Activation of Maf/AP-1 Repressor Bach2 by Oxidative Stress Promotes Apoptosis and Its Interaction with Promyelocytic Leukemia Nuclear Bodies*

The oxidative stress response operates by inducing the expression of genes that counteract the stress. We show here that the oxidative stress-responsive transcription factor Bach2 is a generic inhibitor of gene expression directed by the 12-O-tetradecanoylphorbol-13-acetate response element, the Maf recognition element, and the antioxidant-responsive element. The Bach2-enhanced green fluorescent protein bicistronic retrovirus was used to monitor the fate of Bach2-expressing cells at the single cell level. Bach2 exerted an inhibitory effect on NIH3T3 cell proliferation and caused massive apoptosis upon mild oxidative stress in both NIH3T3 and Raji B-lymphoid cells. Interestingly, Bach1, a highly homologous protein, could not induce cell death, demonstrating the specificity for the apoptosis induction. Although both oxidative stress and leptomycin B, an inhibitor of nuclear export, induce nuclear accumulation of Bach2, the leptomycin B-induced nuclear accumulation of Bach2 was not sufficient to elicit apoptosis. Upon oxidative stress, Bach2 formed nuclear foci that associated with promyelocytic leukemia nuclear bodies. Our results suggest that Bach2 constitutes a cell lineage-specific system that couples oxidative stress and cell death and that inhibition of 12-O-tetradecanoylphorbol-13-acetate response element, the Maf recognition element, and the antioxidant-responsive element upon oxidative stress may be critical determinants for apoptosis.

The oxidative stress response operates by inducing the expression of genes that counteract the stress. We show here that the oxidative stress-responsive transcription factor Bach2 is a generic inhibitor of gene expression directed by the 12-O-tetradecanoylphorbol-13-acetate response element, the Maf recognition element, and the antioxidant-responsive element. The Bach2-enhanced green fluorescent protein bicistronic retrovirus was used to monitor the fate of Bach2-expressing cells at the single cell level. Bach2 exerted an inhibitory effect on NIH3T3 cell proliferation and caused massive apoptosis upon mild oxidative stress in both NIH3T3 and Raji B-lymphoid cells. Interestingly, Bach1, a highly homologous protein, could not induce cell death, demonstrating the specificity for the apoptosis induction. Although both oxidative stress and leptomycin B, an inhibitor of nuclear export, induce nuclear accumulation of Bach2, the leptomycin B-induced nuclear accumulation of Bach2 was not sufficient to elicit apoptosis. Upon oxidative stress, Bach2 formed nuclear foci that associated with promyelocytic leukemia nuclear bodies. Our results suggest that Bach2 constitutes a cell lineage-specific system that couples oxidative stress and cell death and that inhibition of 12-O-tetradecanoylphorbol-13-acetate response element, the Maf recognition element, and the antioxidant-responsive element upon oxidative stress may be critical determinants for apoptosis.
Oxidative stress is characterized by high cellular levels of reactive oxygen species (ROS) 1 that have damaging effects on cellular components and so trigger defensive responses by the cell. In addition, ROS have been shown to act as intracellular second messengers for certain cytokines and growth factors (1)(2)(3). Several observations suggest that ROS may mediate apoptosis. Apoptosis also appears to be induced by p53 in part by the transcriptional induction of redox-related genes that supposedly lead to increased levels of ROS (4). Although the importance of ROS in the execution of apoptosis is still controversial, they are likely to be involved as second messengers in the apoptotic signal transduction pathway. For example, apoptosis signal-regulating kinase 1 is activated by hydrogen peroxide to induce apoptosis (5). However, the specific molecular targets of ROS in apoptosis signaling are still largely unknown, leaving the mechanisms responsible for ROS-induced cell death unclear. We show here that the transcription repressor Bach2 induces cell death upon oxidative stress.
The dimeric AP-1 transcription factor complexes control cell proliferation, differentiation, and apoptosis by regulating gene expression upon exposure to various stimuli (6). Originally, four different subclasses of basic leucine zipper (bZip) dimers were identified (i.e. AP-1, ATF/cAMP response element-binding protein, CCAAT/enhancer-binding protein, and Maf) based upon the DNA sequences they bind (7). Dimers of the oncoprotein, v-Maf, and its related factors bind to Maf recognition elements or MARE (8,9). Because MARE embeds a 12-Otetradecanoylphorbol-13-acetate (TPA) response element (TRE), it is also a binding site for the Jun and Fos family members (8). In addition, the small Maf proteins (i.e. MafF, MafG, and MafK) bind to versions of MARE and activate transcription by forming heterodimers with the CNC (Cap'n'Color) family of bZip proteins including the hematopoietic transcription factor NF-E2 p45 and its related factors Nrf1, Nrf2, and Nrf3 (10 -15).
In a yin-yang scenario typical in biological systems, transcription activators are often opposed by repressors that target the same DNA elements. This is also the case for the Maf-CNC system. Like NF-E2 p45 and others, Bach1 and Bach2 form heterodimers with the small Maf proteins to bind to MARE. However, the resulting heterodimers repress MARE-dependent transcription (16). Furthermore, homodimers of the small Maf and small Maf/Fos heterodimers also function as repressors (9,17,18). Thus, there appears to be a complex transcriptional antagonism regulating MARE-dependent gene expression and its cross-talk with the TRE-dependent system. The presence of a vast number of combinations of MARE-binding factors, including those both positive and negative, suggests that MARE is involved in various biological processes and that deregulation of MARE-mediated gene expression underlies cell transformation by Maf and AP-1 oncoproteins. One clue to understanding the biological function of MARE is its close resemblance to the antioxidant-responsive element, ARE.
Cells protect themselves against ROS with various defensive mechanisms including compounds such as glutathione and metallothioneins and detoxifying enzymes like glutathione Stransferase and heme oxygenase 1. The inducible expression of these genes is mediated at least in part by ARE. When compared with other AP-1-binding DNA sequences, ARE is related most closely to MARE (8, 19 -21). Conversely, MARE confers oxidative stress inducibility upon reporter genes in transfection assays, suggesting that MARE is endowed with an ARElike activity (22,23). In vitro, ARE is bound by various AP-1related transcription factors. Gene targeting experiments have shown that Nrf1 and Nrf2 play important roles in the inducible expression of genes with ARE (24 -27).
Oxidative stress regulates subcellular localization of Bach2 through an oxidative stress-sensitive conditional nuclear export (23), suggesting a role for Bach2 as a biological switch that processes and transduces oxidative stress into the nucleus. In cultured cells, Bach2 is localized in the cytoplasm through its C-terminal evolutionarily conserved cytoplasmic localization signal (CLS). The CLS directs leptomycin B-sensitive and Crm1/Exportin1-dependent nuclear export. However, CLS is distinct from the conventional leucine-rich class of nuclear export signal in that oxidative stress aborts the CLS activity and induces nuclear accumulation of Bach2 (23). Regulation of transcription factor localization is a key aspect of many signal transduction pathways. Considering its regulation, Bach2 appears to play an important role in the oxidative stress response in mammalian cells. To understand the biological function of Bach2 in the oxidative stress response, we first asked whether Bach2 is a specific repressor of MARE or a generic repressor of TRE, MARE, and ARE. We carried out Bach2 overexpression studies and examined its effect and dynamics at the level of a single cell during oxidative stress. Finally, we show dynamic interaction of Bach2 with promyelocytic leukemia (PML) nuclear bodies.

MATERIALS AND METHODS
Plasmids-Retroviral vectors were constructed using the MSCV/ IRES-EGFP vector. FLAG epitope-tagged Bach2, Bach2⌬BTB, Bach2⌬Zip, and Bach1 cDNAs were constructed as follows. An expression plasmid pcDNA3.1FLAG for FLAG fusion proteins was constructed by inserting FLAG-coding cDNA between the NotI and BamHI sites of pcDNA3.1B (Invitrogen). The FLAG cDNA was 5Ј-GGCGGCCGCTCT-AGACCATGGACTACAAGGACGACGATGACAAGGGATCC-3Ј (the NotI and BamHI sites are underlined). An entire open reading frame of mouse Bach2 cDNA (16) was isolated by PCR and inserted into the BamHI site of pcDNA3.1BFLAG, resulting in pcDNAFLAGBach2. The primers possessing BamHI site at each 5Ј end, were 5Ј-GTTAAGGAT-CCATGTCTGTGGATGAGAGACCT-3Ј and 5Ј-GTTAAGGATCCCTAG-GCATAATCTTTCCTGGG-3Ј (initiation and stop codons are underlined). A 2.5-kilobase pair NotI and HindIII fragment was isolated from the pcDNAFLAGBach2, filled in, and inserted into the HpaI site of MSCV/IRES-EGFP (28). To construct Bach2⌬BTBFLAG, the EcoRI-BamHI fragment was isolated from Bach2 cDNA, filled in, and cloned into the blunt-ended KpnI site of pcDNA3.1FLAG, resulting in pF-⌬BT-BB2. A filled-in NotI and HindIII fragment of pF-⌬BTBB2 was cloned into the HpaI site of MSCV/IRES-EGFP. To construct the Bach2⌬Zip retrovirus vector, NotI and PmeI fragments were isolated from the Bach2⌬Zip expression vector (29), filled in, and cloned into the HpaI site of MSCV/IRES-EGFP. The PCR-created mouse Bach1 cDNA fragment (16) was inserted into the BamHI and HindIII site of pcDNA3.1FLAG, generating pF-Bach1. The primers (5Ј-GTTAAGAAT-GATCAATGTCTGTGAGTGAGAGT-3Ј and 5Ј-GTTAAGAAGCTTTTA-CTCGTCAGTAGTGCACTT-3Ј) contained the BclI and HindIII sites, respectively, and were amplified between the initiation and stop codons of Bach1. A 2.5-kilobase pair filled-in NotI and HindIII fragment of pF-Bach1 was inserted into the HpaI site of MSCV/IRES-EGFP.
To generate chimeric cDNAs of Bach1 and Bach2, portions of the respective cDNAs were isolated by PCR and ligated into pcDNA3.1BFLAG. The primers used to generate B1B2A were: 5Ј-GT-TAAGAATGATCAATGTCTGTGAGTGAGAGT-3Ј and 5Ј-GCAGCAGG-TACCCAGGCTAATCACACAAGC-3Ј containing the BclI and KpnI sites, respectively (for Bach1 amplification) and 5Ј-GCAGCAGGTACC-AATTCCAGTGACGAGTCT-3Ј and 5Ј-GCAGCAAAGCTTCTAGGCAT-AATCTTTCCT-3Ј containing the KpnI and HindIII sites, respectively (for Bach2). Amplified DNA were digested with BclI and KpnI or KpnI and HindIII and were inserted between the BamHI and HindIII sites of pcDNA3.1BFLAG, resulting in pcDNAFLAGB1B2(A). The primers used to generate B2B1B were: 5Ј-GTTAAGGATCCATGTCTGTGGAT-GAGAAGCCT-3Ј and 5Ј-GCAGCAGGATCCGTAAGACTGCTCACATT-T-3Ј containing the BamHI site (for Bach2) and 5Ј-GCAGCAGGATCC-GACTCTGAGACGGACACG-3Ј and 5Ј-GTTAAGAAGCTTTTACTCGT-CAGTAGTGCACTT-3Ј containing the BamHI and HindIII sites, respectively (for Bach1). Amplified DNA were digested with BamHI or BamHI and HindIII and were inserted between the BamHI and HindIII sites of pcDNA3.1BFLAG, resulting in pcDNAFLAGB2B1(B). pcDNAFLAGB1B2(C) was constructed by substituting the EcoRI-Hind-III fragment of pcDNAFLAGB1B2(A) with a Bach2 cDNA amplified with a primer set of 5Ј-GCAGCAGAATTCGAAGAGGAGGAAGAAGA-G-3Ј and 5Ј-GCAGCAAAGCTTCTAGGCATAATCTTTCCT-3Ј containing EcoRI and HindIII sites, respectively. To generate retrovirus vectors, NotI-HindIII DNAs were isolated from these chimeric cDNA plasmids and bluntly inserted into the HpaI site of the MSCV/IRES-EGFP vector.
Transfection Reporter Assays-The reporter plasmids are based on a TATA box-luciferase reporter plasmid and were described previously (8,12,30). The MARE reporter 1 possesses three copies of the palindromic MARE, whereas reporter 23 possesses mutated MARE, retaining functional TRE, and thus allows binding of Jun/Fos dimers but not Maf dimers (8). The ARE reporter was described previously (30). Transfection reporter assays were carried out as described previously (23). Three independent experiments, carried out in duplicate, were performed, and the results are averaged and diagrammed with the standard errors.
Transduction of NIH3T3 Cells-Phenix Ecotropic packaging cells were transfected with each retroviral vector construct using FuGENE 6 Transfection Reagent (Roche Molecular Biochemicals), and the viral supernatants were harvested 2 days after transfection. For infection, retroviral-containing supernatants with 4 g/ml Polybrane were added to NIH3T3 cells. After 3 h of incubation in a 5% CO 2 incubator, fresh culture medium was added. Two days post infection, infected NIH3T3 cells were harvested and passaged.
Immunoblotting Analysis-Whole cell extracts were prepared from transduced NIH3T3 cells or transfected QT6 cells, separated on 7.5% SDS-polyacrylamide gels, transferred onto polyvinylidene difluoride membranes (Millipore), and processed for reaction with antibodies as described previously (29). All of the antibody reactions were performed in Tris-buffered saline (0.15 M NaCl, 20 mM Tris-HCl, pH 7.5) with 5% skim milk and 0.05% Tween 20. Detection of peroxidase activity was performed with an ECL system (Amersham Biosciences).
FACS Analysis-NIH3T3 cells were collected using Trypsin-EDTA (Sigma) treatment, and the cells were stained with 2.5 g/ml propidium iodide (Sigma) in PBS. The cells were sorted using a fluorescence cell sorter (FACScalibur; Becton Dickinson). The collected data were analyzed by CellQuest software (Becton Dickinson).
Image Acquisition-The images were taken with a Leica epifluorescence microscope equipped with a charge-coupled device camera controlled by QFluoro software (Leica). Adobe Photoshop was used for presentation of the images.

Effects of Bach2 on the Related Enhancers TRE, MARE, and
ARE-To understand the repertoires of genes that are regulated by Bach2, we first examined the effects of Bach2 upon TRE, MARE, or ARE reporters in co-transfection assays using NIH3T3 cells (Fig. 1). Bach2 repressed MARE-dependent reporter gene expression (palindromic MARE reporter 1 and NF-E2-type MARE from the chicken ␤-globin gene) as well as the TRE reporter (reporter 23). In addition, the ARE reporter that possesses three copies of ARE of GST-Ya gene (30) was also efficiently inhibited. In contrast, Bach2 had no effect on the reporter plasmid carrying mutated MARE sequences (reporter 17), verifying its sequence specificity. The oxidative stressor DEM that depletes glutathione induced MARE-and ARE-dependent expression but not TRE-dependent expression, suggesting a functional distinction among the related enhancers. Co-expression of Bach2 efficiently inhibited the induction of MARE and ARE by DEM. These results indicate that Bach2 inhibits the related enhancers TRE, MARE, and ARE as well as the induction of the latter two elements upon oxidative stress. Bach2 most likely binds to MARE and ARE along with the small Maf protein in transfected cells (16). On the other hand, because Bach2 binds to TRE as a homodimer in vitro (16), its homodimer may repress TRE.
Expression of Bach2 and Bach1-The above results suggest Bach2 as a generic inhibitor of TRE-related enhancers. To evaluate the effect of Bach2 expression on cellular function at the level of a single cell, we employed a bicistronic retroviral system possessing the encephalomyocarditis virus-derived IRES and the EGFP with a murine stem cell virus backbone (28). Virus-infected cells can be monitored individually without selecting clonal stable cell lines. This is important when one is to analyze the effects of toxic gene products. The structures of retroviral vectors used in this study are depicted in Fig. 2A. The wild-type Bach2 cDNA and its derivatives were placed upstream of the IRES-EGFP. Some of the cDNAs were tagged with a FLAG epitope-coding sequence to monitor their expression. Bach2⌬BTB lacked the BTB domain that is specific to Bach1 and Bach2 among the NF-E2-related factors. The BTB (Broad-Complex, Tramtrack, and Bric-a-brac) domain is found in more than 100 human proteins (32) and mediates oligomer formation (33)(34)(35)(36), DNA loop generation (37), and nuclear foci formation (34,35,38). Bach2⌬Zip lacked the leucine zipper region that mediates dimerization with small Maf family factors and is thus essential for DNA binding. To compare their biological functions, we also constructed a FLAG-Bach1 retrovirus. To verify their expression, we prepared extracts from Expression of Bach2 and its derivatives was analyzed by immunoblotting with the Bach2 monoclonal antibody. C, Bach1 and Bach2 retroviral vectors were transfected into QT6 cells. One day after transfection, whole cell extracts were prepared from each culture. Immunoblotting analysis was performed with the anti-FLAG antibody.
infected NIH3T3 cells and performed immunoblot analyses using an anti-Bach2 monoclonal antibody (Fig. 2B) or anti-FLAG antibody ( Fig. 2C and see below). The epitope of the anti-Bach2 antibody resides just upstream of the basic region. 2 The retroviruses expressed antigens of expected sizes and at similar levels (Fig. 2B).
Negative Regulation of Cell Proliferation by Bach2-Activation of TRE-and MARE-dependent transcription appears to lead to transformation (8, 39 -43). Because Bach2 represses both TRE-and MARE-dependent transcription, it is possible that Bach2 is a negative regulator of cell proliferation. To test this possibility, we infected NIH3T3 cells with the retroviruses and monitored changes in the relative numbers of infected and uninfected cells (i.e. EGFP-positive and -negative cells) in each cell culture using a FACS. Four days after infection, more than 50% of the cells were EGFP-positive. Thereafter, levels of Bach2-expressing cells were remarkably reduced as determined by the ratios of EGFP-positive cells (Fig. 3A). Diminution of Bach2-expressing cells was very rapid, and the numbers were reduced to 1.6% by day 14 post-infection (Fig. 3B). The disappearance of Bach2-expressing cells was confirmed by immunoblot analyses of cell extracts with the anti-Bach2 antiserum (Fig. 3C). Four days after infection, the cells expressed Bach2 abundantly. However, by day 12 post-infection, there were no detectable levels of Bach2 being expressed. The relative numbers of Bach2⌬BTB-expressing cells also decreased after infection. However, the kinetics for reduction was slower than that for cells expressing the full-length Bach2 (Fig. 3B). In addition, a significant fraction of cells was still EGFP-positive at 14 days post-infection. The ratios of Bach2⌬Zip-expressing cells did not change (Fig. 3, A and B). Because these Bach2 derivatives are expressed at similar levels (Fig. 2B), the differences in their inhibitory effects on cell proliferation could not be explained by a different expression level of each construct. Even though Bach1 and Bach2 share many biochemical activities, we could not detect any obvious effect of Bach1 on cell proliferation (Fig. 3B). The levels of Bach1 and Bach2 accumulation were compared by immunoblot analysis using an anti-FLAG antibody (Fig. 2C). Even though the band corresponding to Bach1 migrated to a position equal to that of an endogenous reactive band with an unknown identity, it is clearly shown that Bach1 is expressed at higher levels than Bach2 in this system. These results indicate that Bach2 possesses an inhibitory or toxic effect on cell proliferation and that the DNA binding activity of Bach2 is a prerequisite for the effect. the ability to inhibit cell proliferation when expressed in NIH3T3 cells. The possibility that the Bach2-IRES-EGFP transgene underwent selective silencing is unlikely based on the observations described below.
Bach2 Enhances Oxidative Stress-induced Cell Death-To elucidate the biological role of Bach2 during oxidative stress, we examined the fate of Bach2-overexpressing cells upon exposure to oxidative stress. Three days after infection, transduced NIH3T3 cells were treated with 50 or 150 M DEM. After further incubating for 24 h, the cells were stained with propidium iodide (PI), and the numbers of living cells and PIstained dead cells were measured by FACS. Roughly 45-70% of the cells were EGFP-positive at the beginning of the DEM treatment. After 1 day of incubation, we observed a significant reduction in the number of EGFP-positive cells in the cultures infected with the wild-type Bach2 virus (Fig. 4A). In contrast, we observed virtually no effect of DEM treatment in the cultures infected with the EGFP virus. Even though Bach2⌬BTBoverexpressing cultures exhibited a decrease in EGFP-positive cells, there were significant numbers of surviving cells that were EGFP-positive. Overall, we did not observe any change in the numbers of EGFP-positive cells in the cultures infected with the Bach2⌬Zip virus (Fig. 4A). The percentages of PIpositive dead cells in all of the cultures are shown in Fig. 4B. These results clearly indicate that Bach2 caused rapid cell death in the presence of oxidative stress and that the BTB domain was not essential; however, it did play an auxiliary role in the cell death-inducing activity of Bach2. The data obtained with Bach2⌬Zip clearly show that the DNA binding activity of Bach2 is indispensable for the response to oxidative stress.
In addition to DEM, hydrogen peroxide caused a massive and selective death of Bach2-overexpressing cells (Fig. 5A). In contrast to Bach2, however, the expression of Bach1 did not enhance the induction of cell death upon treatment with DEM or hydrogen peroxide, indicating a functional distinction between the related factors. Amounts of Bach1 within cells were not affected by the hydrogen peroxide treatment (Fig. 5B), excluding the possibility that oxidative stress decreased Bach1 pro- tein levels. These observations establish a biological role for Bach2 in the oxidative stress response by its enhancement of cell death.
Nuclear Accumulation of Bach2 Is Not Sufficient for Cell Death-Because oxidative stress induces the nuclear accumulation of Bach2, this alone may be sufficient to induce cell death. Alternatively, Bach2 may cooperate with other regulators that are also activated by the stress. To address these possibilities, we examined the effects of leptomycin B (LMB), which inhibits the exporter protein Crm1 and thus induces the nuclear accumulation of Bach2 (23). NIH3T3 cells were infected with the Bach2-expressing retrovirus and cultured in the absence or presence of LMB, and the number of PI-positive cells were monitored as described above. In contrast to DEM, LMB did not change the number of dead cells (Fig. 5C), which means that the nuclear accumulation of Bach2 induced by LMB is not sufficient to induce cell death. Nuclear accumulation of Bach2 in the presence of LMB was confirmed by immunostaining of the infected cells (Fig. 5D). These results suggest that the activation of additional signaling cascade(s) by oxidative stress is essential for causing cell death in the presence of Bach2. Alternatively, LMB treatment per se may protect cells from cell death. However, this is unlikely because LMB was shown previously to promote apoptosis by the BCR-ABL fusion gene product (44).
Bach2-dependent Cell Death Is Apoptotic-To examine whether oxidative stress-induced cell death in Bach2-expressing cells occurs via an apoptotic mechanism, we monitored the cell surface expression of the phospholipid, phosphatidylserine. NIH3T3 cells infected with the Bach2 or control retroviruses were treated with 150 M DEM for 10 h, and the cultures were stained using phycoerythrin-conjugated annexin V. The cells were viewed under fluorescent microscopy for the binding of annexin V-phycoerythrin. Annexin V binding was clearly detected in Bach2-expressing cells but not in uninfected cells (i.e. EGFP-negative cells; Fig. 6A, left column). In contrast, annexin V binding was not observed in cells infected with the control EGFP virus (Fig 6A, right column). Although 40 -50% of the Bach2-expressing cells were found positive for annexin V binding, virtually no cell was positive in the control virus-infected cultures. To observe changes in nuclear morphology, the cells were stained with Hoechst 33342 following treatment with DEM for 12 h. Nuclear condensations were observed only in the EGFP-positive and, thus, Bach2-overexpressing cells (Fig. 6B). The cells infected with the control virus exhibited no changes in nuclear morphology (data not shown). Taken together, these observations suggest that Bach2 induces cell death upon oxidative stress through an apoptotic pathway.
Bach2-induced Cell Death of B-lymphoma Cells-The effect of Bach2 on the control of cell death may vary depending on the cellular context. Because Bach2 is expressed specifically in B-cells among hematopoietic cells (29), we were interested in whether Bach2 influences the sensitivity to oxidative stress among the B-lineage cells as well. Previously, we reported the establishment of Raji cell clones that overexpress human BACH2 (31). These clones seemed useful for examining the effects of Bach2 overexpression in B-cells because Raji cells do not express an endogenous BACH2 (31). We compared the sensitivities of two BACH2-overexpressing (clones 67 and 75) and two control (pDL2 and pDL3) cell lines to DEM. After treating cells with DEM for 10 h, the cells were stained with PI, and the percentages of dead cells were measured by FACS (Fig.  7). The percentages of PI-stained dead cells increased in BACH2-expressing clones depending on the concentrations of DEM. In contrast, no significant increase in the number of dead cells was observed in the two control cell lines. These data suggest that expression of Bach2 increases the susceptibility of B-cells to oxidative stress as well. Taken together, these results suggest that one of the biological functions for Bach2 is in setting a threshold for cell death induced by oxidative stress.
The N-terminal Region of Bach2 Specifies Pro-apoptotic Function-The above results indicate that Bach2 but not Bach1 functions as a pro-apoptotic factor activated upon oxidative stress. This raises an interesting question regarding a functional distinction between Bach1 and Bach2 because both repress MARE-dependent gene expression in transfection assays (16). Because the bZip domain determines the specificity of dimer formation and target gene selection, whereas the BTB domain mediates protein-protein interactions, we asked whether or not these domains of Bach1 and Bach2 are functionally equivalent. To this end, we replaced the N-or Cterminal regions of Bach2 with the corresponding regions of Bach1 (Fig. 8A). As shown in Fig. 8B, the chimeric proteins were expressed at similar levels within transfected cells. NIH3T3 cells were infected with retroviruses and treated with DEM, and the extent of cell death was compared (Fig. 8C). Among the chimeric proteins, the one containing the Bach1 bZip domain (B2B1B) elicited an efficient induction of cell death. On the other hand, the chimeric proteins containing the Bach1 N-terminal region (B1B2A) or the Bach1 BTB domain (B1B2C) resulted in a significantly reduced efficiency in the induction of cell death. These results indicate that the bZip domains of Bach1 and Bach2 function similarly in the inducible cell death assay. Rather, the functional specificity of Bach2 in terms of its cell death-inducing activity resides within its Nterminal region, which includes the BTB domain.
Bach2 Interaction with PML Nuclear Bodies-Proteins with the BTB/POZ (Poxvirus and zinc finger) domain are known to form nuclear domains (45). These domains may represent the accumulation of molecules into specific regulatory and/or functional sites within the nucleus. One of the well characterized nuclear foci is nuclear bodies (NBs) consisting of PML gene product (46,47). Because PML is involved in the regulation of apoptosis (48,49), we examined the spatial relationships of Bach2 with PML NBs upon oxidative stress. Human embryonic kidney 293T cells were transfected with the Bach2 expression plasmid and treated with DEM, and overexpressed Bach2 and endogenous PML were detected by indirect immunofluorescence staining. As shown in Fig. 9A, Bach2 showed a mainly cytoplasmic localization, whereas PML was detected as clear dots within nuclei in the absence of oxidative stress. Upon treating cells with DEM, Bach2 translocated from the cytoplasm into nuclei, forming small foci that were in close association with PML NBs (Fig. 9B). By 2 h after the DEM treatment, about 70% of Bach2-positive cells contained closely associated or co-localized Bach2 domains with PML NBs (Fig. 9C and Table I). In contrast, Bach2⌬BTB-expressing cells exhibited a constitutive nuclear accumulation without any formation of nuclear dots both in the absence or presence of DEM ( Fig. 9D and data not shown). Consistently, we did not observe any nuclear foci with a GFP-Bach2 fusion protein lacking the BTB domain (23). These results indicate that the BTB domain is important in regulating the subcellular localization of Bach2 in two aspects. First, in the absence of oxidative stress, the BTB domain is essential for the cytoplasmic localization of Bach2. Second, in the presence of oxidative stress, it directs the spatial interaction of the Bach2 foci with PML NBs.

DISCUSSION
The balance between cellular life and death is, in part, a function of the ability of a cell to control oxidant insult (50). Because responses to oxidative stress vary in various cell lineages, there must be factors that modify the response of a cell to oxidative stress. Our results provide strong evidence that Bach2 can induce apoptosis in response to oxidative stress and may explain the fact that different cells have different capacities to cope with oxidative stress. In contrast, the related protein Bach1 does not appear to have a similar function, indicating their functional specification. The issue of whether this is likely to reflect a normal physiological function of Bach2 or an artifact of overexpression needs to be addressed using a loss of function approach and is now underway in our laboratory. However, various properties of Bach2 as discussed below strongly suggest its involvement in oxidative stress-induced cell death.
Involvement of Bach2 in oxidative stress-induced cell death is in agreement with the inhibitory effect of Bach2 on ARE (Fig.  1). ARE has been found to associate with oxidative stressinducible genes such as glutathione S-transferase, heme oxygenase-1, and peroxiredoxin (21,51). In mice, Nrf2 functions as an activator of oxidative stress-inducible genes by binding to ARE (22,26,27,51). Considering the fact that Nrf2 and Bach2 regulate ARE in opposite ways, the presence of Bach2 may inhibit the deployment of protective responses against oxidative stress through ARE. Inhibition of protective genetic mechanisms that guard against oxidative stress probably plays an important role in the fate of the cell determining survival and death. This model, which involves transcriptional antagonism in the regulation of the ARE-dependent oxidative stress response, can be extended to include the possibility that Bach2 induces apoptosis in a more direct manner such as the inhibi- tion of anti-apoptotic gene expression by ARE/MARE. Indeed, some of the anti-apoptotic genes possess MAREs. 3 In addition to ARE, we report for the first time that Bach2 also represses TRE-dependent gene expression (Fig. 1). It has been suggested that AP-1 maintains a homeostatic function that reacts to changes in environmental conditions to alter gene expression profiles to adapt to new environments (6). Thus, inhibition of TRE-dependent gene functions may be required for Bach2 to regulate apoptosis. Because AP-1 has been implicated in apoptosis (6), combinations of repression of ARE-, TRE-, and MAREdependent gene functions appear to underlie the rapid and efficient apoptosis induced by oxidative stress.
On the other hand, however, we also show that although Bach1 does not induce apoptosis, the bZip domains of Bach1 and Bach2 are interchangeable in terms of apoptosis induction. This observation indicates that even though the DNA binding ability of Bach2 is a necessary precondition for the induction of apoptosis, it is insufficient for inducing apoptosis. Rather, the N-terminal region, including the BTB domain, appears to specify the pro-apoptotic function of Bach2. These observations suggest that even though both Bach1 and Bach2 are evolutionarily related to each other and they both repress transcription (16,52), there is another level of regulation that dictates Bach2-mediated cell death. Such a regulation most likely depends upon the N-terminal region of Bach2. In this respect, the association of the Bach2 foci with PML NBs under conditions of oxidative stress is intriguing. NBs are nuclear structures that contain a number of proteins including PML and are disrupted in promyelocytic leukemia (46,47,53). PML has been shown to exhibit growth-suppressive properties and to constitute a nuclear signaling pathway for apoptosis (48,49,54). Given these findings, the close association of Bach2 foci with NBs during oxidative stress may be relevant to the role of Bach2 in the regulation of apoptosis. We recently found that Bach1 does not form nuclear foci upon oxidative stress. 2 In further accord with this idea, although Bach2⌬BTB exhibited constitutive nuclear localization, it failed to form nuclear foci that associate with PML NBs and exhibited a reduced efficacy in the inhibition of proliferation and the induction of apoptosis upon oxidative stress. Furthermore, the substitution of the N-terminal region of Bach2 with that of Bach1 (i.e. B1B2A; Fig. 8) significantly reduced its activity. Thus, the requirement for the Bach2-BTB domain and the adjacent region fits well with the hypothesis that DNA binding alone is insufficient and that the interaction of Bach2 with other molecules including NBs plays an auxiliary role in regulating cell proliferation and apoptosis. A simple 3 E. Ito, unpublished observations. model would be that the function of Bach2 is modulated by its interaction with NBs. At present, however, we do not know which of the NB proteins interacts directly with Bach2. Further analysis is required to clarify the role of Bach2 in its association with NBs and to determine the mechanism of NBmediated apoptosis. Recently, it has been reported that both PML and NBs are involved in the repression of transcription (55,56). It will be interesting to determine whether TRE-, MARE-, and/or ARE-dependent gene expression in a chromatin environment is regulated by NBs.
Because the expression of Bach2 is specific to neural cells and B lymphoid cells (16,29), the pro-apoptotic function of Bach2 may be important in these cells. Bach2 is abundantly expressed in the early stages of B-cell differentiation and suppressed in terminally differentiated plasma cells (29). Together with the expression profile for Bach2, our findings raise the possibility that B lymphoid cells in later stages of differentiation may be more resistant to oxidative stress than cells in earlier stages that are expressing Bach2. Mature B-cells migrate from lymphoid organs and target areas of inflammation. In these areas reside oxidants generated by neutrophils and other phagocytic cells (3). The absence of Bach2 in terminally differentiated B-cells may allow for the induction of genes with ARE/MARE to cope with oxidative stress. On the other hand, the presence of Bach2 in the early stages of differentiation may serve to sensitize cells to oxidative stress and/or other forms of stress, thus allowing for the effective elimination of cells during differentiation. Such a system may be important as a quality control for B-cells during differentiation. The involvement of Bach2 in apoptosis agrees with previous observations that Bach2 is a candidate tumor suppressor of B-cell lymphoma (31). Further, inhibition of BCR/ABL kinase activity using STI571 in chronic myeloid leukemia cell lines and CD34 ϩ cells from chronic myeloid leukemia patients during a lymphoid crisis, induces BACH2 expression (57). Because the most striking consequence of suppressing BCR/ABL tyrosine kinase activity is the inhibition of proliferation and subsequent induction of apoptosis (44,58,59), these observations also imply a pro-apoptotic role for Bach2. The loss of Bach2 function may predispose cells toward unlimited proliferation because such an event would be expected to increase the threshold for oxi-