Heme Positively Regulates the Expression of β-Globin at the Locus Control Region via the Transcriptional Factor Bach1 in Erythroid Cells*

The transcription factor Bach1 heterodimerizes with small Maf proteins to repress Maf recognition element (MARE)-dependent gene expression. The repressor activity of Bach1 is inhibited by the direct binding of heme. To investigate the involvement of Bach1 in the heme-dependent regulation of the expression of the β-globin gene, mouse erythroleukemia (MEL) cells were cultured with succinylacetone (SA), a specific inhibitor of heme biosynthesis, and the level of β-globin mRNA was examined. A marked decrease of β-globin mRNA in SA-treated cells was observed, and this decrease was reversed by the addition of hemin. An iron chelator, desferrioxamine, also lowered the level of β-globin mRNA. The heme-dependent expression of β-globin is a transcriptional event since the expression of the human β-globin gene promoter-reporter gene containing the microlocus control region (μLCR) was inhibited when human erythroleukemia K562 cells and MEL cells were cultured with SA. Hemin treatment restored the decrease in promoter activity caused by SA. The control of the μLCR-β-globin promoter reporter gene by heme was dependent on DNase I-hypersensitive site 2 (HS2), which contains MARE. The MARE binding activity of Bach1 in K562 and MEL cells increased upon SA treatment, and the increase was diminished by the treatment with hemin. Transient expression of Bach1 suppressed the μLCR activity, and this repressor activity was cancelled by treatment with hemin. The expression of a mutated Bach1 lacking heme-binding sites led to a loss in the heme responsiveness of the μLCR. Furthermore, chromatin immunoprecipitation experiments revealed that Bach1 bound to the MARE of HS2 increased by the treatment of MEL cells with SA, and this was cancelled by hemin. We propose that heme positively regulates the β-globin gene expression by blocking the interaction of Bach1 with the MARE in the LCR.

The biochemical role of heme is related to either the transport or the utilization of oxygen, and therefore, heme exerts regulatory effects on various cell functions that sense oxygen. In addition, oxygen is an essential requirement of heme bio-synthesis. In yeast, heme regulates the expression of genes involved in the respiratory chain, heme biosynthesis, and oxidative stress, at the transcriptional level, via heme-responsive transcription factors HAP1 and HAP2/3/4/5p (1). In mammals, heme has a profound effect on the proliferation and differentiation of hematopoietic progenitors. Heme not only is incorporated as a structural component of hemoglobin but also causes an increase in the expression of globin as well as enzymes of the heme biosynthetic pathway in erythroid cells (2)(3)(4). Hemin (the ferric chloride salt of heme) treatment also increases both the number of transferrin receptor and the ferritin content (5,6). Thus, heme plays a key role in the coordinated expression of several genes during the differentiation of erythroid cells.
The human globin gene cluster spans a region of 70 kb containing five developmentally regulated genes including ⑀, ␥ G , ␥ A , ␦, and ␤. The entire region is controlled by the microlocus control region (LCR) 1 (7). Namely, high level expression of globin genes requires the LCR, an upstream region that enhances globin gene transcription and insulates the locus from the influence of flanking elements (8,9). Evidence from transgenic mice indicates that switching of the globin genes involves competition between globin promoters for enhancement by the LCR (9). The human and mouse LCRs located upstream of the ⑀-globin gene consist of five DNase I hypersensitive sites (HS) (10). Targeted disruption of HS in both mouse culture cells and mice has revealed that the LCR is not required to initiate or maintain open chromatin and basal transcription, suggesting that the main function of the LCR is the enhancement of transcription (9,11). Each HS contains AP-1, Sp-1, and/or GATA-like sites. The AP-1-like sites in the mouse HS1-4 are closely related to the Maf recognition elements (MAREs: TGCTGA(T/GT)TCAGCA) (12). Various leucine zipper proteins including p45 NF-E2, Nrf1, Nrf2, and Bach1 can interact with MAREs as heterodimers with the small Maf proteins (13). HS2 is sufficient to confer high level and tissue-specific expression on a linked ␤-globin gene of transgenic mice (14). Two tandem MAREs strongly contribute to the overall activity of HS2 (15,16).
Among MARE-associated proteins, Bach1 is unique in that it has a BTB/POZ domain (17). Bach1 forms a multivalent DNAbinding complex, raising the possibility that it can act as an architectural component that is simultaneously able to crosslink multiple MAREs, resulting in the repression of transcrip-tion (18,19). Bach1 is a heme-binding protein, and the DNA binding activity is negatively regulated by heme (20). Furthermore, Bach1 is involved in the regulation of HO-1, a ratelimiting enzyme of heme catabolism. The heme-mediated inhibition of Bach1 confers heme inducibility upon HO-1 (21). These observations suggest that Bach1 is involved in the regulation of other heme-responsive genes besides ho-1. We were particularly interested in the fact that the enhancer activity of the LCR is induced by hemin treatment of erythroid cells (15,16). To clarify whether Bach1 regulates ␤-globin gene expression, we examined heme-dependent expression of the ␤-globin gene in human erythroleukemia K562 and MEL cells and the regulation of the interaction of Bach1 with the LCR by heme.
Here we demonstrated that heme positively regulates the ␤-globin gene expression by disrupting the interaction of Bach1 with the LCR.

EXPERIMENTAL PROCEDURES
Materials-[␥-32 P]ATP, [␣-32 P]dCTP, and poly(dI-dC) were purchased from Amersham Biosciences. Restriction endonucleases and DNA-modifying enzymes were obtained from Takara Co. and Toyobo Co. Nylon membranes (Biodiene Type B) were products of Pall Co. The transfection reagent LipofectAMINE was from Invitrogen. Antibodies for c-Fos and p45 NF-E2 were purchased from Oncogene Science Co. and Santa Crutz Biotechnology Inc., respectively. Antibodies specific for Bach1 were described elsewhere, 2 and antibodies for MafK were as described previously (21). All other chemicals were of analytical grade.
Plasmids-An 838-bp human ␤-globin gene promoter from BGT14LCR (22) was cloned into the BglII-HindIII site of pGL3B, resulting in pGL␤p. A 3-kb NotI-SalI fragment of the LCR cassette that contains 5ЈHS2, 5ЈHS3, and 5ЈHS4 of the human ␤-globin gene promoter (19,22) was also isolated and inserted into a blushed SacI site of pGL␤p to give pGLCRh␤. A reporter plasmid pGLCRh␤HS2mut, lacking the tandem MAREs within the mouse HS2, was constructed as follows. Using pGLCRh␤ (19) as a template, flanking regions of the tandem MAREs were amplified by PCR. Primers used were: 5Ј-GAAT-TCGCCGCTCCCACCTCCAGCTTAGGGTGTGTGCCCAGATGTTC-3Ј (primer a) and 5Ј-CGAGCCCGGGCTAGCACGCGTA-3Ј (primer b) for one side, and 5Ј-CTGGAGGTGGGAGCGGCGAATTCGCTTGAGCCAG-AAGGTTTGCTTAG-3Ј (primer c) and 5Ј-CTCAGAGCCTGATGTTAAT-TTAGC-3Ј (primer d) for the other side. Amplified DNA were purified and mixed, and subsequent PCR was carried out using the above primers b and d. The resulting HS2 DNA lacked the tandem MAREs where the EcoRI site was introduced instead. The mutated HS2 DNA was digested with SpeI and used to replace the SpeI fragment, containing the wild-type HS2 DNA, on the pGLCRh␤. The resulting pGLCRh␤H-S2mut carried the HS4, HS3, and the mutated HS2 fused with the ␤-globin promoter and firefly luciferase gene. Mammalian expression vectors carrying wild-type Bach1 and its derivative lacking the hemebinding sites (Bach1mCP1-6) were described previously (17,20).
Cell Cultures-Mouse BALB/3T3 cells and MEL cells were grown in Dulbecco's modified Eagle's medium supplemented with 7% fetal calf serum and antibiotics. Human erythroleukemia K562 cells were grown in RPMI 1640 medium supplemented with 10% fetal calf serum and antibiotics. To reduce the level of heme in cells, the cells were incubated in medium supplemented with 7% fetal calf serum in the presence of 1 mM SA for 16 h. The amounts of heme and porphyrin in the cells were determined as described previously (23). For the differentiation of MEL cells, the cells were cultured with 2% Me 2 SO for 72 h and then collected.
Reporter Assay-Cells were transfected with the reporter plasmids pGL3B, pGL␤p, pGLCRh␤ and pGLCRh␤HS2mut, and pRL-CMV (Promega) using a LipofectAMINE reagent, according to the manufacturer's recommendations. The cells were incubated with 1 mM SA for 16 h after the transfection and washed twice with phosphate-buffered saline. They were then lysed in a Reporter lysis buffer (Promega), the lysate was centrifuged, and the supernatants were assayed for luciferase. The Photinus and Renilla luciferase assays were performed according to the protocol for the dual luciferase assay system (Promega). Transfection efficiency was normalized on the basis of Renilla luciferase activity.
Gel-shift Assay-Nuclear extracts were prepared by the method of Schneider et al. (24) with some modifications. Briefly, MEL cells (1 ϫ 10 7 ) and K562 cells (1 ϫ 10 7 ) left untreated or treated with 1 mM SA for 16 h were washed with 10 mM Tris-HCl, pH 7.6, containing 0.15 M NaCl, and then lysed with 10 mM HEPES, pH 7.9, containing 0.6% Nonidet P-40, 10 mM KCl, 0.1 mM EGTA, 1 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride. The homogenates were centrifuged, and the nuclear pellet was resuspended in 100 l of 20 mM HEPES, pH 7.9, containing 400 mM KCl, 0.5 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride. The nuclear extracts were obtained by centrifugation at 4°C. Nuclear extracts were also obtained from MEL cells cultured with 2% Me 2 SO for 48 and 72 h. The oligonucleotide probe containing the MARE/NF-E2 site from the ␤-globin enhancer was described previously (25) and was end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase. Nuclear extracts were incubated with 32 P-labeled probe (35,000 cpm) in a reaction buffer containing 25 mM HEPES, pH 7.9, 0.5 mM EDTA, 50 mM KCl, 10% glycerol, 0.5 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride, and 100 g/ml poly(dI-dC) (26). After a 10-min incubation at 37°C, the reaction mixture was loaded onto 4% polyacrylamide gels containing 50 mM Tris, 380 mM glycine, and 1 mM EDTA, pH 8.5, and electrophoresed at 100 V at room temperature. The gels were exposed to x-ray film at Ϫ80°C.
RNA Blots-Total RNA was isolated from the cells by the guanidium isothiocyanate method (23). The RNA (20 g) was applied to a 1% agarose gel, electrophoresed, and subsequently transferred onto a nylon membrane (Biodiene Type B) for hybridization with biotin-labeled antisense ␤-globin and actin RNAs. The filters were hybridized and washed, according to the method of Suzuki et al. (27). Hybridized RNA was incubated with alkaline phophatase-conjugated avidin, washed, and examined for positive signals with a luminol reagent (Roche Applied Science). The RNA concentration was quantified using an Advantec DMU-33c densitometer.
Chromatin Immunoprecipitations-Chromatin was fixed and purified, according to the method of Shang et al. (28). Briefly, MEL cells (1 ϫ 10 8 ) were fixed in 50 ml of Dulbecco's modified Eagle's medium with 1% formaldehyde at room temperature for 10 min. Immunoprecipitations with anti-Bach1 and NF-E2 antibodies were performed followed by isolation of immunoprecipitates with protein A-Sepharose beads. The PCR reaction was carried out in the reaction mixture containing 1 Ci (1 Ci ϭ 37GBq) of [␣-32 P]dCTP and 25 cycles of amplification was used. Primers used were 5Ј-CACTTCTTCATATTCTCTCTCTAG-3Ј and 5Ј-CTTATTTTCTTTTCACCTTCCCTG-3Ј to amplify the DNA fragment of HS2 (29). After PCR products were electrophoresed on 6% polyacrylamide gels, the gels were exposed to x-ray film at Ϫ80°C

Effects of SA and Hemin on ␤-Globin mRNA in MEL
Cells-SA is known as a competitive inhibitor of ␦-aminolevulinic acid dehydratase, the second enzyme of the heme biosynthetic pathway. MEL cells were cultured in the presence of 1 mM SA or an iron chelator, DFO (100 M), for 16 h, after which the amount of heme in the cells was measured. The heme content of SA-and DFO-treated cells decreased to about 70 and 45%, respectively, of that of control cells (Fig. 1A). Although the expression of globin genes is markedly induced upon the differentiation of MEL cells treated with Me 2 SO, ␣and ␤-globin are known to be synthesized at lower levels in uninduced MEL cells. To examine whether SA or DFO causes any change in the ␤-globin expression in the absence of the differentiation-inducer Me 2 SO, Northern blot analysis of ␤-globin mRNA in uninduced cells was carried out. As shown in Fig. 1B, ␤-globin mRNA in SA-treated cells decreased to about 40%. Since the level was restored on the addition of 50 M hemin, the effect of SA was through its inhibition of heme synthesis. Treatment of the cells with 100 M DFO also resulted in a decrease of ␤-globin mRNA. Again, culture of the cells with 50 M hemin reversed the decrease caused by DFO. When MEL cells were induced to differentiate with 2% Me 2 SO for 48 h, ␤-globin mRNA was expressed (Fig. 1C). The increase in ␤-globin mRNA upon the differentiation of MEL cell was markedly inhibited when the cells were cultured with 1 mM SA in addition to Me 2 SO. The decreased ␤-globin expression by the inhibition of heme biosynthesis was reversed by culture with 50 M hemin, suggesting that the expression of ␤-globin mRNA in MEL cells is controlled by heme in both uninduced and induced cells.
To clarify whether the heme-dependent expression of ␤-globin mRNA is regulated at the transcriptional level, we investigated the effect of SA on the human ␤-globin gene promoter/ LCR reporters in human erythroleukemia K562 cells. The reporter plasmid pGL␤p contains the upstream 836bp of the human ␤-globin gene, whereas pGLCRh␤ was made by inserting human LCR (30) into pGL␤p (19). The reporter activity was weak in pGL␤p-transfected K562 cells, and no change in the activity was observed on treatment with 1 mM SA or 50 M hemin for 16 h (Fig. 2A). The reporter activity was strengthened when pGLCRh␤ was transfected, consistent with the enhancer activity of the LCR. The LCR-dependent reporter gene activity decreased to 60% when the culture was treated with 1 mM SA. This loss of reporter activity was restored upon the addition of 50 M hemin. To confirm these observations in other cells, MEL cells were transfected with pGL␤p and pGL-CRh␤, and the effects of SA and hemin on the reporter activity were also examined. As shown in Fig. 2B, the decrease in pGLCRh␤ activity caused by SA, and its cancellation by hemin, was similar to that in K562 cells. In contrast, when BALB/3T3 cells were examined, neither the LCR-dependent activation nor the SA-dependent inhibition of the ␤-globin gene reporter was observed (Fig. 2C). Thus, heme induces the expression of the ␤-globin gene by stimulating the enhancer activity of the ␤-globin LCR in erythroid cells.
Involvement of HS2 of the LCR in the Regulation of Hemindependent Expression in K562 and MEL Cells-Among the DNase I-hypersensitive sites (HSs), HS3 and HS4 harbor single MAREs, and HS2 contains tandem duplicated MAREs. It was shown previously that hemin induces the enhancer activity of the tandem MAREs of HS-2 (previously referred to as the NF-E2 site) in K562 cells (15,16). To verify that the tandem MAREs are involved in hemin-dependent regulation of the reporter gene, activities of a reporter plasmid pGLCRh␤HS2mut carrying the mutated MAREs of HS2 in the LCR were compared using cells treated with SA or hemin. As shown in Fig.  3A, the basal activity of pGLCRh␤HS2mut in K562 cells was significantly less than that of the wild-type control pGLCRh␤. Moreover, the reporter activity of pGLCRh␤HS2mut was not affected by SA or hemin. When MEL cells were induced to differentiate with 2% Me 2 SO for 24 h, the reporter activity of pGLCRh␤ was markedly increased as compared with that in uninduced cells (Fig. 3B). The reporter activity of pGLCRh␤ was inhibited by SA, and the effect of SA was cancelled by simultaneously adding hemin. In contrast, the reporter activity of pGLCRh␤HS2mut did not change significantly in the presence of SA or SA plus hemin (Fig. 3B). These results indicate that a deficiency of heme in the cells leads to the suppression of the enhancer activity of the LCR and that this regulation by heme involves the tandem MAREs within the HS2.
Regulation of Hemin-dependent Expression of ␤-Globin Gene in Erythroid Cells by Bach1-The above results confirm previous observations that the tandem MAREs were essential for the Me 2 SO-induced expression of the LCR reporter. Bach1 interacts with the tandem MAREs in vitro and competes for the sites with p45 NF-E2 (18). Since Bach1 acts as a repressor of MARE-dependent transcription and its DNA binding activity is inhibited by hemin (18,20,21), it may be involved in the regulation of the enhancer activity of HS2 by heme. We then conducted a conventional gel-shift assay using a synthetic oligonucleotide probe containing a MARE from the chicken ␤-globin 3Ј enhancer to examine whether the DNA binding activity of endogenous Bach1 is indeed regulated by heme. To characterize the binding of Bach1 and p45 NF-E2, the MARE probe was incubated with nuclear extracts of untreated and Me 2 SOtreated MEL cells. As reported previously (18), we observed two specific retarding bands (Fig. 4A, lane 2). The upper band disappeared upon preincubation of the nuclear extracts with anti-MafK and Bach1, indicating that the binding complex was composed of Bach1 and small Maf protein. The DNA binding activity of the Bach1/small Maf complex was slightly and markedly decreased in 48-and 72-h-treated cells, respectively (Fig.  4A). The lower band, NF-E2 (p45 and small Maf heterodimer), which disappeared with anti-p45 NF-E2 antibodies, intensified upon Me 2 SO-induced differentiation. The specific band corresponding to Bach1 intensified when MEL cells were treated with 1 mM SA (Fig. 4B). The DNA binding activity of Bach1 was decreased by treating the cells with SA plus 50 M hemin. In this experiment, a weak band corresponding to NF-E2 was found, and the activity remained unchanged by a 16-h incuba- tion. In Me 2 SO-treated MEL cells, SA caused an increase in the Bach1 activity but inhibited the increase in the NF-E2 activity (Fig. 4A). When nuclear extracts from K562 cells were used, the probe containing MAREs exhibited a binding pattern similar to that seen with MEL cells (Fig. 4C). No change in the bindings of Bach1 and NF-E2 to the MARE occurred by the addition of SA in vitro (data not shown).
We next examined whether Bach1 affects the hemin-dependent activity of the LCR. The reporter activity was measured in K562 cells in the presence of the wild-type and heme-insensitive Bach1-expression plasmids. The Bach1 mutant (Bach1mCP1-6) carrying multiple changes in the heme-binding cysteine-proline (CP) motifs does not bind heme, and its DNA binding activity is not inhibited by heme (20,21). As shown in Fig. 5A, the reporter activity of pGLCRh␤ was suppressed by the wild-type Bach1. Consistent with the previous results using a synthetic reporter and HO-1 reporter (20,21), the repressor activity of Bach1 was lost by treating cells with hemin (Fig. 5A). In contrast, the repressor activity of Bach1mCP1-6 was not affected by hemin treatment (Fig. 5A). As shown in Fig. 5B, the repressor activity of Bach1 further dropped in SA-treated cells. Unlike the case for the wild-type Bach1, the reporter activity was not further reduced by SA in Bach1mCP1-6 expressing cells (Fig. 5B). When Bach1 was expressed in MEL cells, a heme-dependent regulation of the LCR activity, similar to that in the case of K562 cells, was observed (data not shown). These results suggest that Bach1 regulates the heme-dependent activity of the HS2 enhancer and that the repressive ability of Bach1 on the LCR activity is independent of the heme responsiveness.

Occupancy of HS2 of the ␤-Globin LCR Enhancer by Bach1
and NF-E2-We finally examined the recruitment of Bach1 and NF-E2 to the MARE of HS2 in MEL cells treated with SA and hemin. The status of the endogenous transcription complexes present on the HS2 was determined using chromatin immunoprecipitations. The presence of the HS2 in the chromatin immunoprecipitates was analyzed by semiquantitative PCR using specific pairs of primers spanning the HS2 of the LCR. As shown in Fig. 6A, the PCR product increased dependent on 22-28 cycles, and then we analyzed the products with 25 cycles of PCR to evaluate the binding of Bach1 and NF-E2 to HS2. Bach1 and NF-E2 were equally bound to the HS2, and an increase in the binding of Bach1 was observed when cells were treated with 1 mM SA (Fig. 6B). The increase of Bach1 was cancelled by treatment with 50 M hemin, whereas the binding of NF-E2 to the HS2 increased in the same cells.

DISCUSSION
The present study showed that Bach1 plays an important role in the differentiation of erythroid cells and directly regulates the expression of ␤-globin, mediated by hemin. The expression of ␤-globin mRNA was reduced when the content of heme in uninduced and Me 2 SO-treated MEL cells was decreased by treatment with SA or DFO. Exogenously added hemin restored the level of ␤-globin mRNA. These findings were similar to previous observations that the expression of globin chains in MEL and human erythroid progenitor cells was reduced when these cells were cultured with SA (31,32) and support the concept of the positive effect of heme on ␤-globin synthesis (33,34). Using a promoter assay of ␤-globin, the present study confirmed that the expression of ␤-globin mRNA is regulated by hemin at the transcriptional level. There are also many reports that heme induces erythropoiesis by the coordinate expression of ␣-, and ␤-globin, transferrin receptors, ferrochelatase, and erythroid specific ␦-aminolevulinic acidsynthase (4,32,35). In erythroid cells, it is reported that heme acts as not only a transcription factor NF-Y for ferritin synthesis (36) but also as a transcription factor for kinases including the heme-regulated inhibitor kinase, heme-regulated eIF2␣ kinase (37). It is also known that heme-regulated eIF2␣ kinase mRNA increases in induced MEL cells but decreases when the heme biosynthesis is blocked (38). Furthermore, dominant negative heme-regulated eIF2␣ kinase mutant expressed in MEL cells resulted in the increase in hemoglobin synthesis (38). Thus, heme is well known to regulate the synthesis of globin chains at the translational step. We have shown for the first time that the hemin-dependent activation of ␤-globin expression at the transcriptional level occurred at the HS2 of the LCR, where it interacted with Bach1.
Analysis by atomic force microscopy revealed large looped DNA structures between MAREs located in different regulatory elements within the human ␤-globin LCR formed by Bach1/MafK heterodimer (19). Based on the observations that the formation of these loops required the Bach1 BTB/POZ protein interaction domain (19), Bach1 would function as an architectural factor and cause the repression of the enhancer activity of the LCR. We confirmed that Bach1 suppresses the activity of the LCR and demonstrated that the heme binding ability of Bach1 contributes to the hemin-dependent induction of the LCR enhancer (Fig. 5). Consistent with previous findings that the formation of a multimetric and multivalent DNA complex with multiple MAREs occurs simultaneously (18) and that ␤-globin LCR holocomplex can be formed by structurally con- necting the MAREs present in HS2, HS3, and HS4 of the LCR (9,19), Bach1 contributes to the generation of a multiprotein complex as a repressosome among the HS of the LCR (39). The present data by gel-shift assay and chromatin immunoprecipitations clearly showed the physiological role of the interaction of Bach1 with HS2 in the regulation of the expression of ␤-globin.
Recently, chromatin immunoprecipitation experiments with anti-MafK and anti-p45 NF-E2 showed that the LCR was occupied by small Maf proteins in uninduced MEL cells where the synthesis of globin is suppressed (29) and that p45 NF-E2 was recruited to the LCR as well as the active globin promoters on erythroid differentiation. Using chromatin immunoprecipitations, we found that the recruitments of Bach1 and NF-E2 to the MARE of HS2 were regulated by the intracellular level of heme, indicating that Bach1 and NF-E2 share MAREs and effectively compete with them. Furthermore, the present study and a previous study (18) showed that the DNA complex with Bach1 was down-regulated at the late stage of Me 2 SO-induced MEL cell differentiation, whereas the NF-E2 complex was in-duced to form by Me 2 SO treatment. Bach1 may be involved in the assembly of the LCR complex at the early stages of hematopoietic cell differentiation, whereas p45 NF-E2 contributes to the activation of globin genes per se at the late stage. Based on the fact that the level of intracellular heme was markedly elevated in Me 2 SO-treated cells and that a marked decrease in Bach1 activity was observed in hemin-treated cells, heme contributes to the loss of function of Bach1 during erythroid differentiation. This is supported by the finding that the expression of Bach1 and Bach1mCP1-6 led to a decrease in the activity for the LCR enhancement, but the hemin-dependent restoration of the LCR activity was only observed in wild-type Bach1-expressing cells (Fig. 5). In fact, the treatment of MEL cells with Me 2 SO increased NF-E2 activity significantly (40), which is also supported by the present observation that the reporter activity of pGLCRh␤ in Me 2 SO-induced MEL cells was markedly increased as compared with that in uninduced cells (Fig. 3B). Thus, the induction of the p45 NF-E2 binding activity during erythroid differentiation implies that the replacement of the Bach1 associated with small Maf molecules with the NF-E2 complex occurs dependent on the stage of the differentiation. Similarly, a recent study (21) clarified the mechanisms involved in the induction of HO-1 by hemin. Namely, Bach1 represses the expression of HO-1 under physiological conditions, and an increased level of heme displaces Bach1 from the enhancers by inhibiting DNA binding, allowing activators to bind the enhancers. Thus, heterodimers of small Maf-and p45 NF-E2-related activators including Nrf2 are most likely the form binding to the HO-1 enhancers upon transcriptional activation.
We now provide evidence of a new function of heme directly regulating the transcription of genes. For a long time, it has been considered that intracellular heme has a significant role in the transcriptional up-regulation of several erythroid-specific proteins, including globin chains, transferrin receptors, ferritin, and enzymes of the heme biosynthetic pathway (3). We identified the LCR ␤-globin gene as one of the targets of Bach1 where heme acts as a positive regulator of the transcription of the ␤-globin gene by regulating the interaction of Bach1 with the MARE region of the LCR (HS2). We propose that transcriptional regulation of ␤-globin is a direct sensing of heme levels during the terminal differentiation of erythroid cells. Since Bach1 is ubiquitously expressed in a variety of tissues, there may be activation systems for various genes involving the replacement of the repressor Bach1 with some enhancer protein, the event being triggered by heme. FIG. 5. Transcriptional activity of wild-type and mutated Bach1 for the LCR ␤-globin gene in K562 cells. As shown in A, K562 cells were co-transfected with pGLCRh␤ vector and plasmids carrying the wild-type Bach1 or Bach1 mutations in the cysteineproline motifs (Bach1mCP1-6) (0.5-1.5 g). The cells were cultured with or without 50 M hemin. As shown in B, pGLCRh␤ was also transfected with Bach1 or Bach1mCP1-6 (1.0 g) into K562 cells. The cells were treated with 0.25-0.5 mM SA for 16 h. The reporter gene assay was carried out, as above. Data are the average of three separate experiments.
FIG. 6. Occupancy of HS2 of the LCR enhancer by Bach1 and NF-E2 in MEL cells. A, relation of the amount of PCR products with PCR cycles. The HS2 of the LCR was examined for the occupancy by Bach1 and NF-E2 with the corresponding antibodies and the PCR primers shown under "Experimental Procedures." Input corresponds to PCR reactions containing 0.5% total amount of chromatin used in immunoprecipitation reactions. The PCR was done with the indicated cycles. B, effect of SA or hemin on the occupancy of HS2 by Bach1 and NF-E2. MEL cells were treated without or with 1 mM SA or 1 mM SA plus 50 M hemin for 16 h. Chromatin immunoprecipitation and PCR with 25 cycles were done as above. Mock immunoprecipitation with irrelevant and control antibodies (anti-ferrochelatase) is shown.