Research article

Novel insights into the regulation of antioxidant-response-elementmediated gene expression by electrophiles: induction of the transcriptional repressor BACH1 by Nrf2

Henna-Kaisa Jyrkkänen, Suvi Kuosmanen, Merja Heinäniemi, Heidi Laitinen, Emilia Kansanen, Eero Mella-Aho, Hanna Leinonen, Seppo Ylä-Herttuala, Anna-Liisa Levonen


A central mechanism in cellular defence against oxidative or electrophilic stress is mediated by transcriptional induction of genes via the ARE (antioxidant-response element), a cis-acting sequence present in the regulatory regions of genes involved in the detoxification and elimination of reactive oxidants and electrophiles. The ARE binds different bZIP (basic-region leucine zipper) transcription factors, most notably Nrf2 (nuclear factor-erythroid 2-related factor 2) that functions as a transcriptional activator via heterodimerization with small Maf proteins. Although ARE activation by Nrf2 is relatively well understood, the mechanisms by which ARE-mediated signalling is down-regulated are poorly known. Transcription factor BACH1 [BTB (broad-complex, tramtrack and bric-a-brac) and CNC (cap'n'collar protein) homology 1] binds to ARE-like sequences, functioning as a transcriptional repressor in a subset of ARE-regulated genes, thus antagonizing the activator function of Nrf2. In the present study, we have demonstrated that BACH1 itself is regulated by Nrf2 as it is induced by Nrf2 overexpression and by Nrf2-activating agents in an Nrf2-dependent manner. Furthermore, a functional ARE site was identified at +1411 from the transcription start site of transcript variant 2 of BACH1. We conclude that BACH1 is a bona fide Nrf2 target gene and that induction of BACH1 by Nrf2 may serve as a feedback-inhibitory mechanism for ARE-mediated gene regulation.

  • antioxidant-response element (ARE)
  • BTB and CNC homology 1 (BACH1)
  • gene expression
  • gene regulation
  • nuclear factor-erythroid 2-related factor 2 (Nrf2)
  • transcription factor


Cells have an inherent ability to sense and respond to environmental stress via a variety of gene-regulatory pathways. A number of antioxidant and detoxification enzymes are transcriptionally induced during oxidative or electrophilic stress via the ARE (antioxidant-response element) [1]. The ARE was initially characterized as having a cis-acting sequence with a consensus core sequence TGACnnnGC [2], and it was characterized further and extended to 5′-TMAnnRTGAYnnnGCRwwww-3′ (the core ARE is in bold) [3]. Subsequent analyses of the human GCLM (glutamate–cysteine ligase modifier) [4] and the mouse NAD(P)H:quinone oxidoreductase [5] genes suggest that a universally applicable ARE may not be attainable, although certain core nucleotides appear to be highly conserved [6]. Multiple bZIP (basic-region leucine zipper) transcription factors, such as those of the CNC (cap'n'collar protein) family {Nrf [NF-E2 (nuclear factor-erythroid 2)-related factor] 1, Nrf2, Nrf3 and p45 NF-E2}, and BACH [BTB (broad-complex tramtrack and bric-a-brac) and CNC homology] proteins BACH1 and BACH2 bind to the ARE [7]. All members from both protein families form heterodimers with small Maf proteins, which are necessary for DNA binding [7]. Transcription factors can either transactivate or repress gene expression via binding to the ARE.

Nrf2 (NFE2L2) is one of the main ARE-binding transactivators, and it regulates a wide variety of genes involved in antioxidant defence and xenobiotic metabolism, including haem oxygenase-1 (HMOX1) and NAD(P)H:quinone oxidoreductase-1 (NQO1) [8]. Under basal conditions, Nrf2 is bound to its inhibitor Keap1 [Kelch-like ECH (erythroid cell-derived protein with CNC homology)-associated protein 1], where Keap1 acts as an adapter between Nrf2 and Cul3-based E3 ubiquitin ligase, facilitating proteasomal degradation of Nrf2. On exposure to oxidants or electrophiles, Nrf2 accumulates in the nucleus and drives gene expression via the ARE [9]. BACH1 is a repressor of transcription via competitive binding to the MARE (Maf-recognition element), a regulatory element closely related to ARE [10]. BACH1 contains bZIP and BTB domains that are essential for DNA binding and heterodimerization. However, BACH1 lacks the transactivation domain and is therefore unable to support transcription [11]. Under unstressed conditions, BACH1 binds to its target sequence to repress gene expression. DNA binding is inhibited by haem, which interacts with several CP (cysteine-proline) motifs in the BTB domain [12]. The presence of haem also triggers the Crm1 (chromosome region maintenance 1)-dependent nuclear export of BACH1 [13]. It has also been proposed that oxidation of cysteine residues that are critical for DNA binding triggers dissociation of BACH1 from its target sequence [14]. Remarkably little is known about the target genes of BACH1, but both human and mouse HMOX1 have been unequivocally shown to be repressed by BACH1 [15,16].

Transcriptional regulation of the BACH1 gene itself is poorly understood. BACH1 has three different transcript variants (1, 2 and 3). Transcript variants 1 and 2 have different 5′-UTRs (untranslated regions), but they code for the same protein: isoform A. Isoform B differs from the other two by having a 5′-UTR, a 3′-coding region and a 3′-UTR that are distinctive, resulting in a shorter protein that lacks the DNA-binding properties [17]. The proposed function of the shorter isoform is to recruit the longer active isoform to the nucleus through interaction with the BTB domain. The promoter area of mouse Bach1 contains two GC boxes that bind Sp1 (specificity protein 1), and these seem to be important for the regulation of basal transcription [18]. In mouse and human hepatoma cell lines, Bach1 is induced by tBHQ (t-butylhydroquinone), which also triggers rapid nuclear export of BACH1 by phosphorylation of Tyr486 [19]. Human BACH1 mRNA is also induced by hypoxia through an unknown mechanism [20].

We have studied the effect of Nrf2 overexpression and OA-NO2 (nitro-oleic acid)-mediated activation of Nrf2 on the endothelial cell transcriptome, for the purpose of characterizing novel target genes of Nrf2 ([21] and H.K. Jyrkkänen, E. Kansanen, S.K. Häkkinen, S. Ylä-Herttuala and A. L. Levonen, unpublished work). In the course of these studies, we identified BACH1 as one of the genes potentially regulated by Nrf2. In the present study, we therefore investigated whether BACH1 is indeed a bona fide Nrf2 target gene. In the present paper, we report that adenoviral Nrf2 overexpression and Nrf2-activating compounds induce BACH1 in an Nrf2-dependent manner. Moreover, in silico analysis of the human BACH1 promoter revealed two putative AREs, located at +1411 and +1270 downstream of the TSS (transcription start site) of transcript variant 2. ARE1, located at +1411, was proven to be functional using electrophoretic mobility shift and promoter reporter assays, as well as by ChIP (chromatin immunoprecipitation). Our results provide a novel feedback-inhibitory mechanism by which Nrf2-dependent gene expression can be inhibited via transcriptional induction of BACH1.


Cell culture and Nrf2 inducers

HUVECs (human umbilical vein endothelial cells) isolated from umbilical cords obtained from the maternity ward of Kuopio University Hospital, by the approval of the Kuopio University Hospital Ethics Committee, were cultivated as described previously [22]. HEK (human embryonic kidney)-293T cells were purchased from the A.T.C.C. (Manassas, VA, U.S.A.). HEK-293T cells were maintained in Dulbecco's modified Eagle's medium (Sigma–Aldrich), supplemented with 10% (v/v) fetal bovine serum (HyClone) and 1% penicillin/streptomycin (Invitrogen-Gibco). Cloning and production of Ad-Nrf2 (Nrf2-overexpressing adenovirus) were performed as described previously [23]. Nrf2 activation was induced with SFN (sulforaphane) (Sigma–Aldrich), tBHQ (Sigma–Aldrich) or OA-NO2 (a gift from Dr Bruce Freeman, University of Pittsburgh, Pittsburgh, PA, U.S.A.) [24].

Cloning of plasmids

For the luciferase assays, BACH1-ARE1, mutated elements BACH1-M1 and BACH1-M2, BACH1-ARE2, wild-type NQO1-ARE and mutated NQO1-M1 were ligated into the pGL4 promoter vector (Promega) containing the minimal thymidine kinase promoter [25]. Complementary ARE-containing oligonucleotides having KpnI and SacI restriction sites were synthesized (by TAG Copenhagen, Copenhagen, Denmark), annealed and ligated into the vector and verified by sequencing. The sequences of the sense strands were as follows: BACH1-ARE1, 5′-CAGCTGGGGCGAATGACTCAGCAACCCAACTTTTGAGCT-3′; BACH1-M1, 5′-CAGCTGGGGCGAATAACTCAGCAACCCAACTTTTGAGCT-3′; BACH1-M2, 5′-CAGCTGGGGCGAATGACTCATTTACCCAACTTTTGAGCT-3′; BACH1-ARE2, 5′-CAAAGGCTCCCGGTGAGTCAGCTTGGAGTAGGGTGAGCT-3′; NQO1-ARE, 5′- CGATCCAGTCACAGTGACTCAGCAGAATCTGGAGCT-3′; and NQO1-M1, 5′- CGATCCAGTCACAGTAACTCAGCAGAATCTGGAGCT-3′.

RNA isolation and quantitative real-time PCR

HUVECs were transduced with Ad-Nrf2 as described in [23] and harvested 48 h after transduction for RNA extraction. Confluent HUVECs were used for experiments with SFN, tBHQ or OA-NO2. Total RNA was extracted with TRI Reagent (Sigma–Aldrich), and 1 μg of RNA was used for the cDNA synthesis using random hexamer primers (Promega) and ReverAid M-MuLV (moloney murine leukaemia virus) reverse transcriptase (Fermentas Life Sciences). The relative expression levels of mRNA encoding BACH1 transcript variants 1 and 2 (Hs00230917_m1), BACH1 transcript variant 1 (Custom TaqMan Genomic assay), BACH1 transcript variant 3 (Hs01110003_s1), BACH1 all transcript variants (Hs00895421_m1) and NQO1 (Hs00168547_m1) were measured following the manufacturer's protocol by quantitative real-time PCR (StepOnePlus™ Real-Time PCR systems, Applied Biosystems), using specific Assays-on-Demand (Applied Biosystems) target mixtures. The expression levels were normalized to the level of B2M (β2-microglobulin) (Hs00187842_m1) (Applied Biosystems).

siRNA (small interfering RNA) transfection

HUVECs were transfected with siRNA oligonucleotides targeting Nrf1 or Nrf2, or a non-specific siRNA control (20 nmol/l, Invitrogen) as described in [26]. At 48 h after transfection, cells were treated with Nrf2-activating agents for 6 h and mRNA levels of BACH1, NFE2L1, NFE2L2 and NQO1 were measured with quantitative real-time PCR.

Luciferase reporter gene assay

Activation of BACH-ARE or mutated elements was measured using a luciferase reporter assay. NQO1-ARE and NQO1-M1 were used as controls. HEK-293T cells grown in 96-well plates were transfected by using a calcium phosphate method with 100 ng of DNA/well. BACH1-ARE or NQO1-ARE was transfected together with β-galactosidase expression vector (pCMVβ; Invitrogen). In addition, pCI-Nrf2 and p3xFLAG-Keap1 plasmids were co-transfected in some experiments. Cells were treated with 10 μM SFN, 50 μM tBHQ or 10 μM OA-NO2 24 h after transfection for 16 h and then the luciferase activity was measured using the PerkinElmer luciferase assay system according to the manufacturer's protocol. Luciferase activity was normalized to β-galactosidase activity as described previously [27].

EMSA (electrophoretic mobility-shift assay)

For EMSA, the DNA-binding region of Nrf2 (amino acids 327–605) was amplified by PCR from pCI-Nrf2neo [28] and cloned into the KpnI and XbaI restriction sites of the pcDNA3 vector, with a Kozak sequence. MafG cDNA (German Research Center for Genome Research, Berlin, Germany) was PCR-amplified and cloned into the HindIII and EcoRI restriction sites of pcDNA3 vector with a Kozak sequence. The following primers were used for cloning: Nrf2 sense, 5′-AATGGTACCGCCACCATGGCTTTCAACCAAAACCACCC-3′; Nrf2 antisense, 5′-AAATCTAGACTAGTTTTTCTTAACATCTGGC-3′; MafG sense, 5′-AATAAGCTTGCCACCATGACGACCCCCAATAAAGG-3′; and MafG antisense, 5′-AAAGAATTCCTACGATCGGGCATCCGT-3′. Nrf2 and MafG proteins were generated by coupled in vitro transcription/translation using their respective pcDNA3-based cDNA expression constructs and TNT® Quick Coupled Transcription/Translation kit as recommended by the supplier (Promega). The proteins were incubated for 10 min in a total volume of 20 μl of binding buffer (150 mM KCl, 1 mM dithiothreitol, 25 ng/μl herring sperm DNA, 5% glycerol and 10 mM Hepes, pH 7.9). 32P-labelled double-stranded oligonucleotides containing one copy of the respective response element were added to the mixture, and incubation was continued for 15 min at room temperature (20°C). Protein–DNA complexes were resolved by electrophoresis through 8% non-denaturing polyacrylamide gels (19:1 mono-/bis-acrylamide ratio) in 0.5×TBE (45 mM Tris, 45 mM boric acid and 1 mM EDTA, pH 8.3) for 100 min at 200 V and quantified on a FLA-3000 reader (Fuji) using ScienceLab99 software (Fuji). In the competition experiments, unlabelled DNA at approximately 10/50/100-fold molar excess was pre-incubated with protein samples for 10 min at 4°C before the addition of the labelled probe. The reaction mixtures were incubated for an additional 15 min with the labelled probe and resolved on 8% non-denaturing polyacrylamide gels.

Chromatin immunoprecipitation

ChIP was performed as described previously [26], with modifications. HEK-293T cells were treated with 10 μM SFN or DMSO as a control for 2 h. Nuclear proteins were cross-linked to DNA by adding formaldehyde directly to the medium to a final concentration of 1% and incubating for 10 min at room temperature on a rocking platform. Cross-linking was stopped by adding glycine to a final concentration of 0.125 M and incubating for 5 min at room temperature on a rocking platform. Medium was removed, and the cells were washed twice with ice-cold PBS. The cells were collected and lysed with 1 ml of SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris/HCl, pH 8.1, and protease inhibitors) and incubated at room temperature for 10 min. Lysates were sonicated using a Bioruptor UCD-200 instrument (Diagenode) to result in DNA fragments of 200–1000 bp in length. Cellular debris was removed by centrifugation at 20000 g for 15 min at 4°C. After centrifugation, 20 μl of each sample was separated for input control and the remaining sample was diluted 1:10 in ChIP dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 167 mM NaCl, 16.7 mM Tris/HCl, pH 8.1, and protease inhibitors) and divided into 100 μl aliquots for immunoprecipitation. Then, 2.5 μl of BSA (100 mg/ml) was added to each aliquot. Chromatin solutions were incubated overnight at 4°C on a rocking platform with 3 μg of an Nrf2-specific antibody (sc-722; Santa Cruz Biotechnology), 10 μg of a MafF/G/K-specific antibody (sc-22831; Santa Cruz Biotechnology), 10 μg of an antibody against BACH1 (sc-14700; Santa Cruz Biotechnology) or 1 μg of non-specific IgG (anti-rabbit IgG, Upstate Biotechnology). The immunocomplexes were collected with 20 μl of Magna ChIP Protein A Magnetic Beads (Millipore) for 1.5 h at 4°C with rotation. The beads were separated with a magnetic rack and washed sequentially for 3 min with 700 μl of the following buffers: low-salt wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 150 mM NaCl and 20 mM Tris/HCl, pH 8.1), high-salt wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 500 mM NaCl and 20 mM Tris/HCl, pH 8.1) and LiCl wash buffer (0.25 M LiCl, 1% Nonidet P40, 1% sodium deoxycholate, 1 mM EDTA and 10 mM Tris/HCl, pH 8.1). Finally, the beads were washed twice with 700 μl of TE buffer (10 mM Tris/HCl, pH 8.1, and 1 mM EDTA). The immunocomplexes were then eluted by adding 500 μl of elution buffer (100 mM NaHCO3 and 1% SDS) and incubating for 30 min at room temperature. Proteins were digested from the eluate by adding 2 μl of proteinase K (934 units/ml, Fermentas) and incubating overnight at 64°C. DNA was recovered by phenol/chloroform/3-methylbutan-1-ol (25:24:1, by vol.) extractions and precipitated with 1:10 volume of 3 M sodium acetate buffer (pH 5.2) and 2 volumes of ethanol using glycogen as a carrier. Immunoprecipitated chromatin DNA was then used as a template for real-time quantitative PCR.

PCR of chromatin templates

Real-time quantitative PCR of ChIP templates was performed using chromatin region-specific primers of NQO1 (primers 5′-TCCAAATCCGCAGTCACAGT-3′ and 5′-TTGGCACGAAATGGAGC-3′), BACH1 (5′-TCAGCCTTCTGAAGCAACCTC-3′ and 5′-TTGGAGTAGGGTGACTTTGGC-3′), HMOX1 (5′TGAGTAATCCTTTCCCGAGC-3′ and 5′-GTGACTCAGCGAAAACAGACA-3′) and actin (5′-AACTCTCCCTCCTCCTCTTCCTC-3′ and 5′-GAGCCATAAAAGGCAACTTTCGG-3′), and Maxima™ SYBR Green/ROX qPCR Master Mix in a total volume of 10 μl in a LightCycler® 480 System. The PCR cycling conditions were: pre-incubation for 10 min at 95°C, 40 cycles of denaturation for 20 s at 95°C, annealing for 20 s at 58°C, elongation for 20 s at 72°C and a final elongation for 5 min at 72°C. The PCR products were also resolved on 2% agarose gels to control for correct product size. Relative association of chromatin-bound protein was calculated using the formula 2−(ΔCt)×100, where ΔCt is Ct(output)Ct(input), output is the immunoprecipitated DNA and input is the purified genomic DNA from the starting material of the ChIP assay. IgG control values were subtracted from output values.

Statistical analysis

Statistical analyses were performed with GraphPad Prism Software. Statistical significance was evaluated by the unpaired Student's t test or one-way ANOVA using Bonferroni's post hoc comparisons. Results are expressed as means±S.E.M. and differences are considered significant for P<0.05.


BACH1 mRNA expression is induced by Nrf2 overexpression and Nrf2-inducing agents

Our previous study using genome-wide expression profiling indicates that BACH1 mRNA is induced by the Nrf2-activator OA-NO2, suggesting that BACH1 is a transcriptional target of Nrf2 [21]. In order to verify this finding, HUVECs were transduced with Ad-Nrf2 with MOIs (multiplicities of infection) of 10–100, and BACH1 mRNA was measured 64 h after transduction. Initially, we used three different assays: an assay that measures both transcript variants 1 and 2, and assays for transcript variant 1 and transcript variant 3. However, the expression of transcript variant 1 was not measurable (results not shown) and we therefore concluded that the assay for both variants measures transcript variant 2 in HUVECs. Both BACH1 transcript variants 2 and 3 were significantly induced by AdNrf2 with MOIs of 50 and 100 (Figure 1A). Also, three structurally different Nrf2-activating agents, SFN, tBHQ and OA-NO2, increased BACH1 mRNA at 6 h (Figures 1B, 1C and 1D). Using 10 μM SFN as the inducer, the maximum induction was observed at 6 h and the mRNA levels remained elevated for 24 h (Figure 1E). In addition, we examined whether the increased BACH1 mRNA expression by SFN or OA-NO2 was a consequence of increased transcription of mRNA. For this, HUVECs were pre-treated for 1 h with a transcriptional inhibitor actinomycin D (2 μg/ml) or DMSO as a solvent control. SFN or OA-NO2 was then added and incubation was continued for 6 h. SFN and OA-NO2 significantly induced BACH1 mRNA in cells incubated with DMSO, whereas pre-treatment with actinomycin D abolished induction (Figure 1F).

Figure 1 Adenoviral Nrf2 overexpression and Nrf2 activation induce BACH1 mRNA

(AE) HUVECs were transduced with Ad-Nrf2 using MOIs of 10–100 for 64 h (A), or were treated with SFN (B), tBHQ (C) or OA-NO2 (D) for 6 h or with 10 μM SFN for 0–24 h (E). (F) After 1 h of pre-treatment with actinomycin D (2 μg/ml) or DMSO, HUVECs were treated with 10 μM SFN or 5 μM OA-NO2 for 6 h. BACH1 mRNA expression was measured with quantitative real-time PCR and normalized to B2M expression. Results are means±S.E.M. (n=3). *P<0.05, **P<0.01, ***P<0.001 compared with control. ¤P<0.05, ¤¤¤P<0.001, DMSO compared with actinomycin D.

Identification of a functional ARE-binding site in the BACH1 promoter

Considering BACH1 was effectively induced by Nrf2 overexpression and Nrf2-inducing agents, we next sought to identify whether BACH1 is a direct target of Nrf2. We therefore searched for putative ARE-binding sites from the BACH1 gene in close proximity to all three BACH1 transcript variant TSSs by in silico analysis performed with the MatInspector software (Genomatix), using a position weight matrix generated by combining 57 functional ARE sequences reported previously [6]. This analysis revealed two putative AREs within the first intron of the BACH1 transcript variants 1 and 2, and in the promoter region of transcript variant 3. These two variants that are located +1411 (ARE1) and +1270 (ARE2) from the TSS of transcript variant 2 are phylogenetically conserved (Figure 2). ARE2 is closely similar to the variant ARE described by Nioi et al. [5].

Figure 2 BACH1 gene structure, sequence and promoter alignment

(A) BACH1 gene structure. BACH1 has three transcript variants (1–3). Transcript variants 1 and 2 have the same exon structure and code for the same protein. Transcript variant 3 lacks the first exon and has two more exons at the 3′-end in comparison with the others. Location of the two putative AREs in the BACH1 gene are presented. (B) Sequences of the putative BACH1 AREs were aligned with the ARE consensus sequence [3] and the NQO1-ARE sequence. The BACH1 ARE1 is in reverse orientation. (C) Human BACH1 promoter sequence aligned with the mouse sequence.

In order to examine the functionality of the two putative binding sites in the human BACH1 gene, the binding of Nrf2 to ARE1 and ARE2 was first analysed by EMSA. An ARE from a well-characterized Nrf2 target gene, NQO1, was used as a positive control. The binding of the MafG homodimer and Nrf2–MafG heterodimer to the BACH1-ARE1 was similar to that of NQO1-ARE, whereas BACH1-ARE2 did not bind to the complex (Figure 3A). The specificity of Nrf2–MafG binding to the BACH1-ARE1 was studied by using mutated BACH1-ARE1 oligonucleotides BACH1-M1 and BACH1-M2 (Figure 3B). Mutations were directed to the nucleotides known to be important for Nrf2–MafG binding [3] (Figure 3C). The Nrf2–MafG complex failed to bind either BACH1-M1 or BACH1-M2 (Figure 3B, lanes 8 and 12). Furthermore, binding of the Nrf2–MafG complex to the BACH1-ARE1 was inhibited with increasing concentrations of unlabelled NQO1-ARE (Figure 3D, lanes 3–5), whereas NQO1-M1 or a non-specific competitor [NF-κB (nuclear factor κB)-binding element] were not able to compete with BACH1-ARE1 binding (Figure 3D, lanes 6–11).

Figure 3 Nrf2–MafG protein complex binds to the human BACH1 ARE

(A) Labelled oligonucleotides were incubated with in vitro-translated MafG (lanes 2, 6 and 10), Nrf2 (lanes 3, 7 and 11) or both (lanes 4, 6 and 12). (B) 32P-labelled BACH1-ARE1 or ARE1 mutations M1 or M2 (sequences in C) were incubated with in vitro-translated MafG (lanes 2, 6 and 10), Nrf2 (lanes 3, 7 and 11) or both (lanes 4, 8 and 12). (D) 32P-labelled BACH1-ARE was incubated with increasing concentrations of unlabelled NQO1-ARE (lanes 3–5), NQO1-M1 (lanes 6–8) or NF-κB element (lanes 9–11).

To study the functionality of the two putative BACH1 AREs in a cellular context, the oligonucleotides that were used in EMSA were cloned into the luciferase vector and the effects of Nrf2-activating agents were tested, using NQO1-ARE as a positive control. Luciferase activity of BACH1-ARE1 and NQO1-ARE was increased by all agents used (Figure 4A). The basal activity of BACH1-ARE2 was substantially lower, but increased by inducing agents (Figure 4A). Of note, the basal and inducible activity of the BACH1-ARE1 was substantially lower than that of the NQO1-ARE control. Moreover, the BACH1-ARE1 harbouring the G/A mutation within the core ARE sequence (mutation M1; Figure 3C) and respective (G/T) mutation within the NQO1 ARE greatly attenuated both the basal as well as the inducible ARE activity (Figure 4A). In addition to Nrf2-inducing agents, Nrf2 overexpression also increased the activity of BACH1-ARE1 as well as NQO1-ARE as a control (Figures 4B and 4C). This increase was attenuated by Keap1 co-transfection and restored by SFN and OA-NO2 treatment, further affirming the notion that BACH1-ARE1 is functionally active.

Figure 4 Nrf2 co-transfection or Nrf2 activators increase human BACH1-ARE activity

(A) HEK-293T cells were transfected with wild-type or mutated BACH1-ARE1 or BACH1-ARE2 cloned into pGL4 promoter vectors. The cells were treated with SFN, tBHQ or OA-NO2 24 h after transfection, and luciferase activity was measured 16 h thereafter. NQO1-ARE and NQO1-M1 were used as controls. (B and C) HEK-293T cells were transfected with BACH1-ARE1 (B) or NQO1-ARE (C) constructs with Nrf2- and Keap1-expressing plasmids. At 24 h after transfection, cells were treated with Nrf2-activating agents for 16 h and luciferase activities were measured. An empty pGL4 promoter vector was used as a control in all experiments. Luciferase activities were normalized to β-galactosidase activity. Results are shown relative to control and are means±S.E.M. (n=5–6). *P<0.05 compared with the relative control.

To gain insight into the relevance of the findings in EMSA and reporter assays in intact cells, binding of Nrf2 and MafG with BACH1-ARE1 was studied by ChIP. The binding of Nrf2 to the BACH1-ARE1 significantly increased when HEK-293T cells were treated with SFN (Figure 5A). In line with the reporter assay data, both basal as well as inducible binding of Nrf2 to NQO1-ARE relative to input control was substantially higher than to BACH1-ARE1 (Figure 5A). ChIP using an anti-Maf antibody recognizing all three small Mafs (G/F/K) indicated that small Mafs bind to both BACH1-ARE1 and NQO1-ARE with no significant increase in binding in the presence of SFN (Figure 5B). In order to examine whether BACH1 binds to its own promoter, we performed ChIP using an antibody against BACH1 and the distal enhancer element in the HMOX1 promoter as a positive control [16,29]. Whereas there was significant binding to the distal enhancer region of the HMOX1 gene (0.68±0.10% enrichment relative to input control under basal conditions; n=3), there was no detectable binding of BACH1 to the ARE of the BACH1 gene, suggesting that BACH1 does not regulate its own transcription.

Figure 5 Binding of Nrf2 and MafG to the ARE of human BACH1 and NQO1

(A and B) HEK-293T cells were treated with 10 μM SFN for 2 h and ChIP was performed using anti-Nrf2 (A) and anti-MafF/G/K (B) antibodies. Real-time quantitative PCR was performed using primers specific for an ARE of BACH1, NQO1 and actin genes. Binding is depicted as the percentage of input values from three independent experiments. Results are means±S.E.M. (n=3). *P<0.05 compared with the control treatment.

Induction of BACH1 expression is Nrf2-dependent

Finally, we examined the differential role of Nrf1 and Nrf2 in the regulation of BACH1, as both factors have been shown to have a role in the regulation of ARE-dependent genes [30,31]. To this end, we used the siRNA approach (Figure 6). The expression of NFE2L2 (Nrf2) mRNA was reduced by 83% and NFE2L1 (Nrf1) mRNA by 89% with respective siRNAs under basal conditions (Figures 6A and 6B). Transfection with siNrf2, but not siNrf1, totally abolished both basal as well as inducible expression of NQO1 mRNA, used as a positive control (Figure 6C). SFN treatment induced BACH1 mRNA expression 1.8±0.2 fold (n=3) in control siRNA-transfected cells (Figure 6D). Transfection with siNrf1 had no impact on BACH1 mRNA induction by SFN, whereas siNrf2 transfection abolished inducible expression (86.5±20.4% increase by SFN treatment in siControl compared with a 14.5±9.5% increase in siNrf2-transfected cells; P=0.03) (Figure 6C). Basal BACH1 expression was not significantly affected by either siNrf1 or siNrf2, suggesting that they do not play a role in the regulation of BACH1 under basal conditions.

Figure 6 SFN-induced BACH1 expression is Nrf2-dependent

HUVECs were transfected with control, Nrf1 or Nrf2 siRNA. At 48 h after transfection, cells were treated with SFN for 6 h and mRNA levels of NFE2L2 (A), NFE2L1 (B) NQO1 (C) and BACH1 (D) genes were measured with quantitative real-time PCR. The results are normalized to B2M and are means±S.E.M. (n=3). (A and B) Results are depicted as fold change compared with control siRNA-transfected untreated cells (#P<0.05) or with control siRNA-transfected SFN-treated cells (¤P<0.05). (C and D) Results are depicted as fold change of untreated cells compared with treated cells (*P<0.05).


BACH1 is a transcriptional repressor that is regulated at multiple levels. Haem binds to BACH1 and inhibits its DNA-binding activity and induces nuclear export and proteasomal degradation [12,32]. Furthermore, BACH1 protein translation is attenuated by specific microRNAs of endogenous or viral origin [3335]. In the present study, we have shown that the BACH1 gene is transcriptionally induced by Nrf2, which adds yet another layer of regulation to BACH1 expression.

To our knowledge, ours is the first study on the transcriptional regulation of the human BACH1 gene that identifies a specific transcription factor playing a regulatory role. Hypoxia has been shown to increase BACH1 mRNA with a concomitant decrease in HMOX1 expression, and it has been postulated that the human gene is regulated by hypoxia-response element(s) that are not present in other species [20]. However, such elements have not been identified. Also, a recent study by Kaspar and Jaiswal [19] that studied the nuclear export mechanisms of BACH1 in human hepatoblastoma HepG2 cells after treatment with tBHQ showed that, after an initial decline, nuclear levels of BACH1 protein are restored in 4 h via de novo synthesis. They also show that BACH1 mRNA expression is increased by tBHQ in a manner inhibitable by actinomycin D [19], suggesting that increased transcription is required. Our findings are in line with this previous study and identify the mechanism by which tBHQ affects BACH1 expression.

We found a functional ARE located in the intronic region, +1411 nucleotides from the TSS of transcript variant 2. Although, to our knowledge, this is the first demonstration of a functional ARE within an intron, enhancer elements are frequently found remote from the proximal 5′-promoters of genes [36]. Factors that bind to these more distal regions regulate gene expression via a looping mechanism, in which they form protein–protein interactions with the general transcriptional machinery bound at the TSS. Notably, data derived from the Encyclopedia of DNA Elements (ENCODE, project reveals the presence of open chromatin using FAIRE-seq (formaldehyde-assisted isolation of regulatory elements coupled with high-throughput sequencing) in HUVECs and clustering of markers of active chromatin at the vicinity of the identified ARE sequence [37] (see Supplementary Figure S1 at In addition, binding of other transcription factors was demonstrated downstream of the TSS by ChIP-sequencing, supporting further the notion that this intron region is a highly active regulatory region.

In the present study, we also examined the potential role of Nrf1 in regulating BACH1 expression. Nrf1, like other CNC-bZIP transcription factors, heterodimerizes with small Maf proteins and binds to the ARE sequence. Although Nrf1 and Nrf2 have been reported to have partially overlapping roles in the regulation of certain ARE-dependent genes such as the glutathione synthetic enzymes [38], their functions are not entirely redundant. Gene targeting of Nrf1 [39] and Nrf2 [40] in mice yields an entirely different phenotype. Moreover, gene expression profiling of hepatocyte-specific Nrf1-deficient mice revealed a distinct set of Nrf1-specific genes, such as metallothioneins-1 and -2 [30]. In the present study, despite a robust reduction in Nrf1 mRNA expression, neither basal nor inducible BACH1 and NQO1 mRNA expression was significantly affected by siNrf1 (Figure 6), indicating a non-redundant role of Nrf2 and Nrf1 in regulating these two genes in human endothelial cells.

Nrf2-dependent gene activation (and its inhibition by Keap1) has been a subject of intense study and it is relatively well characterized. However, little is known about the other mechanisms by which ARE-dependent gene expression is attenuated. BACH1 binds to the ARE-like elements within the HMOX1 promoter and represses the transcription of both human and mouse HMOX1 genes [41]. Derepression of BACH1-mediated inhibition of haem oxygenase 1 transcription temporally precedes Nrf2 binding and is a prerequisite for the induction of HMOX1 by Nrf2 [16,42]. Whether BACH1 affects the expression of other ARE-dependent genes is a matter of debate. ARE-dependent transcription of NQO1 [43] as well as ferritin and thioredoxin reductase [44] have been reported to be down-regulated by BACH1. However, in studies in which BACH1 was silenced by siRNA and the gene expression was examined either by microarray analysis or quantitative PCR, little or no effect on the expression of genes other than HMOX1 was found [42,45]. In the mouse Hmox1 gene, all three ARE-like sequences of the distal enhancer located approximately 9 kb upstream of TSS need to be intact for BACH1-mediated repression, and it has therefore been suggested that multiple ARE-like elements are needed for the binding of BACH1 [46]. The mechanism of this is not known, but it has been postulated that homo-oligomerization of BACH1 clustered in multiple ARE-like elements is necessary for efficient repression.

In conclusion, we have shown that BACH1 is a bona fide Nrf2 target gene regulated by an intronic ARE located at +1411 from the TSS of transcript variant 2. We propose that, after initial nuclear export and degradation of BACH1 protein, de novo synthesis via Nrf2-dependent transcriptional induction of BACH1 provides a feedback mechanism by which nuclear BACH1 levels are restored.


Henna-Kaisa Jyrkkänen designed and performed the experiments, analysed the data and wrote the paper. Suvi Kuosmanen, Heidi Laitinen, Emilia Kansanen and Hanna Leinonen performed the experiments. Merja Heinäniemi and Eero Mella-Aho provided data and genomic sequence analyses. Seppo Ylä-Herttuala provided research materials and supervised the writing of the paper. Anna-Liisa Levonen designed the experiments, analysed the data and wrote the paper.


This work was supported by the Academy of Finland (A.-L.L.), the Sigrid Juselius Foundation (A.-L.L.), the Finnish Cultural Foundation (H.-K.J. and A.-L..L), the Finnish Foundation for Cardiovascular Research (H.-K.J. and A.-L.L.), the Finnish Medical Foundation (H.-K.J.), the Foundation of Aarne and Aili Turunen (H.-K.J.) and the Finnish Cancer Organizations (A.-L.L.).


We thank Dr Sami Väisänen (Department of Biosciences, University of Eastern Finland) for the pGL4 promoter vector and Dr Bruce A. Freeman and Dr Steven Woodcock (both at Department of Pharmacology and Chemical Biology, University of Pittsburgh, Pittsburgh, PA, U.S.A.) for the gift of OA-NO2.

Abbreviations: Ad-Nrf2, Nrf2-overexpressing adenovirus; ARE, antioxidant-response element; B2M, β2-microglobulin; BTB, broad-complex, tramtrack and bric-a-brac; bZIP, basic-region leucine zipper; ChIP, chromatin immunoprecipitation; CNC, cap'n'collar protein; BACH, BTB and CNC homology; EMSA, electrophoretic mobility-shift assay; HEK, human embryonic kidney; HUVEC, human umbilical vein endothelial cell; Keap1, Kelch-like ECH (erythroid cell-derived protein with CNC homology)-associated protein 1; MOI, multiplicity of infection; NF-E2, nuclear factor-erythroid 2; NF-κB, nuclear factor κB; Nrf, NF-E2-related factor; OA-NO2, nitro-oleic acid; SFN, sulforaphane; siRNA, small interfering RNA; tBHQ, t-butylhydroquinone; TSS, transcription start site; UTR, untranslated region


View Abstract