3′-flanking regions of Mrp genes. When these results are considered together, Nrf2 appears to be an important transcription factor for the basal and inducible expression of many Mrp transporters.

This study suggests that Nrf2 can regulate several members of the Mrp family, including Mrp2, Mrp3, and Mrp4, in response to a wide variety of chemicals and conditions. Thus, Nrf2 can regulate (1) phase I detoxification enzymes, such as epoxide hydrolase and Nqo1, (2) phase II conjugation enzymes, such as uridine 5′-diphospho-glucuronosyltransferases, sulfotransferases, and GSTs, and (3) efflux transporters, such as Mrp2, Mrp3, and Mrp4 (Fig. 11); this makes it an important transcription factor in hepatic metabolism and efflux. Furthermore, the regulation of Mrp transporters by Nrf2 may be an important aspect in liver diseases. Oxidative stress has been implicated in the chemical toxicity of the liver and in the pathogenesis of numerous hepatic diseases, including cholestasis, fibrosis, tyrosinemia, and cancer. Interestingly, compensatory increases in Nrf2 targets and Mrp transporters have been reported in rodent models of each of these diseases, suggesting that Nrf2 is coordinately regulating detoxification and transport pathways to mitigate cellular injury. Thus, Nrf2-mediated regulation of Mrps not only may be important in understanding hepatic transport but also may serve as an important mechanism for treating patients with liver diseases that exhibit an oxidative stress component.

Figure 11.

Summary of Nrf2 regulation of drug disposition and metabolism by regulation of phase I and II metabolism, along with efflux transport.


The authors thank the following: Nathan Cherrington for his assistance and for scientific discussions, Fumiki Katsuoka and Takafumi Suzuki for their vital help in conducting ChIP assays, Xingguo Cheng for help with sample processing, Bruno Stieger for generously providing the Mrp2 antibody, Hozumi Motohashi for scientific discussions, Amy Jakowski for her assistance with double immunofluorescence imaging, and Terry Kavanagh for generously providing the Gclc antibody. They also thank Sarah Barnes for her help with the western blot analysis.