DNA and mRNA elements with complementary responses to hemin, antioxidant inducers, and iron control ferritin-L expression

Experimental Procedures

Design and Construction of Reporters. Human ferritin-L promoter. Promoter sequences were determined by scanning genomic sequences upstream of the known transcription start site. The putative ferritin-L promoter was cloned by PCR using sequence-specific PCR primers containing Mlu1 and XhoI restriction sites, and chromosome 19 BAC (GenBank accession no. AC026803.7) as the DNA template. The PCR product was blunt-end cloned into pSTBLUE1 (Novagen), digested with Mlu1 and XhoI, gel purified, then cloned into pGL3-Basic (Promega) Mlu1 and XhoI sites to create vector pFTL-promoter-LUC. The same cloning strategy was used to generate a series of 0.2-kb nested deletions by PCR using the original 1.4-kb construct as a template, in addition to a construct missing that 0.2 kb proximal to the luciferase coding region that contained the TATA sequence. A MARE/ARE mutation (TGAC/GACA) was created by using Quikchange XL Site-Directed Mutagenesis kit (Stratagene) to create pFTL-MARE/ARE^m-LUC. Cloning of the TRR and QR LUC reporters, each with a functional ARE, has been described (9).

pFTLUC plus FTL-5′ UTR mRNA. The entire 5′ UTR region of the human ferritin-L gene (+49 to +227; GenBank accession no. X03742), which contains the IRE, was amplified by PCR using primers to incorporate XhoI and NcoI restriction sites and a chromosome 19 BAC (GenBank accession no. AC026803.7) as DNA template. The PCR product was subcloned into pSTBLUE1. XhoI and NcoI fragments were recovered and then inserted between the 3′ end of the ferritin-L promoter and the luciferase start codon of pFTLUC to create pFTL- +IRE-LUC, a chimera where luciferase synthesis is under the control of both the ferritin-L promoter (MARE/ARE) and mRNA (IRE) regulator. The 5 bp inserted as the XhoI site, upstream of the 5′UTR, is a region of variable length and sequence in ferritin mRNA (44). This strategy was selected because attempts to amplify the entire ferritin-L promoter and noncoding, 5′ ferritin-L mRNA sequences failed under a variety of experimental conditions. All plasmid sequences were confirmed by sequencing the DNA.

Cell Culture and Luminescence Assay. HepG2 (ATCC HB-8065) or Caco2 (ATCC HTB-37) cells were seeded in collagen coated, 12-well plates (160,000 cells per well) by using MEM media with 10% FBS (Atlanta Biologicals) and 100 units/ml penicillin/streptomycin (Gibco), and cultured in 95% ambient air/5% CO₂. One day after seeding, cells were cotransfected (FuGENE 6 Transfection Reagent, Roche Molecular Biochemicals) with pRLSV40 Renilla luciferase, a housekeeping control for transfection (Promega), and test reporter constructs using DNA ratios suggested by manufacturer. One day after transfection, cells were treated with experimental media. SF, TBHQ, hemin, protoporphyrin IX (PPIX), hydrogen peroxide, FAC, and DMSO were obtained from Sigma. SF, TBHQ, hemin and PPIX were dissolved in DMSO, and FAC was dissolved in water. All test media, including that for control and FAC-treated cells, contained a final concentration of 0.1% DMSO. Cells were harvested 24 h after treatment and assayed for firefly and Renilla luciferase (DuaLuciferase kit; Promega) on a Turner 20/20 luminometer. Data are the ratio of firefly to Renilla luciferase activity.

Quantitative PCR. HepG2 cells were seeded in six-well plates (400,000 cells per well) and treated with DMSO (control) or DMSO with hemin (80 μM) 48 h after seeding. Twenty-four hours later, total mRNA was isolated from the cells by using the RNeasy Mini kit (Qiagen, Valencia, CA). cDNA was prepared from total RNA by using the High-Capacity cDNA Archive kit (Applied Biosystems) according to the manufacturers instructions; mRNA concentrations of ferritin-L, ferritin-H, heme-oxygenase-1, and β-actin were assessed by using TaqMan Gene Expression Assay kits (Applied Bioscience) and an Applied Biosystems 7900 Sequence Detection system (Applied Bioscience) according to manufacturer's instructions. Differences in mRNA concentrations between hemin-treated and control cells were quantified by the cycles to fluorescence midpoint cycle threshold calculation (2^{–[ΔCt experimental gene–ΔCt housekeeping gene]}), using β-actin as the housekeeping gene.

Results

A consensus MARE element, embedded in an ARE element in the human ferritin-L gene, 1.35 kb upstream from the transcription start site (Fig. 1A), was identified in a search of human genome sequences (chromosome 19; 19q13.3-q13.4). MARE/ARE combination elements (TGAG/CTCA) are also found in the promoters of heme responsive genes such as β-globin (15), heme oxygenase (16), and QR (17). A search of sequences upstream of the human ferritin-H gene also revealed an MARE/ARE at –4.4 kb, possibly analogous to the reported murine ferritin-H gene ARE (–4.13 kb) that is regulated by the antioxidant response inducers oltipraz and TBHQ (18).

The putative 1.4-kb ferritin-L promoter containing the MARE/ARE sequence, was cloned into a luciferase reporter and characterized by deletion analysis (Fig. 1B), followed by construction of a reporter containing both the ferritin-L DNA promoter and the mRNA IRE sequence (Fig. 1C). The ferritin-L MARE/ARE sequence regulated expression even under normal culture conditions, based on the requirement of wild-type MARE/ARE sequence for maximum luciferase expression in HepG2 cells (Fig. 1B). Apparently, some MARE/ARE responses are occurring at all times during culture. Deletion of 200 bp, immediately adjacent to the transcription start site and containing the TATA box, completely ablated luciferase activity, as previously observed with a CAT reporter and deletion of sequences to –260 bp (19). Sequences between the MARE/ARE and TATA box, –0.2 to –0.6 kb, had increased luciferase activity relative to the 1.4-kb mutated promoter or deletions of the ARE in promoter fragments between –1.2 and –0.6 (Fig. 1B). The simplest explanation of the variation in activity among the MARE/ARE deletion constructs of different promoter length is the presence of ARE-independent, positive control elements between –0.2 and –0.6 kb, although an alternative explanation is the presence of negative control elements between –0.6 and –1.2 kb that the MARE/ARE overrides.

High sensitivity of the human ferritin-L promoter (1.4 kb) to an inducer of antioxidant responses (SF) contrasted with low sensitivity to iron (FAC) (Fig. 1D); equivalent responses required only 2 μM SF compared to 1,600 μM FAC (Fig. 1D), an 800-fold difference; only ≈100–200 μM FAC is needed to induce ferritin protein synthesis (24). The relative insensitivity of ferritin-L transcription to FAC has also been observed in whole animal experiments (20–22). Hydrogen peroxide (200 μM) had no effect on luciferase activity of the human ferritin-L promoter construct, extending the earlier results observed earlier with endogenous mouse ferritin-L mRNA by Torti and colleagues (12). Hemin induced similar responses in the ferritin-L promoter compared with those in the QR and TRR reporters, which contain ARE promoter sequences (Fig. 2). SF and TBHQ also induced similar responses for all three promoters (23) (Fig. 2). PPIX (heme without Fe) and hemin had similar effects on the ferritin-L promoter (Fig. 2C), indicating that the porphyrin ring, not Fe, was the major recognition feature for ferritin-L promoter regulation.

Effects of combining DNA (MARE/ARE) and RNA (IRE) regulatory elements were determined with a chimeric reporter vector containing both elements (Fig. 3 A–C). When the DNA-MARE/ARE and the mRNA-IRE were both present, hemin increased luciferase activity ≈9-fold (Fig. 3B), consistent with endogenous ferritin-L expression in hemin-treated Friend's leukemia cells (24) and contrasting with the 2- to 3-fold hemin or PPIX induction with the DNA-MARE/ARE alone (Fig. 3A) and when the IRE was coupled to an inactive MARE/ARE (Fig. 3D). Hemin also increased the concentration of endogenous ferritin-L mRNA ≈3-fold and heme oxygenase-1 mRNA ≈10-fold, measured by RT-PCR and normalized to β-actin (Fig. 4); in a single experiment, ferritin-H mRNA also increased ≈3-fold. Mutation of the MARE/ARE (TGAC/GACA) abrogated the hemin effect on the ferritin-L promoter (Fig. 3C) and, with the IRE, gave equal responses to FAC or hemin (Fig. 3D), confirming earlier observations that hemin inhibits IRE/IRP interactions (25) and that FAC and hemin both increase ferritin synthesis (24). The response to hemin of the combined MARE/ARE and IRE regulatory elements (Fig. 3B) was significantly higher (P < 0.05) than the sum of the effects for the MARE/ARE (Fig. 3A) and IRE (Fig. 3D) individually. The large effects of hemin when the ARE/MARE plus IRE are present (Fig. 3B), the decreased effect of hemin when the MARE/ARE is mutated (Fig. 3 C and D), and the similarity of the hemin, TBHQ, and PPIX effects on the MARE/ARE minus the IRE (Fig. 3A) show that hemin is affecting both the MARE/ARE and IRE pathways.

Discussion

Ferritin-L plays an important role in protection against oxidative damage, based on the similar responses of the ferritin-L DNA promoter and the promoters of antioxidant response proteins, such as QR and TRR, all of which have ARE sequences (Figs. 1A and 2), (8, 9, 26). The ARE sequence in the ferritin-L promoter overlaps with an MARE sequence (Fig. 1A). MARE and ARE elements are both members of the AP-1 response family that bind leucine zipper “cap n' collar” factors to regulate transcription (8, 27). The presence of MARE in the ferritin-L and -H ARE sequences (Fig. 1A) and regulation of ferritin-L by hemin (Fig. 2C), which also regulates heme oxygenase 1 (16) and β-globin (15), indicates that ferritin may also influence cellular dioxygen metabolism, possibly through the entrapment of dioxygen in the mineral. Ferritin-H transcription, in contrast to ferritin-L, is induced by hydrogen peroxide activation (12), which may relate to the release of hydrogen peroxide associated with ferritin-H catalytic activity (28, 29). Other observations emphasizing the importance of ferritin in cell functions beyond storing and concentrating iron include: induction by c-myc and TNF-α (ferritin-H) (13, 18), RNA interference knockdown effects on cell growth and division (ferritin-L) (30), effects of MARE/ARE mutation on basal expression (ferritin-L) (Fig. 1B), and induction by TBHQ (ferritin-H and -L) (Figs. 1D, 2C, and 3 A and B) (11, 12). Hemin effects equivalent to those of TBHQ, SF, and metal-free protoporphyrin IX (Figs. 1D, 2C, and 3 A and B) or co-PPIX (31) all emphasize iron-independent regulation of the ferritin-L DNA ARE/MARE.

Antioxidant inducers and iron appear to selectively target the MARE/ARE (DNA) and IRE (mRNA), respectively, with a low level of cross-talk for ferritin-L with high concentrations of FAC (Fig. 1D) and, for ferritin-H, with hydrogen peroxide (12). Hemin, as both an iron complex and antioxidant inducer (32), targets the ferritin-L DNA-MARE/ARE as an antioxidant inducer (Figs. 2 and 3) and the mRNA-IRE as an iron complex (Fig. 3D), with effects significantly (P < 0.05) greater than the sum of the antioxidant and iron responses (Fig. 3). Direct binding of hemin to the DNA and mRNA repressors is a possible mechanism of the enhanced hemin effects when both DNA and mRNA regulators are present. Hemin is known to bind to both IRP 1 and IRP 2, the ferritin-L mRNA repressors, and Bach 1, a DNA transcription repressor for MARE/ARE genes (33–35). Little is known about the molecular effects of hemin binding to IRP proteins, other than changing protein stability (33, 36). In contrast, it is known that the Bach 1/hemin interaction relieves Bach 1 inhibition of Maf/Nrf2 interactions with MARE/ARE (5′-TGAC/GTCA-3′) DNA sequences (Fig. 1 A) to enhance transcription (35), as shown for QR (17) and heme oxygenase 1 (35, 37), where the hemin response is comparable to ferritin-L (Figs. 2 and 4). PPIX, which binds to proteins such as apomyoglobin (38–40), could also target Bach 1. In addition to the possible direct effects of hemin on ferritin-L DNA and mRNA repressors, the several signaling molecules released during hemin catabolism, such as biliverdin/bilirubin carbon monoxide and iron (37, 41, 42), may also contribute to ferritin-L regulation.

Dual regulation of ferritin-L by a transcriptional regulator (MARE/ARE) sensitive to antioxidant inducers and hemin, linked to a translational regulator (IRE), sensitive to iron and hemin, connects ferritin-L gene regulation to three groups of genes at the intersection of iron and oxygen metabolism: (i) oxygen metabolism, e.g., β-globin and heme oxygenase; (ii) oxidative stress, e.g., TRR and QR (9, 43); and (iii) iron metabolism, e.g., ferroportin, DMT1 and mitochondrial aconitase (4, 44). Cross-talk among genes, mediated by hemin/regulatory protein interactions, can be complemented by iron or dioxygen effects on regulatory proteins exemplified by dioxygen sensing of 2-oxoglutarate reductases (45, 46) and by anoxia effects on IRP proteins (46, 47). Ferritin-L gene regulation, particularly with hemin as a signal, is a rich source of information showing coordination of dual DNA (MARE/ARE) and mRNA (IRE) genetic mechanisms, possibly through hemin binding to specific DNA and mRNA repressors (Fig. 5), indicating a mechanism to integrate iron and oxygen metabolism and modeling DNA and mRNA regulation at other metabolic crossroads.