Oxidative Stress Activates Metal-responsive Transcription Factor-1 Binding Activity OCCUPANCY IN VIVO OF METAL RESPONSE ELEMENTS IN THE METALLOTHIONEIN-I GENE PROMOTER*

Oxidative stress ( tert -butylhydroquinone) rapidly in- duced metallothionein-I gene expression in mouse Hepa cells, and this effect was mediated predominantly through metal response promoter elements in transient transfection assays. In vivo genomic footprinting of the mouse metallothionein-I promoter after treatment of Hepa cells with hydrogen peroxide, tert -butylhydroqui- none, or zinc suggested a rapid increase in occupancy of the metal response elements. More subtle changes also occurred in the constitutive genomic footprint at the composite major late transcription factor/antioxidant response element. This element may, in part, mediate induction by hydrogen peroxide. Electrophoretic mobil- ity shift assays demonstrated a rapid (30 min) increase in the DNA binding activity of metal-responsive tran- scription factor-1 in Hepa cells treated with any of these inducers. In control cells, upstream stimulatory factor binding with the major late transcription factor site, and a nuclear protein complex distinct from AP-1, but specific for the antioxidant response element, were detected. The amounts of these complexes were not altered after these treatments. These studies indicate that met-al-responsive transcription factor-1 plays a role in acti- vating mouse metallothionein-I gene transcription in response to reactive oxygen species. Reactive oxygen species (ROS) 1 are generated as a result of many normal and pathological processes (1), and their gen-eration has been shown to promote a number of deleterious cellular and organismal pathologies including aging, cancer, and cell death (2). Cells synthesize a battery of proteins and other antioxidants gene

Oxidative stress (tert-butylhydroquinone) rapidly induced metallothionein-I gene expression in mouse Hepa cells, and this effect was mediated predominantly through metal response promoter elements in transient transfection assays. In vivo genomic footprinting of the mouse metallothionein-I promoter after treatment of Hepa cells with hydrogen peroxide, tert-butylhydroquinone, or zinc suggested a rapid increase in occupancy of the metal response elements. More subtle changes also occurred in the constitutive genomic footprint at the composite major late transcription factor/antioxidant response element. This element may, in part, mediate induction by hydrogen peroxide. Electrophoretic mobility shift assays demonstrated a rapid (30 min) increase in the DNA binding activity of metal-responsive transcription factor-1 in Hepa cells treated with any of these inducers. In control cells, upstream stimulatory factor binding with the major late transcription factor site, and a nuclear protein complex distinct from AP-1, but specific for the antioxidant response element, were detected. The amounts of these complexes were not altered after these treatments. These studies indicate that metal-responsive transcription factor-1 plays a role in activating mouse metallothionein-I gene transcription in response to reactive oxygen species.
Reactive oxygen species (ROS) 1 are generated as a result of many normal and pathological processes (1), and their generation has been shown to promote a number of deleterious cellular and organismal pathologies including aging, cancer, and cell death (2). Cells synthesize a battery of proteins and other antioxidants whose function is to protect from ROS (1).
Metallothioneins (MT) constitute a family of small cysteinerich heavy metal-binding proteins. MT-I and -II display a broad tissue distribution and were initially characterized for their ability to sequester and protect cells against heavy metals (3). The functional roles for MT include zinc homeostasis (4) and protection from ROS (5)(6)(7).
A hallmark of MT-I and MT-II genes is their rapid transcriptional induction both in vivo and in vitro by agents against which MT protects. Essential to induction by heavy metals are cis-acting sequences termed metal response elements (MRE) (8 -10). The metal-responsive transcription factor, termed MTF-I, that binds to MREs and activates MT gene transcription has been cloned from mouse and human (11,12). The mechanisms by which metals regulate transcription through MTF-1 are poorly understood.
It is well documented that agents that redox cycle, such as paraquat, agents that deplete glutathione, such as diethylmaleate, and inflammatory substances, such as lipopolysaccharide and cytokines, induce ROS production (1) and MT (13,14) in rodents. We found that transcription of the mouse MT-I gene in mouse Hepa cells is rapidly induced by hydrogen peroxide (H 2 O 2 ) (15). In that study transient transfection analysis using the mouse MT-I promoter and isolated transcription elements driving the CAT reporter gene implicated MREs and a composite major late transcription factor/antioxidant response element (MLTF/ARE) in the response of the MT-I gene to ROS. The MLTF binding site has been shown to effect basal transcription from the mouse MT-I promoter (16) and upstream stimulatory factor (USF), a member of the bHLH-bZip protein superfamily, can bind to this site (17,18). The ARE is an element best characterized in the promoters of phase II drug metabolizing enzyme genes and directs response to electrophilic xenobiotics and H 2 O 2 (19). The transcription factor(s) responsible for transactivation through the ARE is not clear. Some reports implicate AP-1 complexes and some a novel factor (19).
In the present study, we employ transient transfection analysis, in vivo footprint analysis, and electrophoretic mobility shift assays (EMSA) to elucidate the transcription factors involved in the transcriptional response of the mouse MT-I gene to ROS. Our results indicate that pro-oxidant conditions activate mouse MTF-1 binding activity and MT-I gene transcription during oxidative stress. concentrated stocks of H 2 O 2 and ZnSO 4 as described (15). tert-Butylhydroquinone (tBHQ) and ␤-naphthoflavone (␤-NF) were dissolved in Me 2 SO as a 1000 times concentrated stock. Hepa cells were transfected using the calcium phosphate precipitation method and reporter gene activity (CAT) was normalized to ␤-gal activity by cotransfection of SV-␤-gal as described (15).
Isolation of RNA and Northern Blot Hybridization-RNA was isolated, size fractionated by formaldehyde agarose gel electrophoresis, transferred to nylon membranes and hybridized with 32 P-labeled mouse MT-I cRNA probe as described (15,20). The probe had a specific activity of approximately 2 ϫ 10 9 cpm/g.
Fusion Gene Construction-The construction of fusion genes used in this study has been described previously (15).
In Vivo Genomic Footprinting-After treatment with the indicated inducer, Hepa cells were exposed to 0.1% dimethyl sulfate (Aldrich) for 2 min at 37°C; genomic DNA was purified and subjected to piperidine (Aldrich) cleavage at positions of methylated guanines as described (21). The cleaved DNA (2 g) was then amplified by ligation-mediated PCR (LM-PCR), using mouse MT-I promoter-specific primers that have been described previously (22,23), except that the following primer (GATAGGCCGTAATATCGGGGAAAGCAC) was used as the third primer in LM-PCR. Reaction conditions were as described (22,23), with the following exceptions. 1) The initial primer extension reaction contained 2 units of Vent Exo DNA polymerase (New England Biolabs, Beverly, MA), and denaturation was conducted for 5 min at 95°C, annealing was for 30 min at 60°C and primer extension was for 10 min at 72°C; 2) during PCR, denaturation, annealing, and primer extension temperatures were as above; 3) the labeling reaction contained 2.3 pmol of end-labeled oligonucleotide (4 -8 ϫ 10 6 cpm/pmol) and was denatured for 3.5 min at 95°C, annealed for 2 min at 66°C, and extended for 10 min at 76°C; and 4) at the end of the labeling cycle, one half of the sample was frozen at Ϫ20°C for later use, and the remainder was fractionated on a 6% DNA sequencing gel. Based on band intensities in each lane of the initial sequencing gel, a second gel was run in which the radioactivity in each lane was "normalized." Samples with low signal were subjected to another cycle of labeling before further analysis.
Quantitation of Footprints-Dried sequencing gels were exposed to a phosphor screen, which was scanned with a Phosphorimager SI, and quantitated using the ImageQuaNT software (Molecular Dynamics, Sunnyvale, CA). To adjust for variations in the amount of radioactivity among the samples, the amount of radioactivity in bases within the footprinted region (base) was compared with that of bases flanking the footprinted region (standard), and the ratio (P base /P standard ) was calculated for each base as described (24). Data were verified for consistency by examining the ratios obtained using several standard bases. Protection or hypersensitivity of individual guanines was determined by subtracting the ratio for that base in the uninduced in vivo control sample, from that in the induced sample, and the difference was divided by the ratio of the in vivo control.
Preparation of Nuclear Extracts-Nuclear extracts were prepared with modifications of the method of Dignam et al. (25). After treatment, cells (1 ϫ 10 8 ) were placed on ice, the medium was removed, and the cells were washed once with ice-cold phosphate-buffered saline. Cells were scraped off the dish in 5 ml of cold phosphate-buffered saline, and collected by centrifugation at 1500 ϫ g for 5 min at 4°C. The cell pellet was resuspended in 5 ml of cell lysis buffer (10 mM HEPES (pH 7.9), 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM DTT, and 0.2 mM phenylmethylsulfonyl fluoride) and immediately centrifuged as above. Cells were resuspended in 2 times the original packed cell volume of cell lysis buffer, allowed to swell on ice for 10 min, and homogenized with 10 strokes of a Dounce homogenizer (B pestle). Nuclei were collected by centrifugation at 3300 ϫ g for 15 min at 4°C, the supernatant was removed and the nuclei were suspended, using 6 strokes of a Teflon-glass homogenizer, in 3 volumes (about 750 l) of nuclear extraction buffer (20 mM HEPES (pH 7.9), 1.5 mM MgCl 2 , 400 mM KCl, 0.5 mM DTT, 0.2 mM phenylmethylsulfonyl fluoride, and 25% glycerol). The nuclear suspension was stirred for 30 min on ice, and then centrifuged at 89,000 ϫ g for 30 min at 4°C. The supernatant was collected, and concentrated in a Micro-con3 microconcentrator (Amicon Inc., Beverly, MA) by centrifugation at 12,000 ϫ g for 3 h at 4°C. The concentrated extract was diluted in half with diluent buffer (extraction buffer without KCl) and frozen in aliquots at Ϫ80°C. Proteins were determined using a Bradford protein assay (Bio-Rad) with rabbit IgG as a standard. A nuclear extract from TPA-treated NIH 3T3 cells, and antibodies against the c-jun family and USF (recognizes both USF-1 and USF-2; Ref. 26) were purchased from Santa Cruz Biochemicals (Santa Cruz, CA).
EMSA-Nuclear proteins (5-10 g) were incubated in binding reaction buffer containing 12 mM HEPES (pH 7.9), 60 mM KCl, 0.5 mM DTT, 12% glycerol, 5 mM MgCl 2 , 0.2 g of dI-dC/g of protein, 2-4 fmol of end-labeled double-stranded oligonucleotide (5000 cpm/fmol) (27), and where indicated 60 M ZnSO 4 , in a total volume of 20 l for, unless otherwise stated, 20 min at room temperature. In binding site competition experiments, binding reactions also contained a 250-fold molar excess of the indicated unlabeled oligonucleotide. In antibody competition experiments, the indicated antiserum (1 g) was added to the complete binding reaction and the mixture was incubated for 20 min at room temperature. The oligonucleotide sequences were as follows (bold bases denote the functional core, and underlined bases denote mutations).
Protein-DNA complexes were separated electrophoretically at 4°C in a 4% polyacrylamide gel (acrylamide:bisacrylamide/80:1) at 15 V/cm. The gel was polymerized in, and the running buffer consisted of 0.19 M glycine (pH 8.5), 25 mM Tris, 0.5 mM EDTA, and where indicated 60 M ZnSO 4 . After electrophoresis, the gel was dried and labeled complexes were detected by autoradiography.
Proteolytic EMSA-The effects of digestion with thrombin or trypsin on MRE-s binding activity were examined as follows. Nuclear proteins (30 g) from H 2 O 2 -and ZnSO 4 -treated Hepa cells (1 h after treatment) and recombinant MTF-1 (0.25 l of a 50-l reaction described below) were added to the complete binding reaction containing labeled MRE-s and incubated at 4°C for 20 min to allow DNA binding. In some binding reactions, a solution of 2.5% trypsin (Life Technologies, Inc.) diluted 1:100 in Hanks' balanced salt solution was added (1 l/20 l binding reaction), and they were incubated for 15 min at 37°C. Soybean trypsin inhibitor (Life Technologies, Inc.; 100 g in 1 l) was then added, and the samples were analyzed by gel electrophoresis. In other binding reactions, thrombin (Boehringer Mannheim) was added (8 ϫ 10 Ϫ3 units/2 l) and the mixtures were incubated at room temperature for 15 min. These samples were immediately chilled on ice and analyzed by gel electrophoresis. Specificity of the binding was confirmed by competition with excess unlabeled MRE-s, and protein-DNA complexes were separated electrophoretically, as described above.
In Vitro Transcription/Translation of MTF-1-A cDNA clone for mouse MTF-1 was generated using reverse transcriptase-PCR from mouse liver RNA with the following primers.

SEQUENCES 9 AND 10
Primers contained XbaI or ClaI sites (bold bases), plus two terminal stabilizing bases, for ease of cloning and they directed amplification of the complete coding sequence of mouse MTF-1. Reverse transcriptase-PCR was conducted as described (33) using mouse liver total RNA. The MTF-1 cDNA was cloned between the XbaI-ClaI sites of pGEM7 (Promega Biotech, Madison, WI) and sequenced using Sequenase (Stratagene). MTF-1 was synthesized in vitro using a TnT coupled reticulocyte lysate system (Promega Biotech), containing 1 g of the MTF-1 plasmid, according to the manufacturer's suggestions.

Activation of Transcription from the MT-I Promoter by tBHQ Involves
MREs-We previously reported that transcriptional induction of the mouse MT-I gene by H 2 O 2 requires the promoter region between Ϫ153 and Ϫ42, relative to the transcription start point (15). This region contains five MREs and a composite MLTF(USF)/ARE. We also reported that four or five tandem copies of these elements independently conferred response to H 2 O 2 on a reporter gene. In contrast, only MREs conferred response to zinc. To further examine the regulation of MT-I gene expression by oxidative stress, the effects of the planar aromatic compound ␤-NF and the phenolic antioxidant tBHQ on MT-I gene expression were determined. ␤-NF is a bifunctional inducer that activates both phase I (e.g. CYP1a1 P-450) and phase II (e.g. glutathione S-transferase Ya) detoxification systems, whereas tBHQ is monofunctional and induces phase II genes (34 -36). ␤-NF must be metabolized to yield an electrophilic compound which then activates gene expression (37,38). In contrast, tBHQ undergoes auto-oxidation to form a redox-active semiquinone anion radical (39,40).
Incubation of Hepa cells in medium containing ␤-NF (50 or 100 M) did not induce MT-I mRNA, which suggested that this compound may not be metabolized by these Hepa cells. Thus, effects of ␤-NF were not further examined. In contrast, tBHQ caused a concentration-dependent induction of MT-I mRNA ( Fig. 1), and 400 M tBHQ caused an approximately 40-fold increase in MT-I mRNA by 8 h. Treatment of cells with tBHQ also dramatically induced heme oxygenase-1 mRNA, but did not effect the levels of ␤-actin or Cu,Zn-superoxide dismutase mRNAs (data not shown). tBHQ also induces MT-I mRNA in NIH 3T3 and L929 cells (data not shown).
Transient transfection assays were performed in Hepa cells to delineate the promoter elements involved in response of the MT-I gene to tBHQ (Fig. 2). A construct containing 153 base pairs of the MT-I promoter fused to the chloramphenicol acetyltransferase (CAT) gene was responsive to tBHQ, and removal of the region distal to Ϫ43 completely abolished response. Response of the 153-base pair promoter to tBHQ was submaximal at 200 M and apparently maximal at 400 M tBHQ. The higher concentration was moderately cytotoxic in the transfected cells. Selective deletion of the MLTF/ARE (Ϫ100 to Ϫ89) in Ϫ153-CAT only modestly reduced response to a tBHQ (p Ͻ 0.04; 200 M tBHQ). In contrast, deletion of this region reduces response to H 2 O 2 by 50% (15). These results suggested that response of the MT-I promoter to tBHQ is mediated predominantly by MREs. To further examine this possibility, multiple copies of MRE-d (MRE-dЈ 5 ) or MLTF/ARE (MLTF/ARE 4 ) were analyzed for the ability to confer response to tBHQ on a minimal promoter (Ϫ42 base pairs of the mouse MT-I promoter) driving CAT. Treatment of Hepa cells with 200 M tBHQ resulted in 4.4-fold increase in CAT activity from the MRE-dЈ 5 -CAT construct, but did not result in significant induction of the MLTF/ARE 4 -CAT construct. Treatment with tBHQ did not increase CAT activity from an RSV-CAT vector or from the minimal MT promoter construct (Ϫ42). These results demonstrate the mouse MT-I gene is transcriptionally regulated by tBHQ, consistent with a role for ROS in regulating MT gene expression, and suggest that this induction is mediated by MREs.
In Vivo Genomic Footprinting of the MT-I Promoter Reveals That Occupancy of the MREs Is Induced in Response to H 2 O 2 , tBHQ, and Zinc-In vivo genomic footprint analysis was accomplished by LM-PCR of bases Ϫ200 to Ϫ30 in the MT-I promoter (Figs. 3 and 4). Cells, before or after treatment with an inducer, were incubated with dimethyl sulfate, and genomic DNA was isolated, cleaved with piperidine, and MT-I promoter fragments were specifically amplified. Guanine residues involved in protein-DNA interactions were visualized as either less intense (protected) or more intense (hypersensitive) compared with invariant G residues in the promoter, and by comparison with DNA from untreated control cells. Naked genomic DNA was methylated and similarly amplified to provide a G-ladder and to reveal constitutive interactions in the MT-I promoter in control cells. Footprint experiments with each inducer were conducted several times, and the results were reproducible.
In control cells, a strong constitutive footprint was evident on the guanine-rich sense strand (Fig. 3A) over an Sp1 binding Ϫ42 minimal MT-I promoter. After a recovery period of 28 h, the transfected cells were incubated for 8 h in medium containing the indicated concentration of tBHQ. Cells were harvested, and cell lysates were assayed for CAT and ␤-gal activities (15). CAT activity was normalized to ␤-gal activity in each cell lysate, and the fold induction of CAT was calculated as the ratio of the normalized CAT activity in treated cells to that in control cells. Data represent the mean Ϯ S.D. of three independent transfection assays for each construct. * indicates p Ͻ 0.04 (Student's t test).
site (Ϫ187 to Ϫ179); in addition, there was a constitutive footprint over the MLTF/ARE (Figs. 3, A and B, and 4c). Similar results have been reported previously (41). In contrast, little constitutive footprinting of MRE core sequences, with the exception of a weak footprint over MRE-d, was detected (Figs. 3,  A and B, and 4b). An apparently stronger constitutive footprint over MRE-d was reported previously (41), which may reflect the cell type or culture conditions in those experiments. Please note that the histograms do not indicate constitutive footprints, as the data were derived by comparisons of footprints from induced cells with those from uninduced cells (in vivo control).
After treatment with H 2 O 2 , tBHQ, or zinc, footprints were rapidly induced over five of the MREs (Fig. 3, A and B). Footprinted regions at the time of maximal occupancy are shown enlarged and quantitated for MRE-c and MRE-d (Fig. 4, panels  a and b), because these two MREs were more strongly footprinted. In contrast, the footprint over MRE-e was weak. It is noteworthy that G residues within each MRE core sequence (TGCRCnC) that are functionally essential (28) were protected from methylation. Remarkably, with all three inducers the footprint patterns were essentially identical over the MREs, but the temporal appearance of these footprints differed. Footprints induced by H 2 O 2 appeared within 30 min of treatment, but were absent by 5 h. This is consistent with the rapid, transient induction of MT-I gene transcription by H 2 O 2 (15). In contrast, the MRE footprints induced by 200 M tBHQ were detected by 1 h, but were maintained for at least 5 h. These footprints were notably less intense than those induced by H 2 O 2 or zinc. However, 400 M tBHQ, which maximally induced MT mRNA (Fig. 1), induced strong and prolonged footprints over these MREs. It is certainly conceivable that MRE occupancy and subsequent transcriptional activation of the MT gene are both modulated.
In addition to footprints induced over MREs, all of these treatments also induced footprints over the MLTF/ARE region (Fig. 3, A and B; Fig. 4, panel c). Footprint patterns with the various inducers were qualitatively similar, but quantitatively distinct. Major changes in guanine methylation were noted in residues immediately upstream and within the MLTF binding site and immediately downstream of the ARE. The footprint , or tBHQ. After the indicated treatment period, Hepa cells were incubated in medium containing 0.1% dimethyl sulfate to methylate guanine residues and genomic DNA was then purified. To generate a G-ladder, purified MT-I promoter DNA was methylated in vitro. Methylated DNA was cleaved with piperidine, and the MT-I promoter region from Ϫ200 to Ϫ30 of the sense strand (A) and antisense strand (B) were specifically amplified using LM-PCR. LM-PCR products were separated on a 6% sequencing gel and detected by autoradiography. Locations of known regulatory elements in the MT-I promoter are indicated. Sp1, Sp1 binding site; MRE, metal regulatory elements (-a, -b, -c, -d, and -e); MLTF/ARE, overlapping major late transcription factor binding site and antioxidant response element. Constitutive footprints can be detected in the 0 h control. over the MLTF/ARE induced by H 2 O 2 was transient, while those induced by tBHQ and zinc were prolonged. It is somewhat surprising that zinc induced a footprint over the MLTF/ ARE; however, a similar result was reported previously (41). This region does not contain an MRE core sequence, and an oligonucleotide with this sequence does not interact with MTF-1 (see below). These results revealed little evidence of protein interactions with guanine residues within the ARE functional core (GTGACnnnGC) (31,42).
H 2 O 2 , tBHQ, and Zinc Each Rapidly Induce MRE Binding Activity in Hepa Cells-EMSA was used to detect MRE binding activity in nuclear extracts from Hepa cells. MRE-d is the best studied MRE in the mouse MT-I promoter (8,29,43), but it is bound by two transcription factors, Sp1 and MTF-1 (43). To reduce the complexity of analyzing MTF-1 binding to the MRE, we employed a consensus MRE oligonucleotide (MRE-s) that is a strong MTF-1 binding site, but does not bind Sp1 (11).
EMSA analysis using nuclear extracts obtained from Hepa cells treated for 45 min with H 2 O 2 revealed an 8-fold increase in MRE-s complex formed (Fig. 5A). Formation of this complex was effectively inhibited by a 250-fold molar excess of unlabeled MRE-s or MRE-d oligonucleotides, but not by a similar molar excess of a mutant MRE in which the bases that constitute the MRE functional core were mutated (Fig. 5, A and B). Furthermore, Sp1 binding site and MLTF/ARE oligonucleotides did not compete for MRE-s complex formation. Thus, formation of this complex displayed similar DNA sequence specificity to recombinant MTF-1 (11,43) and was rapidly increased in amount during oxidative stress. This complex has the same electrophoretic mobility shift as that formed between MRE-s and mouse MTF-1 synthesized in vitro in a coupled transcription/translation system (Fig. 6).
Antibodies that supershift MTF-1 are not available; therefore, to provide further evidence that the MRE-s binding activity in nuclear extracts from Hepa cells induced with H 2 O 2 is MTF-1, a proteolytic EMSA was performed in which the proteinase-resistant MRE-s complexes from Hepa cells were compared with those from recombinant MTF-1 (Fig. 6, A and B). Trypsin digestion of preformed MRE-s complexes resulted in the rapid appearance of a predominant smaller complex that displayed the same mobility shift using Hepa cell nuclear extracts and recombinant MTF-1 (Fig. 6A). Thrombin digestion of preformed MRE-s complexes resulted in the appearance of a cluster of three predominant smaller complexes that displayed the same mobility shifts using Hepa cell nuclear extracts and recombinant MTF-1 (Fig. 6B). Identical results were obtained using nuclear extracts from Hepa cells treated with zinc. A relatively weak signal in the proteinase digested nuclear extract lanes (Fig. 6, A and B, lane 4) was noted in both experiments, but we found that mixing those extracts with recombinant MTF-1 before EMSA also resulted in a weakened signal (data not shown). Proteinase digestion of these nuclear extracts and of recombinant MTF-1 in the absence of MRE-s led to the rapid and complete loss of subsequent MRE-s binding activity (data not shown). These results strongly suggest that the major MRE-s complex in Hepa cell nuclear extracts during oxidative stress contains MTF-1 (Fig. 6). Saturation analysis of MRE-s binding activity in nuclear extracts from control and H 2 O 2 -treated Hepa cells suggested that differences in MTF-1 activity between these extracts reflected increased binding activity, without an apparent change in the affinity of MTF-1 binding (data not shown). Although MTF-1 binding activity was increased in nuclear extracts from H 2 O 2 -treated cells, the binding activities of Sp1 (Fig. 7) and USF (Figs. 8 and 9A) were not.
During preparation of Hepa cell nuclear extracts (25), EDTA was omitted from all buffers to avoid potentially removing zinc from MTF-1. During the EMSA shown in Fig. 5, zinc (60 M) was included in the binding reaction and electrophoresis buffer, as described previously (11). However, exogenous zinc The in vitro translated MTF-1 was activated to bind DNA by the addition of 30 M zinc to the binding reaction, followed by incubation at 37°C for 15 min before EMSA. In the absence of exogenous zinc, as well as in the "coupled" reaction lacking MTF-1 cDNA template, no specific MRE-s binding activity (left side arrow) was detected (data not shown). Specificity of binding was confirmed by competition with excess unlabeled MRE-s as described above (data not shown). A, the effects of trypsin on MRE-s-protein complexes with recombinant MTF-1 (lanes 1 and 3) and in nuclear extracts from H 2 O 2treated Hepa cells (lanes 2 and 4) were examined. Nuclear proteins (30 g) and recombinant MTF-1 in the complete binding reaction were incubated with trypsin (described under "Experimental Procedures") for 15 min at 37°C. Soybean trypsin inhibitor was then added, and the samples were analyzed by gel electrophoresis. The right side arrow indicates the trypsin-resistant specific MRE-s protein complex. B, thrombin was added to binding reactions as in A, which were incubated at room temperature for 15 min immediately before gel electrophoresis. Lanes 1 and 3, recombinant MTF-1; lanes 2 and 4, Hepa nuclear extract. The right side arrows indicate thrombin-resistant specific MRE-s protein complexes.
was not required for MRE-s complex formation in these nuclear extracts (Fig. 7), and it was omitted in subsequent EMSA. Under these experimental conditions, nuclear extracts from H 2 O 2 -treated (30 min) cells had increased MTF-1 binding activity, as did those from cells treated with zinc. Furthermore, tBHQ-treated (200 M) cells also displayed increased MTF-1 binding activity. Although the amount of complex formed was much less than that in extracts from zinc-or H 2 O 2 -treated cells, it was noted that 400 M tBHQ induced much more MTF-1 binding activity (data not shown). The electrophoretic mobility shift of the MRE-s complex from zinc-, tBHQ-, and H 2 O 2 -treated cells was identical. In Fig. 7, MTF-1 and Sp1 binding activities in the same extracts were compared to confirm the specificity of activation of MTF-1 by zinc and oxidative stress, and as an internal control for extract loading and qual-ity. The HSV Sp1 binding site was used in this analysis. This oligonucleotide has two Sp1 binding sites, and thus gives rise to a complex shift pattern as reported previously (43).
The Binding of USF and Another Factor to the MLTF(USF)/ ARE Composite Element Are Not Rapidly Increased by H 2 O 2 or tBHQ Treatment-Two complexes were detected by EMSA using the mouse MLTF/ARE oligonucleotide and Hepa cell nuclear extracts. In repeated experiments, these complexes did not differ in amount or mobility among nuclear extracts from control and treated cells (Figs. 8 and 9). These constitutive Nuclear proteins (10 g) were extracted and EMSA was performed using 32 Plabeled MRE-s and Sp1 oligonucleotides, as indicated. EMSA was performed as described in the legend to Fig. 5, except that exogenous zinc was omitted from all buffers. These results were reproduced in three separate experiments.

FIG. 8. Detection of MLTF/ARE binding activities in nuclear extracts from control and H 2 O 2 -treated Hepa cells.
Hepa cells were incubated for 30 min in medium containing H 2 O 2 (2.5 mM). Nuclei from control and treated cells were then isolated and nuclear extracts prepared. EMSA was performed using a 32 P-labeled MLTF/ARE oligonucleotide. The arrows point to two specific protein-DNA complexes. Where indicated, the binding reaction also contained a 250-fold molar excess of the following unlabeled competitor oligonucleotide: MLTF/ ARE, MLTF/mutARE in which the terminal GC nucleotides in the ARE consensus sequence were mutated, and mutMLTF/ARE in which the first three bases in the MLTF binding site were mutated. MLTF Ab denotes an reaction in which antibody against USF-1 and -2 was added to the binding reaction before EMSA as detailed under "Experimental Procedures."

FIG. 9. Examination of AP-1 binding activity in nuclear extracts from TPA-treated NIH 3T3 cells and from Hepa cells treated with H 2 O 2 or tBHQ. NIH 3T3 cells incubated in medium
containing phorbol ester (10 ng of TPA/ml) for 1 h before nuclear extract preparation. Nuclear extracts were prepared from Hepa cells incubated in medium containing H 2 O 2 (2.5 mM) or tBHQ (200 M) for 30 min. A, EMSA was performed using a 32 P-labeled MLTF/ARE oligonucleotide. B, EMSA was performed using a 32 P-labeled TRE oligonucleotide. AP-1 Ab denotes an antibody competition experiment in which antibody against the c-jun family was added to the binding reaction before EMSA, as described under "Experimental Procedures." Arrows point to specific protein-DNA complexes.
binding activities were sequence-specific, and complex formation was inhibited by a 250-fold molar excess of unlabeled MLTF/ARE oligonucleotide (Fig. 8). Mutation of the two functionally important (31,42) terminal bases (GC) in the ARE consensus sequence (GTGACnnnGC) in the competitor oligonucleotide (MLTF/mutARE) completely inhibited formation of the faster migrating complex, but had less of an inhibitory effect on formation of the slower migrating complex. In contrast, mutation of the functionally important (16) first three bases of the MLTF binding site (CGCGTGAC) in the competitor oligonucleotide completely inhibited formation of the slower migrating complex, but had no effect on the faster migrating complex. These results suggested that two factors bind to the MLTF/ARE. The faster migrating complex contains USF, whereas the slower migrating complex likely contains AREspecific binding activity. That the faster migrating complex contained USF was confirmed by antibody competition analysis (Fig. 8). In these experiments, the USF antiserum did not alter the ARE-specific complex or Sp1 binding activity (data not shown). The identity of the factors that bind to the ARE is investigated below.

AP-1 Is Not a Major ARE Binding Activity and Is Not Rapidly Induced in H 2 O 2 -or tBHQ-treated Hepa
Cells-AP-1 has been shown to bind to AREs (19), but several studies suggest that ARE function is not mediated by nor dependent on AP-1 (31,35,36). To examine AP-1 activity in Hepa cell extracts, an antibody against the c-jun family, that blocks binding of c-juncontaining AP-1 complexes, was used in EMSA. Nuclear proteins from TPA-treated NIH 3T3 cells served as a positive control for AP-1 activity, and using the MLTF/ARE oligonucleotide, formation of one major complex was detected in this extract. The AP-1 antibody inhibited formation of this complex by greater than 90% (Fig. 9A). Thus, Jun-containing AP-1 complexes can also bind to the mouse MT-I MLTF/ARE. However, this AP-1 antibody did not significantly prevent the formation of either of the two MLTF/ARE-specific complexes in nuclear extracts isolated from control, H 2 O 2 -, or tBHQ-treated Hepa cells. In addition, the AP-1 antibody did not effect Sp1 binding activity in these extracts (data not shown).
To further examine AP-1 activity in Hepa cells during oxidative stress, nuclear extracts from H 2 O 2 -or tBHQ-treated cells were used in EMSA with a consensus TRE oligonucleotide (Fig. 9B). As a positive control, a single specific complex was readily detected in nuclear proteins from TPA-treated NIH 3T3 and the AP-1 antibody inhibited greater that 80% of this binding. In contrast, only a low amount of a TRE-specific complex was detected in nuclear extracts from H 2 O 2 -and tBHQ-treated Hepa cells, and formation of this complex was reduced 30 -50% by the Ap-1 antibody. The amount of this TRE complex was not increased after H 2 O 2 -or tBHQ treatment, and cross-competition experiments demonstrated that excess unlabeled TRE oligonucleotide did not efficiently inhibit formation of the MLTF/ ARE slow migrating complex. Thus, AP-1 activity is not rapidly induced during oxidative stress in Hepa cells, and AP-1 does not represent a major component of the MLTF/ARE binding activity in these cells. DISCUSSION MT protects cells against oxidative stress. In yeast, mammalian MT can functionally replace superoxide dismutase (5), and mouse cells with targeted disruption of the MT-I and -II genes are more susceptible to organic peroxides than are cells with a single or two wild-type alleles (6). Furthermore, overexpression of MT protects mammalian cells against oxidative stress and can dramatically reduce the level of intracellular oxygen radicals (7). In vitro, MT is an efficient scavenger of hydroxyl radicals (44). Consistent with this hypothesized protective function, mouse MT gene transcription is rapidly and dramatically induced by oxidative stress (H 2 O 2 , menadione, tBHQ) in Hepa cells (15), as well as by tBHQ in AR42J, NIH3T3, and L929 cells. 2 Recently, H 2 O 2 has been reported to induce MT in human retinal pigment epithelial cells (45). Thus, induction of MT genes by oxidative stress occurs in many cell types.
Herein, we further explored the mechanisms of regulation of MT-I gene transcription during oxidative stress. A number of convergent experimental approaches yielded results consistent with the conclusion that induction of mouse MT-I gene transcription by tBHQ, as well as by H 2 O 2 (15), is mediated, in part, by MREs in the proximal promoter region. Interestingly, the heme oxygenase-I (46) and several acute-phase genes are also induced by metals (47), which may suggest that MREs are important promoter elements in several protective genes. In vivo genomic footprint analysis and EMSA suggest that MT-I gene MRE activity is modulated during oxidative stress by the binding of the transcription factor MTF-1. Manipulation of MTF-1 expression by targeted deletion of both genes in ES cells (48) or by expression of antisense MTF-1 (49) eliminates metal responsiveness of the MT-I gene. Thus, this transcription factor plays a key role in regulating MT gene expression in response to metal ions and oxidative stress.
Mouse MTF-1 contains six zinc fingers and three different putative transactivation domains (50). Mechanisms of activation of MTF-1 are not well understood. It has been postulated that this transcription factor is constitutively active in the absence of a metal-sensitive inhibitor (11,49), although such an inhibitor has not been isolated. Another model for MTF-1 activation, proposed previously (11), suggests that MTF-1 has a lower affinity for zinc than other transcription factors and that it is reversibly activated in response to free zinc levels in the cell. Our finding that mouse MTF-1 DNA binding activity is rapidly activated (10-fold in 30 min) in Hepa cells treated with zinc or oxidative stress-inducing agents, and the distinct similarities in the genomic footprints of the MT-I promoter induced by zinc and oxidative stress are consistent with a central role for free zinc in the activation of this transcription factor during oxidative stress. Oxidative stress may cause the release of bound intracellular zinc by oxidation of thiols and lipids. Oxidized glutathione, which is increased during oxidative stress, has been shown to mobilize zinc from metallothionein (51).
Data presented herein do not allow one to distinguish between these proposed models for mechanisms of activation of MTF-1 by zinc. Furthermore, other mechanisms of activation of MTF-1 are plausible. If MTF-1 binding is controlled by an inhibitory molecule, a transcription factor that might serve as a paradigm for the control of MTF-1 is NF-B. This transcription factor is regulated by a family of inhibitory proteins, IB, that dissociate in response to a number of different signals including oxidative stress (52). It is also possible that MTF-1 is activated in a manner similar to that of the bacterial transcription factor OxyR (53). OxyR apparently becomes oxidized and activated in response to an ROS-induced shift in the redox environment within the cell. Once MTF-1 is activated to bind to DNA, its transactivating potential may also be regulated. Oxidative stresses can enhance tyrosine phosphorylation of the epidermal growth factor receptor (54), phosphorylation of IB (55,56), and Jun kinase (mitogen-activated protein kinase) activity (57). Mouse MTF-1 contains several potential sites of phosphorylation (50), but no studies of the phosphorylation state of the protein have been reported.
Evidence from other investigators suggests the paramount importance of the ARE in response to oxidative stress (19), but deletion of the MLTF/ARE in the MT-I promoter reduced but did not eliminate response to H 2 O 2 (15), and had little effect on response to tBHQ. We obtained similar results in studies of the chicken MT promoter. 3 Furthermore, the MLTF/ARE element alone (four copies) can direct response of a minimal promoter to H 2 O 2 (15), but not to tBHQ in transient transfection assays. Transcription of the mouse glutathione S-transferase Ya gene during oxidative stress is controlled by two adjacent electrophile response elements that are identical in core sequence to the ARE and similar to AP-1 binding sites. These elements cooperate with the transcription factor Ets to mediate response to oxidative stress (58). Thus, interactions between MT-I promoter elements may modulate activity of the MLTF/ARE. Although the mouse MT-I gene MLTF/ARE was found to interact with AP-1, this was a minor binding activity in Hepa cell nuclear extracts, and it was not increased under these experimental conditions. However, the demonstration that AP-1 can interact with the MLTF/ARE suggests that it could potentially be involved in regulating mouse MT-I gene expression. AP-1 is thought to regulate mouse glutathione S-transferase Ya (58,59) and human quinone reductase (19) gene expression via induced binding to the ARE/TRE. In addition to the well documented regulation of AP-1 at the transcriptional level, AP-1 binding is redox regulated by Ref-1 (60). Furthermore, its ability to stimulate gene transcription during oxidative stress is regulated by phosphorylation (57). Thus, the transactivating potential of preexisting AP-1⅐TRE complexes can be modulated. In this light, it is also important to note that electrophilic quinones, in particular tBHQ, have recently been shown to preferentially induce Fra-1, which forms an inactive heterodimer with Jun, and thus antagonizes AP-1 activity (36). Induction of Fra-1 by tBHQ in Hepa cells might explain the inability of this compound to transactivate through the MT-I MLTF/ARE. H 2 O 2 , on the other hand, is a poor inducer of Fra-1 (61).
In contrast to AP-1 interactions with the ARE, the constitutive binding of as yet uncharacterized proteins to the ARE has been demonstrated (19,31). Consistent with those studies, EMSA demonstrated the binding of USF and a predominant ARE-specific non-AP-1 factor to the mouse MT-I gene MLTF/ ARE. A functional ARE from the rat glutathione S-transferase Ya gene does not bind AP-1, but if it is mutated to an ARE-TRE, it can bind AP-1 yet retain inducibility by oxidative stress (31). Thus, protein interactions with the ARE are complex in several genes, and the functional significance of the ARE may depend upon its context within a promoter, flanking nucleotides, inducer, and/or the cell-type being analyzed. Due to the complex regulation of factors that interact with AREs, understanding the nature of the transcriptional control through the MLTF/ ARE will require further experimentation using alternative approaches.