Small maf (MafG and MafK) proteins negatively regulate antioxidant response element-mediated expression and antioxidant induction of the NAD(P)H:Quinone oxidoreductase1 gene.

The antioxidant response element (ARE) is known to regulate expression and induction of NQO1, GST Ya, and other detoxifying enzyme genes in response to antioxidants and xenobiotics. The nuclear transcription factor Nrf2 and Nrf1 bind to the ARE and positively regulate expression and induction of the NQO1 and GST Ya genes. In this study, we demonstrate that overexpression of small Maf (MafG and MafK) proteins negatively regulate ARE-mediated expression and tert-butyl hydroquinone induction of the NQO1 and GST Ya genes in transfected Hep-G2 cells. In similar experiments, overexpression of small Maf proteins also repressed Nrf2-mediated up-regulation of ARE-mediated NQO1 and GST Ya genes expression in Hep-G2 cells co-transfected with Nrf2 and small Maf proteins. Band and supershift assays with the NQO1 gene ARE and nuclear proteins demonstrate that small MafG and MafK bind to the ARE as Maf-Maf homodimers and Maf-Nrf2 heterodimers. Therefore, Maf-Maf homodimers and possibly Maf-Nrf2 heterodimers play a role in negative regulation of ARE-mediated transcription and antioxidant induction of NQO1 and other detoxifying enzyme genes. In contrast to Maf-Nrf2, the Maf-Nrf1 heterodimers failed to bind with the NQO1 gene ARE and did not demonstrate the repressive effect in transfection assays.

philes and ROS with glutathione (7,8); UDP-glucuronosyl transferases, which catalyze the conjugation of glucuronic acid with xenobiotics and drugs for excretion (9); epoxide hydrolase, which inactivates epoxides (10); ␥-glutamylcysteine synthetase (␥-GCS), which plays a role in glutathione metabolism (11); ferritin-L gene, which plays an important role in iron storage (12); and heme oxygenase-1, which catalyzes the first and ratelimiting step in heme catabolism (13). The coordinated induction of these genes, including NQO1, protects cells against free radical damage, oxidative stress, and neoplasia. It is critical in achieving chemoprevention. Deletion mutagenesis studies of the human NQO1 gene promoter identified 24 base pairs of an antioxidant response element (ARE) between nucleotides Ϫ470 and Ϫ447. This region is required for basal expression and induction of NQO1 in response to ␤-naphthoflavone and t-BHQ (6). ARE elements have also been found in the promoter region of the human NQO2 gene (14), the rat and mouse GST Ya subunit genes (15)(16)(17)(18)(19), the rat GST P gene (20), the human ␥-GCS gene (11), the ferritin-L gene (12), and the human heme oxygenase-1 gene (21). Analysis of the AREs from various genes revealed that they contain AP1/AP1-like elements arranged as inverse or direct repeats. This is followed by a GC box (22). Additional cis-elements and nucleotide sequences, flanking the core sequence, have been shown to contribute to the ARE-mediated expression and induction of detoxifying enzyme genes (12,23,24).
Small Maf (MafG, MafK, and MafF) proteins are leucine zipper proteins that are known to repress, as well as activate, transcription of many eukaryotic genes including ␤-globin gene (38 -41). Their gene products are closely related to v-Maf, especially in the structure of the DNA-binding and leucine zipper domains. The products of these small Maf proteins, however, lack the amino-terminal domain of v-Maf (39 -41

EXPERIMENTAL PROCEDURES
Materials-The enzymes used in this study were purchased from Life Technologies, Inc. pcDNA3.1/V-5-His-TOPO and anti-V5 antibody were purchased from Invitrogen (Carlsbad, CA). ␣-Minimum essential medium was purchased from Life Technologies, Inc. Effectene transfection reagent was purchased from Qiagen (Valencia, CA). The Nrf2 antibodies were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). t-BHQ and all other chemicals were purchased from Sigma. The pGL2 promoter plasmid containing firefly luciferase gene, internal control plasmid pRL-TK that encodes Renilla luciferase, the dual luciferase assay kit, and the TNT T7/T3 coupled rabbit reticulocyte lysate system were obtained from Promega (Madison, WI). The Hybond ECL nitrocellulose membrane, ECL Western blot analysis kit, and Amplify NAMP1000 were purchased from Amersham Pharmacia Biotech.
Plasmid Construction-The NQO1 gene ARE, mutant ARE, and GST Ya ARE were subcloned in the pGL2 vector to generate reporter plasmids pGL2-hARE-Luc, pGL2-mhARE-Luc, and pGL2-GST Ya ARE-Luc. The nucleotide sequences of the various AREs are shown in Fig. 1. The sense and antisense oligonucleotides corresponding to the AREs were synthesized with either NheI/XhoI sites (NQO1 hARE and mutant hARE) or NheI/BamHI sites (GST Ya ARE). The oligonucleotides were annealed, phosphorylated using T4 polynucleotide kinase, and cloned at the respective sites in the pGL2 promoter vector. The GST Ya ARE was cloned at the NheI/BglII site. The various constructs were checked by DNA sequencing.
The mouse Nrf2 and Nrf1 cDNAs were kindly provided by Dr. Jefferson Y. Chan (University of California, San Francisco, CA). The full-length Nrf2 and Nrf1 cDNAs were amplified by polymerase chain reaction and subcloned separately into the mammalian expression vector pcDNA 3.1 to make the expression plasmids pcDNA-Nrf2R, pcDNA-Nrf2C, pcDNA-Nrf1R, and pcDNA-Nrf1C. The chicken MafG and MafK cDNAs were provided by Dr. Makoto Nishizawa (The Scripps Research Institute, La Jolla, CA). The cDNAs encoding MafG and MafK were subcloned separately into the pcDNA 3.1 vector to generate the expression plasmids pcDNA-MafGR, pcDNA-MafGC, pcDNA-MafKR, and pcDNA-MafKC. The plasmids with an R suffix contain cDNAs in the reverse orientation, and those with the suffix C contain cDNAs in the correct orientation. The MafG and MafK cDNAs were also subcloned in-frame with the V5 epitope of the pcDNA 3.1 expression plasmid. These plasmids encode the V5-tagged MafG-V5 and MafK-V5 proteins. The V5 epitope contains 14 amino acids in the sequence Gly-Lys-Pro-Ile-Pro-Asn-Pro-Leu-Leu-Gly-Leu-Asp-Ser-Thr.
Cell Culture and Co-transfection of Reporter and Expression Plasmids-Human hepatoblastoma (Hep-G2) cells were grown in six-well monolayer cultures containing ␣-minimum essential medium supplemented with fetal bovine serum (25). The Effectene transfection reagent kit from Qiagen was used to perform the transfections by procedures as described in the manufacturer's protocol. Briefly, 0.5 g of reporter constructs (hARE-Luc, mutant hARE-Luc, and GST Ya ARE-Luc) were mixed, individually and in combinations, with different concentrations of the pcDNA expression plasmids (Nrf2, Nrf1, MafG, and MafK) and transfected into Hep-G2 cells. The plasmid pRL-TK encoding Renilla luciferase was used as the internal control in each transfection. The various plasmids were mixed with DNA condensation buffer and Enhancer solution from the kit and incubated at room temperature for 5 min. This was followed by addition of Effectene reagent to the mixture. This mixture was incubated for 7 min at room temperature. The DNA-Enhancer-Effectene mixture was added dropwise onto the Hep-G2 cells and incubated at 37°C with 5% CO 2 . 48 h after the transfection, the cells were washed with 1ϫ phosphate-buffered saline and lysed in 1ϫ passive lysis buffer from the kit. The dual-luciferase reporter assay system from Promega was used to assay the samples for luciferase activity as described in the manufacturer's protocol. First, the cell lysate was assayed for the firefly luciferase activity using 100 l of the substrate LARII. Then 100 l of the STOP & GLO reagent was added to quench the firefly luciferase activity and activate the Renilla luciferase activity, which was also measured. The assays were carried out in a Packard luminometer, and the relative luciferase activity was calculated as follows: 100,000/activity of Renilla luciferase (in units) ϫ activity of firefly luciferase (in units). Each set of transfections was repeated three times. For induction studies, the cells were treated with 200 M t-BHQ, dissolved in Me 2 SO for 16 h and analyzed for luciferase activity by procedures as described above. The transfection experiments were performed using pcDNA-Maf and pcDNA-Maf-V5 constructs. Both sets of expression plasmids gave similar results. The presence or absence of V5 epitope-tagged to Maf proteins had no effect on their activity/function. Therefore, we have shown transfection data only on Maf proteins without V5 tag. The addition of V5 tag to Maf proteins enabled us to use antibodies against V5 peptide in Western and band and supershift assays.
Gel Shift/Supershift Assays-The nuclear extracts from Hep-G2 (Me 2 SO control and t-BHQ-treated (200 M for 16 h)) cells were prepared by previously described procedures (25). The Hep-G2 cells were co-transfected with pcDNA-Nrf2C and pcDNA-MafG-V5 in 1:1 ratio to overexpress Nrf2 and MafG-V5. These cells were treated with Me 2 SO and t-BHQ (200 M for 16 h) and nuclear extract prepared by previously described procedures (25). The in vitro transcription/translation of the plasmids encoding Nrf1, Nrf2, MafG-V5, and MafK-V5 were performed using the TNT-coupled rabbit reticulocyte lysate system (Promega) by procedures as suggested in the manufacturer's protocol. Redivue L-[ 35 S]methionine was substituted for methionine in the reactions. After the coupled transcription/translation, the proteins were checked for their correct size on a 10% PAGE and Western analysis. Briefly, 5 l of the translated proteins were resolved on a 10% PAGE, treated with Amplify solution (NAMP 100; Amersham Pharmacia Biotech) to enhance the 35 S signal, dried, and exposed to x-ray film. In a similar experiment, the proteins were transferred onto a Hybond ECL nitrocellulose membrane and probed with Nrf2 and V5 antibodies. V5 antibodies were used to detect the V5 tagged Maf proteins. All of the in vitro translated proteins gave the expected sized products.
The NQO1 gene ARE was end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase. The labeled ARE was incubated with nuclear extracts or in vitro translated proteins. Band and supershift assays were then performed by previously described procedures (25,31). 10 g of the nuclear extract or equimolar concentrations of in vitro translated proteins were used in the gel shift and supershift experiments. The supershift assay with nuclear extracts used 20 g of proteins. 1.5 l of anti-V5 antibody and 3 l of Nrf2 antibody or preimmune serum were used in the supershift assays.

RESULTS
The transfection of Hep-G2 cells with expression plasmids pcDNA-MafG-V5, pcDNA-MafK-V5, and pcDNA-Nrf2(C) led to the overexpression of these respective proteins as determined by SDS-PAGE, Western analysis, and antibody probing (data not shown). Overexpression of the various proteins was in near linear range between 0.1 and 1.0 g of plasmids used for transfecting Hep-G2 cells. In similar experiments, transfection of Hep-G2 cells with cDNA in reverse orientation did not result in overexpression of the respective proteins.
Transfection of Hep-G2 cells with the hARE-Luc plasmid expressed luciferase activity (Fig. 2). Overexpression of MafG in Hep-G2 cells repressed hARE-mediated luciferase activity in transfected cells ( Fig. 2A, left panel). This repression was MafG concentration-dependent. The transfection of 1.0 g of pcDNA-MafG(C) plasmid repressed 83% of hARE-mediated luciferase activity in transfected Hep-G2 cells. Interestingly, transfection of Hep-G2 cells with plasmid pcDNA-MafG(R) expressing antisense MafG RNA significantly increased the hARE-mediated luciferase activity ( Fig. 2A, left panel). The transfection of 0.5 g of plasmid pcDNA-MafG(R) caused 3.4fold increase in the hARE-mediated luciferase activity ( Fig. 2A,  left panel). Contrary to MafG, overexpression of Nrf2 in Hep-G2 cells led to increased expression of the NQO1 gene ARE-mediated luciferase gene ( Fig. 2A). In a similar experiment, the Nrf2-mediated up-regulation of luciferase gene expression was repressed because of the overexpression of MafG ( Fig. 2A, right panel). This repression was also MafG concentration-dependent. Similar results were also observed with Hep-G2 cells overexpressing MafK alone or MafK with Nrf2 (Fig. 2B). The overexpression of MafK repressed hARE-mediated luciferase gene expression and its activation by Nrf2. Replacement of the reporter plasmid hARE-Luc with mutant hARE-Luc resulted in the loss of basal expression and repression of the luciferase gene by MafG and MafK (Fig. 2C). In related experiments, the overexpression of MafG and MafK also repressed GST Ya ARE-mediated luciferase gene expression and up-regulation by Nrf2 (Fig. 3). t-BHQ treatment of Hep-G2 cells, transfected with hARE-Luc and GST Ya ARE-Luc, increased luciferase gene by approximately 2-fold (Fig. 4). Interestingly, t-BHQ induction of hARE-and GST Ya AREmediated luciferase gene expression was also repressed in Hep-G2 cells that overexpressed MafG and MafK (Fig. 4). The repression of t-BHQ induced expression was more or less proportional to the repression of basal expression. In other words, overexpression of small Maf proteins inhibited the basal and t-BHQ induced luciferase activity equally.
The pcDNA-Nrf2C, pcDNA-Nrf1, pcDNA-MafG-V5, and pcDNA-MafK-V5 plasmids were in vitro transcribed and translated with rabbit reticulocyte lysate system. SDS-PAGE analysis of in vitro translated Nrf2, Nrf1, and V5-tagged MafG and MafK is shown in Fig. 5. The in vitro translated Nrf2 and Nrf1 proteins migrated between the 105-and 75-kDa standards. Nrf1 migrated slower than Nrf2. The in vitro translated MafG-V5 and MafK-V5 proteins moved faster than the 25-kDa molecular mass standard. MafK-V5 moved faster than MafG-V5 in SDS-PAGE. The in vitro translated proteins were confirmed by Western blotting and probing with specific antibodies (data not shown).
The binding of in vitro translated MafG-V5, MafK-V5, Nrf2, and Nrf1 to the NQO1 gene ARE was determined by band and supershift assays (Fig. 6). MafG and MafK both bound to the hARE as Maf-Maf homodimers and Nrf2-Maf heterodimers (Fig. 6). The binding of homo-and heterodimers of Maf and Nrf2 were competed with cold hARE (data not shown). The presence of V-5-tagged MafG and MafK in Maf-Maf homodimers and Nrf2-Maf heterodimers were confirmed by supershift assays with anti-V5 antibodies (Fig. 6). The antibodies against Nrf2 also supershifted Nrf2-Maf heterodimers in experiments with Nrf2-MafG and Nrf2-MafK (Fig. 6) The results of band shift assays with hARE and nuclear extract from Hep-G2 cells treated with Me 2 SO (control) and t-BHQ are shown (Fig. 7A). The results demonstrated twoshifted bands as observed with in vitro translated Nrf2 and Maf proteins (Fig. 6). The upper and lower shifted bands corresponded to the Nrf2-Maf heterodimers and Maf-Maf homodimers, respectively. It may be noteworthy that both the lower and upper bands are not exclusively due to Nrf2-Maf and Maf-Maf dimers. These bands are also expected to contain positive factors including Nrf2-Jun (31). The presence of Nrf2 and Maf in the upper and lower complexes was determined by supershift assays with respective antibodies (Fig. 7B). The Hep-G2 cells treated with t-BHQ showed increase in binding of nuclear proteins in upper and lower bands, as compared with Me 2 SO-treated control (Fig. 7A). This increase is presumably due to increased binding of positive factors including Nrf2-Jun (31). The Hep-G2 cells overexpressing cDNA derived Nrf2 and MafG also showed increased binding as compared with control. However, the increase was more significant in binding of upper (containing Nrf2ϩMaf) band, as compared with lower (containing Maf-Maf) band. The treatment of Hep-G2 cells overexpressing Nrf2ϩMafG with t-BHQ also resulted in increased binding of upper and lower bands.
The overexpression of Nrf1 in Hep-G2 cells is known to increase the hARE-mediated gene expression in transfected Hep-G2 cells (26). The coexpression of the varying concentrations of Nrf1 with a constant concentration of MafG led an increase in hARE-mediated luciferase gene expression (Fig. 8). This increase was Nrf1 concentration-dependent. However, in the similar experiment, the replacement of Nrf1 with Nrf2 did not lead to increase in hARE-mediated luciferase gene expression in transfected Hep-G2 cells (Fig. 8). The results of the band shift assays with NQO1 gene hARE and in vitro translated Nrf1 are shown in Fig. 9. In band shift experiments, in vitro translated Nrf1, alone and combined with MafG or MafK, failed to bind to the NQO1 gene ARE as homodimers or heterodimers with MafG and MafK (Fig. 9). DISCUSSION Small Maf proteins (MafG, MafK, and MafF) possess leucine zipper DNA binding domains but lack transactivation domains (39 -41, 43). Most studies on the role of small Maf proteins were done with NF-E2 binding site-regulated expression of the ␤-globin gene (43). The small Maf proteins bind to the NF-E2 site of the ␤-globin gene and directly control the DNA binding properties of erythroid specific factor NF-E2 p45 (43). Homodimers of the small Maf proteins act as negative regulators, whereas heterodimers of Maf and p45 support active transcription in vivo (43). Recently, small Maf proteins were also shown to homodimerize and heterodimerize with the ubiquitous factors Nrf1/Nrf2 and regulate the MARE-mediated ␤-globin gene expression (42). Maf-Maf homodimers repressed yet Maf-Nrf heterodimers activated transcription of the ␤-globin gene (42). The Nrf2, Nrf1, and Maf proteins are also known to bind with AREs from several detoxifying enzyme genes (32,37). Nrf2 and Nrf1 are known to up-regulate the ARE-mediated expression and induction of NQO1, GST Ya, ␥-GCS, and heme oxygenase-1 genes (21,26,31,34,37). However, the role of small Maf proteins in the ARE-mediated detoxifying enzyme genes expression especially its mechanism of action remains unknown. It may be noteworthy that ARE is a distinct element than MARE and TRE (22)(23)38). Therefore, the role of Maf-Maf homodimers and Nrf-Maf heterodimers in ARE-mediated regulation of detoxifying enzyme genes expression remains relatively unknown.
In the present report, we demonstrate that small Maf proteins contribute significantly to the regulation of ARE-mediated NQO1 and GST Ya gene expression. The small Maf (MafG and MafK) proteins negatively regulate ARE-mediated expression and antioxidant induction of NQO1 and GST Ya gene expression. This conclusion is based on the following observa- regulator of ARE-mediated gene expression. Therefore, there is some evidence demonstrating that Nrf2-small Maf heterodimers also contributes to the negative regulation of AREmediated gene expression. However, the role of Nrf2-small Maf heterodimers in negative regulation of ARE-mediated gene expression remains to be confirmed by further experiments. Our results on the repressive role of Nrf2-small Maf heterodimers is also supported by a recent report on the repressive role of Nrf2-MafK in ARE-mediated gene expression (44). This report, however, did not detect the Maf-Maf homodimers in band shift assays as observed by us in the present study. The possible explanation is that we ran our gel longer than that run in the published report. The Maf-Maf homodimers run very close to an unspecific band from rabbit reticulocyte lysate and could only be separated by running the band shift gels longer (Fig. 6). In addition, we used antibodies to demonstrate the presence of Maf-Maf homodimers in band shift assays (Fig. 6). The supershift assays using specific antibodies were not done in the recently published report on the role of MafK in AREmediated gene expression (44).
In experiments similar to those described above, Nrf1 be-FIG. 6. Band shift assays. NQO1 gene ARE was end-labeled with [␥-32 P]ATP. 50,000 cpm of the labeled hARE was incubated with in vitro translated MafG-V5 or MafK-V5 either alone or in combination with in vitro translated Nrf2 as shown. The MafG-V5, MafK-V5, and Nrf2 alone and in combinations were preincubated at 37°C for 15 min before incubation with the labeled hARE. The band shift experiment was performed at room temperature. The band shift reaction mixture was incubated with preimmune serum and anti-V5 antibody or Nrf2 antibody for 2 h at 4°C in supershift assays. The band shift and supershift mixtures were analyzed on a 5% nondenaturing polyacrylamide gel. The gel was dried and autoradiographed. The asterisk denotes the nonspecific band from rabbit reticulocyte lysate. haved differently than Nrf2. Interestingly, overexpression of Nrf1, along with MafG, did not reveal a role of Nrf1-MafG in the negative regulation of hARE-mediated NQO1 gene expression. Overexpression of increasing concentration of Nrf1 and Nrf2 with a constant concentration of MafG showed different results. The increase in Nrf1 and not Nrf2 resulted in increased expression of ARE-mediated gene expression in transfected cells. In addition, Nrf1-MafG and Nrf1-MafK heterodimers were not detected in the band and supershift assays. These results led to the conclusion that unlike Nrf2, Nrf1-small Maf heterodimers do not bind to the ARE and have more or less no effect on the gene expression. Nrf1 is known to form heterodimers with hMaf, which binds to NF-E2 binding site and ␥-GCS ARE (42,35). It is possible that Nrf1 formed heterodimers with MafG and MafK in our studies. However, Nrf1-MafG and Nrf1-MafK heterodimers failed to bind with NQO1 gene ARE. Nrf1 has also been shown to bind with ␥-GCS ARE as homodimer (35). However, Nrf1 does not bind with NQO1 gene ARE either as homodimer (Ref. 31 and present studies) or as heterodimer (present studies) with small Maf proteins. The significance of this difference between NQO1 and ␥-GCS ARE remains unknown. In addition, the differential effect of Nrf2 and Nrf1 with Maf proteins on NQO1 gene ARE also remains unknown.
A model to demonstrate the negative role of small Maf (MafG and MafK) proteins in ARE-mediated expression and induction of NQO1 and other detoxifying enzyme genes is depicted in Fig.  10. The metabolism of antioxidants and xenobiotics generates electrophiles and ROS that signal the activation of a battery of genes including NQO1 and GST Ya (1). Electrophiles and/or ROS lead to the transcriptional activation and/or post-transcriptional modification of positive regulatory factors including Nrf2, c-Jun, and unknown factors. The positive factors bind to the ARE and activate transcription of NQO1, GST Ya, and other detoxifying enzyme genes by unknown mechanism(s). Electrophiles and/or ROS may also lead to the transcriptional activation and/or post-transcriptional modification of negative regulatory factors including the small Maf proteins, c-Fos, Fra, and unknown factors, which also bind to ARE and down-regulate the expression of detoxifying enzyme genes. It is possible that the activation of positive factors is an early response and is followed by a late response activation of negative regulators. However, this chain of events requires further study. It is also possible that the concentrations/activities of negative regulatory factors remain unaffected because of a signal from electrophiles and/ROS. This would indicate that ARE-mediated gene expression might be a balance between positive and negative regulators. Further studies are required to test these hypotheses.
It has recently been proposed that Nrf-Maf heterodimers could also possibly activate ARE-mediated ␥-GCS gene expression (35,37). Another report also suggested the positive role of Nrf2-Maf in ARE-mediated gene expression in general (45). However, these reports only speculated the role of Nrf2-Maf in positive regulation of ARE-mediated gene expression (35,37,45). This speculation was exclusively based on in vitro binding of these factors to the ARE. No in vivo transfection data were shown in support of a positive role of Nrf2-Maf in regulation of ARE-mediated gene expression. In contrast, the Nrf2 and Maf heterodimers have been shown to negatively regulate AREmediated GST Ya and QR genes expression and induction (44). This conclusion on the negative role of Nrf2-Maf was supported by in vitro gel shift as well as in vivo transfection experiments (44). Our data in the present manuscript support negative regulation by Maf-Maf homodimers and possibly by Maf-Nrf2 heterodimers. Further studies should reveal more information on the negative or positive role of Nrf2-Maf heterodimers in ARE-mediated gene regulation. The difference in responses among the various genes ARE to Nrf2-Maf heterodimers also remains to be studied.
From this and previous studies, the small Maf proteins join the previously identified negative regulators c-Fos and Fra1 as repressors of ARE-mediated gene expression (26,31,46). These results raise an interesting question regarding the requirement of positive and negative regulators in normal cells. One hypothesis is that small amounts of superoxide and related reactive species are consistently required for keeping cellular defenses active. Because the activation of detoxifying and defensive proteins leads to a significant reduction in the levels of free radicals, cells may require negative regulatory factors to keep the levels of superoxide from falling below the level needed to keep cellular defenses active. FIG. 9. Band shift assays. 50,000 cpm of the labeled hARE was incubated with in vitro translated Nrf1 alone and in combination with MafG-V5 or MafK-V5. Nrf2ϩMafG-V5 or Nrf2ϩMafK-V5 were included as positive control. The Nrf1/Maf and Nrf2/Maf were preincubated at 37°C for 15 min. before incubation with labeled hARE. The band shift experiment was performed at room temperature and analyzed on 5% nondenaturing polyacrylamide gel. The gel was dried and autoradiographed. The asterisk denotes the nonspecific binding of rabbit reticulocyte lysate. In conclusion, small Maf protein homodimers and possibly Nrf2/small Maf heterodimers negatively regulate ARE-mediated expression and antioxidant induction of NQO1 and other detoxifying enzyme genes. Further studies are required to understand the role of positive and negative factors in AREmediated expression and induction of detoxifying enzyme genes.