Role of Nrf2 in the regulation of the Mrp2 (ABCC2) gene

Animal models

Female CF1 mice between 8 and 10 weeks of age (25–28 g) were used in the present study. All mice were cared for in accordance with the Guide for the Care and Use of Laboratory Animals published by the U.S. National Institutes of Health (publication no. 86-23). For 4 days, control mice were fed a commercial rodent diet (Prolab RMH 3000; PMI Nutritional International), whereas the diet of the mice in the experimental groups was supplemented with 0.75% BHA (butylated hydroxyanisole) that had been dissolved previously in ethanol. Bile flow, biliary bile acids and the concentrations of GSH, GSSG and total glutathione (GSH plus GSSG) were analysed in control and BHA-treated mice as described previously [24–27].

Cell cultures and xenobiotics

Mouse Hepa 1-6 and human HepG2 hepatoma cells were cultured in Dulbecco's modified Eagle's medium containing 4.5 g/l D-glucose (Invitrogen), 10% (v/v) foetal bovine serum (Invitrogen), 100 units/ml penicillin G, 100 units/ml streptomycin sulphate and 0.25 μg/ml amphotericin B (Invitrogen). BHA, 2,4,5-T, 2AAF (2-acetylaminofluorene) and β-NF (β-naphthoflavone) were obtained from Sigma Chemicals. All compounds were dissolved in DMSO, and the final DMSO concentration in all treatment conditions was 0.1% in complete medium.

Northern blot analysis and run-on assay

Total RNA was isolated from mouse liver or cells using the guanidinium/phenol method [28], resolved (10 μg/lane) on a 1.5% agarose/2.2 M formaldehyde gel and transferred on to nylon filters (NEN Research Products). The membranes were hybridized with mouse or human ³²P-labelled DNA probes. The DNA probes were synthesized by RT (reverse transcription)–PCR with specific primers using total RNA from mouse liver or HepG2 cells. The Gsta1/Gsta2 probe was synthesized with primers that recognize both Gsta1 and Gsta2 mouse cDNAs (Table 1). The annealing temperature in all PCRs was 58 °C. The DNA probes obtained by RT–PCR were radiolabelled using a random primer labelling system (Promega). A GAPDH (glyceraldehyde-3-phosphate dehydrogenase) DNA probe was used to normalize the RNA loading.

A run-on assay using Mrp2, GCLC, Gsta1/Gsta2 and GAPDH mouse DNA probes was performed as described previously [24,26]. The ø-X174 plasmid DNA was used as a control for non-specific binding in the assay.

Western blot analysis

Proteins were separated by SDS/PAGE (10% gels). Liver plasma membrane fractions enriched in the bile canalicular domain were prepared from control and BHA-treated animals [26]. The Mrp2 protein content was determined by immunoblotting using the anti-Mrp2/cMOAT monoclonal antibody M2 III-5 as described previously [24]. Detection of P-glycoprotein, for controlling membrane protein loading, was performed using the monoclonal antibody C219. GCLC and GST alpha-class protein contents were determined in the cytosolic fraction of liver homogenates [26] using a polyclonal antibody against human GCLC and human polyclonal anti-GST A1-1 (Calbiochem) respectively. Protein loading levels were normalized to the immunodetection of albumin.

Isolation, cloning, sequencing and characterization of the 5′-flanking region of the mouse Mrp2 gene

The 5′-flanking region of the mouse Mrp2 gene was isolated using nested PCR using Mrp2-specific primers and adaptor primers from mouse libraries of uncloned adaptor-ligated genomic DNA fragments (BD Genome Walker Kit; BD Biosciences), according to the manufacturer's instructions. The following Mrp2-specific primers near the translation start site were designed: 5′-GTA-GAGTTGCAGAATTCGTCCATCG-3′ (GSP1N: +23 to −2) and 5′-CGTGATTTCCAGGGCACTTCTGTTA-3′ (GSP2N: −1 to −24), which included the XhoI adapter sequence at the 5′ end. A 4-kb fragment of the 5′-flanking region of Mrp2 was amplified and then digested with MluI and XhoI enzymes to obtain a fragment of approx. 2 kb. This isolated fragment corresponds to the upstream sequence, and extends downstream to nucleotide −1 relative to the translation start site for the Mrp2 gene. The 2-kb fragment was ligated directionally into two reporter plasmids: LUC (luciferase) reporter gene vector pGL3 enhancer and CAT (chloramphenicol acetyltransferase) reporter gene vector (Promega), which had been predigested with the same enzymes.

To functionally characterize the isolated 5′-flanking region, a series of 5′-end deletions of the insert in p-2kb-LUC were constructed using the Erase-a-Base System (Promega) according to the manufacturer's instructions. All of the deleted inserts were sequenced using specific pGL3 primers (RVprimer3 and GLprimer2), and the complete nucleotide sequence of the 2-kb insert obtained corresponded exactly to 1995 bp of the 5′-flanking region of Mrp2. The 285-bp fragment upstream of the translation start site was obtained by digestion of p-2kb-CAT with XbaI and XhoI, and was cloned into the NheI/XhoI site in pCAT-basic, yielding the construct p-285-CAT.

The transcription start site of the Mrp2 gene was determined using the Primer Extension System-AMV Reverse Transcriptase System (Promega). Briefly, 10 μg of total RNA isolated from mouse liver and gene-specific primers GSP1N and GSP2N (end-labelled with [α-³²P]ATP using T4 polynucleotide kinase) were used according to the manufacturer's instructions. The primer-extended cDNA products were analysed on a denaturing 6% polyacrylamide gel alongside the sequencing reaction of p-2kb-LUC with the same primers. The location of the mRNA 5′-end was determined by co-migration with the p-2kb-LUC nucleotide sequence. The transcription initiation site was located 100 bases upstream of the translation start site.

Nrf2 expression plasmids

The full-length cDNA for mouse Nrf2 (GenBank® accession number U20532), a dominant-negative mutant (Nrf2-DN, amino acids 296–581; corresponding to the DNA-binding and leucine zipper dimerization domain [29]) and the N-terminal transactivation domain (Nrf2-TA, amino acids 1–296) [30] were generated by PCR with specific primers from mouse liver cDNA. The amplified products were cloned into pcDNA4/His Max TOPO (Invitrogen). The full-length Nrf2 cDNA was performed with the following primers: forward, 5′-CAGCAGGACATGGATTTGATTGAC-3′ (289–312 bp), and reverse, 5′-AGAAAAGGCTCCATCCTCCCGAAC-3′ (2044–2067 bp). Nrf2-DN was amplified using the following primers: 5′-TGTGACATGTCACTGTGTAAAGC-3′ (1180–1202 bp; where the underlining indicates a change from C to A to create a translation start site) and the Nrf2 reverse primer. To obtain Nrf2-TA, the following primers were used: the Nrf2 forward primer and 5′-GGGTTGAAAGCTTAACACAGTGACAG-3′ (1186–1212 bp; where the underlining indicates a change from T to A to create a stop codon sequence).

Transfections and reporter assays

Hepa 1-6 cells were grown in complete medium, and, 24 h before transfection, the cells were seeded on 12-well plates at a density of 1×10⁵ cells per well. Cells were transfected with 2.6 μg of DNA, 3.5 μl of Lipofectamine™ and 6 μl of Lipofectamine™ Plus reagent (Invitrogen). The total DNA included 0.5 μg of reporter plasmid (p-insert-LUC or p-insert-CAT) and 0.1 μg of transfection control plasmid [pCMV-βgal (Clontech) or pRL-TK Renilla (Promega)] with or without 2 μg of the expression plasmid (p-Nrf2, p-Nrf2-DN or p-Nrf2-TA). The total amount of DNA in each transfection assay was kept constant by using pBR322 plasmid. After 4 h of incubation, the transfection medium was replaced with complete medium or supplemented with each of the tested compounds at various concentrations. Cells were harvested after 24 h, and cell extracts were assayed for enzyme-specific activity of the reporter gene used. LUC activities were normalized to the luminescence of the Renilla internal control, and the activities were measured by the Dual-Luciferase Reporter Assay system (Promega). The CAT activity was determined by the CAT enzyme assay system with reporter lysis buffer (Promega), and was normalized to β-galactosidase activity as determined by the β-galactosidase enzyme assay system with reporter lysis buffer (Promega). All experiments were repeated at least three times.

Isolation of nuclear extracts and EMSAs (electrophoretic mobility-shift assays)

Nuclear protein extracts were isolated from HepG2 and Hepa 1-6 (DMSO control and BHA- or β-NF-treated) cells and from Hepa 1-6 cells overexpressing the DNA-binding and leucine zipper dimerization domain (Nrf2-DN). Briefly, cells were collected by centrifugation at 800 g for 5 min at 4 °C and resuspended in 400 μl of ice-cold buffer [10 mM Hepes (pH 7.8), 10 mM KCl, 1.5 mM MgCl₂, 0.2 mM PMSF and 0.5 mM DTT (dithiothreitol)]. After 15 min of incubation on ice, Nonidet P40 was added to a final concentration of 0.6%, and the cells were vortex-mixed and centrifuged for 1 min at 16000 g. The nuclear pellet was extracted with 50 μl of ice-cold buffer [10 mM Hepes (pH 7.8), 420 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 0.5 mM PMSF, 0.5 mM DTT and 25% glycerol] for 30 min at 4 °C on a rocking platform, and debris was removed by centrifugation at 16000 g for 20 min at 4 °C. The protein concentration was determined using the Bradford assay (Bio-Rad).

A double-stranded DNA probe containing the ARE-1 and ARE-2 sequences or the corresponding ARE-1 mutant oligonucleotide (ARE-1M) was end-labelled with [γ-³²P]ATP using T4 polynucleotide kinase and used for gel shift analysis. The DNA probes were 5′-ACTGGGATGACATAGCATTCATC-3′ (ARE-1) and 5′-TGCTTGCTGACATGGCAAACATC-3′ (ARE-2), where the boldface indicates the ARE core sequence; or 5′-ACTGGGA-GTCAGACGCATTCATC-3′ (ARE-1M; underlining indicates the ARE core mutation). The reaction mixture containing 2 μl of 5× gel-shift binding buffer (Promega), 5 μg of nuclear protein extract and sterile water to a total volume of 9 μl was incubated at room temperature (22 °C) for 10 min. The probe (1 μl; 2×10⁵ c.p.m.) was then added, and DNA-binding reactions were performed overnight at 4 °C. For competition experiments, unlabelled Mrp2 ARE probe was added at a 10-, 25- or 50-fold molar excess before the labelled probe. For the supershift assay, an anti-Nrf2 antibody (sc-722X; Santa Cruz Biotechnology) was added after 4 h of the binding reaction at 4 °C. DNA–protein interactions were detected by electrophoresis on a non-denaturing 5% polyacrylamide gel in 0.5× TBE buffer. Gels were dried, and DNA–protein complexes were visualized by autoradiography.

BHA induces gene expression of Mrp2, GCLC and Gsta1/Gsta2 in the mouse liver

BHA, a classical activator of Nrf2, has been used to study the activation of GCLC [31] and Gsta1/Gsta2 [21] by ARE-dependent mechanisms. The effect of BHA on Mrp2 gene expression was assessed here in vivo. As shown in Figure 1(A), the mRNA levels for the Mrp2 and GCLC genes were higher in the livers of BHA-treated mice than in control mice. Gsta1/Gsta2 mRNA was not detected in the livers of control mice and, as expected, was strongly induced in the livers of BHA-treated mice. The Gsta3 mRNA level did not change in treated mice, as assessed by RT–PCR (results not shown).

Western blotting was used to determine whether increased levels of hepatic Mrp2, GCLC and Gsta1/Gsta2 mRNAs were associated with increased expression of the encoded products. The Mrp2 protein content in the canalicular domain of hepatocytes was 3-fold higher in BHA-treated mice than in control animals, as shown in Figure 1(B). The contents of GCLC and Gsta1/2 proteins were also increased (1.8- and 11-fold respectively) in the cytosolic fraction of proteins in the livers of BHA-treated mice (Figure 1C).

The relative transcription rates of Mrp2 were also examined using a nuclear run-on assay with nuclei isolated from control and BHA-treated mice livers. As seen in Figure 1(D), the transcription rates of Mrp2, GCLC and Gsta1/Gsta2 were higher in BHA-treated mice than in control mice. No changes in the level of GAPDH-labelling RNA transcript, a gene of constitutive expression, were detected. This result indicates that BHA also induces the transcription of the Mrp2 gene.

The hepatocellular concentration of total glutathione was 2-fold higher in BHA-treated mice than in control animals, suggesting an increase of GSH synthesis: total glutathione was 15.49±0.16 compared with 7.99±1.82 μmol/g of wet liver, and GSSG was 0.30±0.49 compared with 0.48±0.38 μmol/g of wet liver (P<0.05, n=3). As evident from Table 2, the bile flow was significantly higher in BHA-treated mice, without changes in the biliary output of bile salts. Remarkably, a significantly higher biliary GSH output was observed in BHA-treated mice, indicating that the newly synthesized Mrp2 transporter was functionally incorporated into the canalicular domain of hepatocytes.

Taken together, these results suggest that BHA induces not only Phase II detoxifying enzymes (Gsta1/Gsta2) and the rate-limiting enzyme of GSH synthesis (GCLC) in the mouse liver as described previously [21,31], but also the Mrp2 transporter via transcriptional activation of the specific genes.

Co-induction of Mrp2 and GCLC genes by xenobiotics in hepatoma cell lines

The above in vivo observations of the co-induction of Mrp2 and GCLC genes in the mouse liver were extended to a cell culture model using the human (HepG2) and mouse (Hepa 1-6) hepatoma cell lines. Cells were cultured with classical activators of the ARE–Nrf2 pathway, such as the synthetic flavonoid β-NF and the antioxidant BHA [16,31], as well as chemical inducers of the Mrp2 gene: 2,4,5-T [24] and 2AAF [32]. The mechanisms of induction of these two compounds remain undefined. Exposure of Hepa 1-6 and HepG2 cells for 24 h to different concentrations of all of the compounds revealed dose-dependent increases in the levels of Mrp2 and GCLC mRNAs (Figures 2A and 2B). The induction of Mrp2 as well as GCLC was time-dependent in the Hepa 1-6 cell line (Figures 2C and 2D). The compounds 2,4,5-T, 2AAF and BHA induced both genes with a similar pattern: induction was first detected at 12 h after treatment, with the maximal induction (2–2.5-fold) occurring after 24 h. The β-NF-mediated induction of both genes showed a similar pattern in this cell line, but the induction occurred as early as 6 h after treatment, and peaked at 12 h.

The similar patterns of co-induction of Mrp2 and GCLC genes in human and mouse hepatoma cell lines by classical activators of Nrf2 (BHA and β-NF) and the chemical inducers (2,4,5-T and 2AAF) suggests conserved molecular mechanisms of transcriptional activation, which may be mediated by the ARE–Nrf2 pathway.

The 5′-flanking region of the mouse Mrp2 gene contains two ARE sequences

To determine the presence of ARE sequences in the 5′-flanking region of the mouse Mrp2 gene, a 1995-bp fragment was cloned and sequenced as described in the Materials and methods section. The BLAST program revealed that the isolated fragment exhibited 99.8% nucleotide identity with a clone of genomic DNA (GenBank® accession number AC132251) located in mouse chromosome 19. As expected, several consensus motifs for the potential transcription-factor-binding sites were identified using the program TRANSFAC [33]. The isolated region contains liver-abundant transcription-factor-binding sites for C/EBP (CCAAT/enhancer-binding protein), HNF-3β (hepatocyte nuclear factor 3β), HNF-4 and PXR (pregnane X receptor)/CAR (constitutively active receptor)/FXR (farnesoid X receptor) (ER-8: everted repeat 8) [34], and ubiquitous factors expressed in multiple tissues including AP-1 (activator protein 1), NF-Y (nuclear factor Y), NF-κB (nuclear factor κB) and SP-1 (specificity protein 1). Interestingly, two ARE-like sequences were found in the isolated region: one at positions −1391 to −1381 (ARE-2; cTGACatgGCa, with lower-case letters indicating non-conserved bases) and the other at positions −95 to −85 (ARE-1; aTGACataGCa) from the transcription start site. The ARE-1 sequence is identical with the previously reported minimal ARE enhancer sequence (a/g)TGA(C/T/G)nnnGC(a/g) [17]. In contrast, the ARE-2 sequence differs from the consensus sequence in the first base. TATA box sequences were not found in the 5′-flanking region of the mouse Mrp2 gene. The complete sequence of the isolated fragment and the putative binding sites for the transcription factors are shown in Figure 3 (GenBank® accession number AY905402).

Deletion analysis of the 5′-flanking region of the Mrp2 gene

In order to identify the sequences in the 5′-flanking region of the Mrp2 gene that mediate constitutive gene expression, LUC or CAT reporter gene assays with constructs containing the complete isolated flanking region or 5′-unidirectionally deleted inserts were performed in Hepa 1-6-transfected cells. As shown in Figure 4(A), the deleted constructs p−1598/+99-LUC, p−995/+99-LUC, p−445/+99-LUC and p−255/+99-LUC exhibited LUC activity similar to that of the full-length construct (p−1895/+99-LUC). However, the construct p−94/+99-LUC resulted in a 70% decrease in the LUC activity (P<0.05, n=3). Interestingly, this construct contains the ARE-1 sequence truncated in the first base pair. These results suggest that the ARE-2 sequence is not relevant to the constitutive Mrp2 gene expression. On the other hand, the activity of a minor construct (p−185/+99-CAT) was similar to the activity of the p−255/+99-LUC (78–80%) (Figure 4B), indicating that positions −185 to +99 of the 5′-flanking region contain the elements that are responsible for the basal transcriptional activity of the Mrp2 gene. This region contains the putative binding sites for the transcription factors C/EBP, SP-1, NF-κB and ARE-1, as shown in the representative map of the Mrp2 5′-flanking region illustrated in Figure 4(C).

Activity of the Mrp2 reporter gene constructs is induced by xenobiotics

To determine whether the isolated 5′-flanking region of the Mrp2 gene contains the sequences that mediate the chemical induction of the Mrp2 gene, Hepa 1-6 cells were transfected with either the construct containing the full region and both ARE-like elements (p−1895/+99-CAT) or the construct responsible for the basal transcriptional activity of the Mrp2 gene (p−185/+99-CAT). The transfected cells were exposed to the following xenobiotics: 250 μM BHA, 500 μM 2,4,5-T or 25 μM β-NF, or to 0.1% DMSO as a control. After 24 h, the CAT reporter activity of both constructs was measured in cell extracts (Figure 5A). We observed that all of the compounds studied induced CAT activity (at different levels) in both constructs compared with control (DMSO). Interestingly, the relative amount of induced CAT activity (p−1895/+99-CAT or p−185/+99-CAT) was similar in both constructs for each compound tested, suggesting that the proximal region, containing the ARE-1 sequence, is responsible for the xenobiotic-mediated induction of the Mrp2 gene (Figure 5B). These experiments also suggest that the ARE-2 sequence is not involved in the xenobiotic-mediated induction of this gene.

Overexpression of Nrf2 enhances the activity of the Mrp2 5′-flanking region

Transient co-transfection assays were used to determine whether the ARE-like sequences present in the Mrp2 gene are activated directly by Nrf2. The two constructs of the Mrp2 5′-flanking region, p−1895/+99-CAT or p−185/+99-CAT, were co-transfected into Hepa 1-6 cells with either an Nrf2 expression plasmid or one of two Nrf2-mutant-expressing plasmids (p-Nrf2-TA or p-Nrf2-DN). Overexpression of the Nrf2 protein significantly enhanced the CAT activity of both constructs compared with controls (Figure 6): 2.2-fold for p−1895/+99-CAT and 3.5-fold for p−185/+99-CAT compared with control (P<0.05, n=3). On the other hand, overexpression of both Nrf2-TA and Nrf2-DN mutant proteins significantly decreased CAT activity to below basal values (to less than 50% in both constructs). Taken together, these results indicate that the ARE-1 sequence (located at positions −95 to −85 relative to the transcription start site of Mrp2) is the only ARE sequence activated by the Nrf2 transcription factor.

ARE sequences of the 5′-flanking region of the Mrp2 gene bind Nrf2 complexes

EMSAs were performed to characterize the protein binding of ARE-1 and ARE-2 sequences present in the 5′-flanking region of the Mrp2 gene. Nuclear extract proteins obtained from Hepa 1-6 cells (control) or transfected cells overexpressing Nrf2-DN, and oligonucleotides (23 nucleotides long) containing ARE-1 or ARE-2 sequences, were used in the assay. As shown in Figure 7(A), nuclear extract proteins from control Hepa 1-6 cells bind, with different affinities, to the ARE-1 and ARE-2 sequences, determining a main DNA–protein complex (band a in Figure 7A). Only ARE-1 showed an additional upper DNA–protein complex (Figure 7A, band b). These bands are specific, because they are effectively competed for by an excess amount of corresponding unlabelled ARE and by Nrf2-DN protein in the nuclear extract (Figure 7A, band c). This result indicates that both ARE sequences are recognized by the DNA-binding and leucine zipper dimerization domain (Nrf2-DN) of Nrf2 (Figure 7A, band c), and also suggests that ARE-1 has a higher affinity for forming different Nrf2–protein complexes (Figure 7A, bands a and b) than does ARE-2. In addition, to demonstrate that the ARE-1 sequence binds specifically to Nrf2 complexes, the ARE-1 core was replaced by the mutant core of the human NQO1 ARE sequence, which has been used previously to assess the specificity of binding of Nrf2 to the ARE sequences [35]. As shown in Figure 7(B), when the EMSA was performed using the mutated ARE-1M oligonucleotide, only a weak main DNA–protein complex (band a) was detected using both control and Nrf2-DN nuclear extracts. The absence of additional bands (Figure 7B, bands b and c) in this assay suggests that the ARE-1 sequence specifically binds Nrf2–protein complexes. We propose that the main complex (Figure 7B, band a) is formed by the ARE-1 sequence binding the heterodimer of Nrf2 with Maf proteins, and the upper complex (Figure 7B, band b) corresponds to a second complex in which additional unknown proteins have been recruited. The three DNA–protein complexes (Figure 7B, bands a, b and c) were effectively competed with the unlabelled ARE-1 oligonucleotide in both nuclear extracts.

To assess whether Nrf2 activated by β-NF and BHA is able to bind the Mrp2 ARE sequences, an EMSA was performed using nuclear extracts isolated from HepG2 cells treated with 25 μM β-NF or 250 μM BHA for 3 h. Remarkably, as shown in Figure 8(A), the presence of β-NF and BHA increased the binding of Nrf2–protein complexes to the ARE-1 sequence, indicating that ARE-2 has a lower affinity for forming Nrf2–protein complexes than does ARE-1. The presence of Nrf2 in ARE-1–protein complexes was determined by a supershift assay using anti-Nrf2 antibody (SC-722x) and nuclear extracts isolated from HepG2 cells treated with 250 μM BHA for 3 h. As shown in Figure 8(B), ARE-1 bound more protein complex from BHA-treated cells than control nuclear extracts. The addition of anti-Nrf2 antibody resulted in weak supershift bands in both extracts, indicating that Nrf2 is present in the ARE-1–protein complex. The supershift assay was also performed with nuclear extracts isolated from Hepa 1-6 cells. However, we were unable to obtain supershift bands in Hepa 1-6 cells (results not shown) because this antibody did not bind reliably to mouse Nrf2; this behaviour has also been reported by other authors [36].