Hemoprotein Bach1 regulates enhancer availability of heme oxygenase-1 gene

Abstract

Heme oxygenase-1 (HO-1) protects cells from various insults including oxidative stress. Transcriptional activators, including the Nrf2/Maf heterodimer, have been the focus of studies on the inducible expression of ho-1. Here we show that a heme-binding factor, Bach1, is a critical physiological repressor of ho-1. Bach1 bound to the multiple Maf recognition elements (MAREs) of ho-1 enhancers with MafK in vitro and repressed their activity in vivo, while heme abrogated this repressor function of Bach1 by inhibiting its binding to the ho-1 enhancers. Gene targeting experiments in mice revealed that, in the absence of Bach1, ho-1 became expressed constitutively at high levels in various tissues under normal physiological conditions. By analyzing bach1/nrf2 compound-deficient mice, we documented antagonistic activities of Bach1 and Nrf2 in several tissues. Chromatin immunoprecipitation revealed that small Maf proteins participate in both repression and activation of ho-1. Thus, regulation of ho-1 involves a direct sensing of heme levels by Bach1 (by analogy to lac repressor sensitivity to lactose), generating a simple feedback loop whereby the substrate effects repressor-activator antagonism.

Fig. 1. Repression of HO-1 enhancers by Bach1 is alleviated by heme. (A) Reporter plasmids carrying upstream 15 kb DNA with or without deletions of E1 and/or E2. Comparison of NF-E2 binding site (Andrews et al., 1993a) and MARE-like elements within the E1 (E1M1-3) and E2 enhancers (E2M1-3) is shown below. (B) NIH 3T3 cells were transfected with the wild-type reporter or enhancer-less reporters (0.5 µg) together with (shaded) or without (filled) a Bach1-expression plasmid (0.1 µg). (C) 293 cells were transfected with the wild-type reporter with or without Bach1-expression plasmid. Where indicated, 10 µM hemin was added to the medium. (D and E) 293 cells were transfected with the wild-type reporter along with 0.1 µg each of expression plasmids for Bach1, MafK (D) and Nrf2 (E), as indicated, in the absence or presence of added hemin.

Fig. 2. Heme-regulated binding of Bach1 to the HO-1 enhancers. (A) Footprinting assays were carried out using 562 bp E2 DNA in the presence of 50 ng of MafK (lanes 2–11) and increasing amounts (40, 80, 120, 240 ng) of Bach1 (lanes 3–6) or Bach1ΔBTB (lanes 8–11). Protected regions were identified by sequencing reactions and are shown at the right (E2M1-3). HS indicates hypersensitive sites. (B) EMSA was carried out using the 562 bp E2 DNA in the presence of 50 ng of MafK (lanes 2–8) and increasing amounts (40, 80, 160 ng) of Bach1 (lanes 5–7) or Bach1ΔBTB (lanes 2–4). Bach1 and Bach1ΔBTB (160 ng) were examined in the absence of MafK (lanes 9 and 10). Complexes were separated on 1% agarose gels. (C) Binding of MafK (50 ng) and Bach1ΔBTB (40 ng) to individual MARE-like sites was examined using indicated oligonucleotide probes. Complexes were separated on 4.5% polyacrylamide gels. Heterodimeric complexes are indicated with arrowheads. (D) Agarose gel EMSA was carried out as in (B) using recombinant MafK and Bach1 (lanes 2–5), Bach1mCP1-6 (lanes 6–9) or MafK alone (lane 10). Hemin was added to the reactions at 0, 0.25, 0.75 or 1.5 µM. (E) The wild-type HO-1 reporter plasmid was transfected into 293 cells with MafK-, Bach1- or Bach1mCP1-6-expression plasmids in indicated combinations in the presence or absence of 10 µM hemin. Fold-repression was calculated as: (reporter activity without effectors)/(reporter activity with effectors). (F) Accumulation of Bach1 and Bach1mCP1-6 within transfected cells was compared by immunoblotting analysis.

Fig. 3. Genetic ablation of bach1 in mice. (A) The bach1 genomic structure sorrounding the BTB domain-coding exon that includes the initiation methionine is indicated in the top line. The targeting vector is indicated in the second line. The targeted allele is indicated in the third line. DNA fragments used as probes for Southern blotting analysis are shown above and below the lines. Primers for PCR screening are indicated with arrows a–c. (B) Southern blot hybridization of the 5′ and neo probes to PstI-digested genomic DNA prepared from wild-type, heterozygous and homozygous mutant mice. The 5′ probe hybridized with a 6 kbp (wild-type allele) or a 2.6 kbp (targeted allele) PstI DNA fragment. The neo probe hybridized with the same 2.6 kbp PstI DNA fragment on the targeted allele. Detection of wild-type and mutated alleles by PCR are shown below the Southern blots. (C) Expression of Bach1 mRNA in thymus was examined by RT–PCR using two (bach1^+/+) or three (bach1^–/–) mice. (D) Protein extracts from thymus of bach1^+/+ or bach1^–/– were analyzed for expression of Bach1 using antiserum raised against Bach2 that is weakly reactive with Bach1. Extracts of Qt-6 cells transfected with a Bach1 expression plasmid was loaded in lane 1.

Fig. 4. Expression of HO-1 in bach1-deficient mice. (A) Protein extracts of indicated tissues of bach1^+/+ or bach1^–/– (shown by W and K, respectively) mice were analyzed for expression of HO-1 by immunoblot assays (upper panels). HO-1 is indicated by an arrowhead. Membranes were stained for protein with amido black to verify equal loading (lower panels). (B) HO-1 and HPRT mRNA levels in thymus, heart, lung and liver from various genotypes (indicated above the panels) were compared by RT–PCR. To ensure linearity of amplification, 3-fold dilutions of cDNA (lanes 1–3 and 4–6, and indicated with H and L at the top) were used as templates. Images were recorded using the Agilent BioAnalyzer 2100.

Fig. 5. Nrf2-dependent and -independent expression of ho-1. (A) HO-1 and HPRT mRNA levels were compared as described in Figure 4 using RNA from various tissues with the indicated genotypes. H and L at the top indicate reactions with 3-fold different amounts of templates. Images were recorded using the Agilent BioAnalyzer 2100. (B) Relative levels of HO-1 mRNA were determined and corrected for HPRT mRNA levels by PCR and BioAnalyzer 2100 analysis. Results are the mean of at least two mice for each genotype, with the SEM indicated. (C) Expression of Nrf2 mRNA in thymus and heart was compared between wild-type and bach1-deficient mice by PCR. Similar results were obtained using two mice for each genotype.

Fig. 6. Binding of small Maf to ho-1 enhancers in vivo. (A) Gel images showing PCR amplification of E2, exon 1 and Rag-2 gene fragments recovered in the absence or presence of anti-small Maf antibodies. Chromatin was prepared from the wild-type (WT) or bach1-deficient (KO) thymi. PCR products without DNA template are shown in the right-hand lanes. (B) Fold-enrichment of each genomic DNA region by the small Maf antibodies was determined using the Agilent BioAnalyzer 2100. Results are mean of two experiments using two mice per each genotype. (C) A hypothetical model describing the regulation of ho-1 by Bach1 and heme. Besides MafK, other Maf-related factors may also serve as partners for Bach1. Bach1 makes enhancers inaccessible to Nrf2 and other activators by binding to them. Repression by Bach1 is alleviated upon increase of heme levels, allowing expression of HO-1 followed by degradation of heme.