Oxidant-Induced Cell Death and Nrf2-Dependent Antioxidative Response Are Controlled by Fra-1/AP-1

Abstract

AP-1 (Jun/Fos) transcription factors play key roles in various biological processes, including cell death. Here we report a novel role for Fra-1 in oxidant-induced cell death controlled by modulating antioxidant gene expression. Fra-1-deficient (Fra-1^Δ/Δ) mouse embryonic fibroblasts (MEFs) and primary lung fibroblasts (PLFs) were remarkably resistant to H₂O₂- and diquat-induced cell death, compared to their wild-type (Fra-1^+/+) counterparts. Fra-1 deficiency ablated oxidant-induced mitochondrion-dependent apoptosis. Fra-1^Δ/Δ cells had elevated basal levels of antioxidant enzymes and intracellular glutathione (GSH), which were further stimulated by oxidants. Loss of Fra-1 led to an increased half-life of transcription factor Nrf2 and increased recruitment of this protein to the promoters of antioxidant genes and increased their expression. Depletion of intracellular GSH or RNA interference (RNAi)-mediated knockdown of Nqo1, Hmox1, and Nrf2 restored oxidant-induced cell death in Fra-1^Δ/Δ cells. Thus, Fra-1 appears to increase susceptibility to oxidants and promotes cell death by attenuating Nrf2-driven antioxidant responses.

INTRODUCTION

Oxidative stress-induced cell death is a major factor in the pathogenesis of many acute and chronic diseases (15). The effector pathways that bring about the death of cells in response to diverse exogenous stimuli are well established. Although most oxidant-induced cell death occurs via mitochondrion-dependent activation of caspases 9 and 3, the upstream regulators that control this process are not clearly understood. Oxidative stress-induced dysregulation of gene expression has been implicated in the development of diseases in the lungs and many other organs (2, 7, 16). For example, oxidative stress activates a number of transcription factors and gene products that confer either cell protection or death (16). Among them are the homo- and heterodimeric complexes of the Jun/Fos family of proteins, generically known as AP-1 (13, 45). AP-1 binds to the TPA (12-O-tetradecanoylphorbol-13-acetate)response element (TRE; also known as the AP-1 site) of target gene promoters and stimulates their expression in response to various pro-oxidants and toxicants. These gene products mediate (mitigate and promote) oxidative stress (30, 46).

Studies using genetic models have demonstrated both cooperative and antagonistic effects of the Jun family of proteins in modulating cell death in response to a variety of proapoptotic stimuli. For example, c-Jun^−/− mouse embryonic fibroblasts (MEFs) and liver cells show increased levels of oxidative stress and apoptosis (12, 59). In contrast, human myeloid leukemia cells or neuronal cells expressing c-Jun mutant were less sensitive to stress-induced apoptosis (1, 14, 20, 56). Jun^−/− MEFs exhibited increased resistance to cell death induced by alkylating agents (34). Likewise, c-Fos also participates in both pro- and antiapoptotic activities. For example, c-Fos^−/− MEFs undergo apoptosis when cultured in vitro and also display greater susceptibility to UV-induced cell death (44). However, retinal cells lacking c-Fos showed increased resistance to light-induced apoptosis (18). A distinct regulation of cell death by AP-1 family members in response to proapoptotic stimuli has generally been attributed to the nature of activation of Jun and Fos family members, the duration of the subsequent TRE-mediated induction of genes that control cellular stress and apoptosis, and the cellular context (30, 46).

In contrast to the early activation of Jun and c-Fos (which peaks at 15 to 30 min), the stimulation of Fra-1, a dimeric partner of Jun family members, by various mitogenic and stress stimuli occurs at a notably later time (peaking at 90 to 180 min) (8). This difference in timing has been suggested to play a critical role in modulating chronic transcriptional responses mediated by AP-1 (29, 41). However, the exact roles played by Fra-1 in regulating oxidative stress-induced cell death are not clearly understood. In this study, we investigated the role of Fra-1 in mediating oxidant-induced stress responses by the use of embryonic and primary lung fibroblasts from wild-type (Fra-1^+/+) and knockout (Fra-1^Δ/Δ) mice. We found that cells lacking Fra-1 were resistant to diquat- and H₂O₂-induced cell death, suggesting that Fra-1 is critical to promoting oxidant-induced cellular responses.

MATERIALS AND METHODS

Cell culture.

Primary mouse embryonic fibroblasts (MEFs) derived from Fra-1^+/+ (Fra-1^f/f) and Fra-1^Δ/Δ mice were obtained from Erwin Wagner at passage 2. These cells were subjected to passage until passage 10 was reached. The cells from passage 10 were used in all experiments. Three separate batches of cells were prepared and used to determine reproducibility of results. MEFs were grown in Iscove's modified Dulbecco's medium supplemented with 10% fetal calf serum, streptomycin, and penicillin. Primary lung fibroblasts (PLFs) were isolated from lungs of adult wild-type (Fra-1^+/+) and Fra-1^Δ/Δ mice. Cells were isolated from 3 independent mice of each genotype, and cells from passage 5 were used in all experiments. Fra-1^Δ/Δ mice were generated by crossing Fra-1 floxed mice with Meox2-Cre mice (Jackson Laboratory) as detailed before (11). To isolate PLFs, lung tissues were incubated with dispase solution (Roche Applied Science, Indianapolis, IN) (0.8 U/ml) and incubated at 37°C for 45 min. Lung tissue was gently teased and minced in a 100-mm-diameter culture dish containing 15 ml of HEPES-buffered Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (vol/vol), streptomycin (100 μg/ml), penicillin (100 units/ml), and 2 mM l-glutamine and DNase I (100 U/ml). The cell suspension was passed through 100-μm- and 40-μm-pore-size cell strainers, and fibroblasts were allowed to adhere to tissue culture plates for 10 min. The supernatant containing nonadherent epithelial and other cells was removed, and the fibroblasts were allowed to grow for 7 to 10 days at 37°C in a tissue culture incubator. Fibroblasts that adhered to the plate were trypsinized and subjected to passage for further experimentation. All experimental animal protocols were performed in accordance with guidelines approved by the animal care and use committee at The Johns Hopkins University.

Gene expression analysis.

For analysis of RNA or protein expression levels, cells were treated with H₂O₂ (200 μM) or diquat (100 μM) for the indicated time periods. Total RNA was isolated using TRIzol reagent (Gibco-BRL) and reverse transcribed using a Superscript III cDNA synthesis kit (Invitrogen Corp., San Diego, CA). Target gene expression was assessed by quantitative RT-PCR (qRT-PCR) using TaqMan gene expression assays (Applied Biosystems, Foster City, CA). For immunoblot analyses, total protein was extracted using a lysis buffer consisting of 20 mM Tris (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM Na₃VO₄, 5 mM β-glycerophosphate, and leupeptin (1 μg/ml). Comparable amounts of total protein from each sample were separated using 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and membranes were probed with the indicated antibodies specific for Hmox1, Nqo1, and Gclc (Santa Cruz Biotechnology), caspase 3, caspase 9, and cleaved poly(ADP-ribose) polymerase (PARP; Cell Signaling Technology), and β-actin (Sigma, St. Louis, MO). The blots were developed using an ECL kit (Pierce, Rockford, IL).

Cell viability and cell death assays.

Cells were treated with diquat (200 μM) and H₂O₂ (500 μM) for 8 h. Cell viability and cell death were quantified using CellTiter-Glo and CytoTox-Glo kits, respectively (Promega, Madison, WI). Cell viability and cell death were calculated as percentages of decrease and increase, respectively, compared to vehicle-treated control cell results.

Mitochondrial outer membrane potential (MOMP).

To monitor apoptosis-associated mitochondrial membrane depolarization caused by diquat and H₂O₂, cells were probed with JC-1 dye, also known as Mitotracker, according to the manufacturer's instructions (Cayman Chemical Company, Ann Arbor, MI). Cells were incubated with JC-1 for 15 min at 37°C, washed, and observed under a fluorescence microscope. In healthy cells, JC-1 accumulates in the mitochondria and forms aggregates that emit red fluorescence. In apoptotic cells, due to a loss of mitochondrial membrane potential, JC-1 does not aggregate but exists mainly as a monomer that emits green fluorescence.

Quantification of GSH.

Cells were treated with diquat (100 μM) or H₂O₂ (200 μM) for 6 h, and glutathione (GSH) levels were measured using a total glutathione detection kit (Enzo Life Sciences, Ann Arbor, MI) per the kit protocol. Briefly, following oxidant treatment, cells were counted and suspended in 5% metaphosphoric acid (4 × 10⁶ cells/ml), sonicated, and centrifuged, and the supernatant was used for assaying GSH. A freshly prepared reaction mixture was added to the supernatant, and the absorbance at 405 nm was measured immediately and at 1-min intervals over a 10-min period. The glutathione levels were determined in triplicate and calculated using GSH standards.

siRNA transfection.

siGENOME Smart pool small interfering RNAs (siRNAs) specific for Hmox1 (M-040543-01), Nqo1 (M-042699-01-0005), Nrf2 (M-040766-01), and Fra-1 (M-040704-01) were obtained from Dharmacon (Lafayette, CO). A siGENOME nontargeting siRNA (pool 2; D-001206-14-05) was used as a control. Cells were transfected with 50 nM siRNA oligonucleotides by the use of DharmaFECT1 (Dharmacon, Lafayette, CO). All siRNAs were verified for target sequence specificity before employment in the experiments. At 24 h posttransfection, cells were treated with H₂O₂ or diquat for 8 h, following which cell viability was estimated. To examine decreases in protein expression following siRNA transfection, cells were transfected for 24 h, following which they were treated with H₂O₂ or diquat for 6 h prior to lysis and immunoblot analysis.

ChIP assays.

Chromatin immunoprecipitation (ChIP) assays were performed using an EZ-ChIP assay kit (Millipore). Briefly, MEFs were treated with diquat and cross-linked with formaldehyde, and chromatin fragmentation was carried out as detailed in the kit protocol. Diluted soluble chromatin solution was incubated with rabbit anti-Nrf2, anti-c-jun, or anti-Jun-D antibodies (Santa Cruz Biotechnology) for 18 h at 4°C with rotation. Nonimmune rabbit IgG was used to determine the binding specificity. Following incubation with protein A-agarose, the bound products were washed and DNA was eluted. DNA was subjected to qRT-PCR with primers encompassing the functional antioxidant response elements (AREs) of the Hmox1 (F, 5′-GCTCATTTCCTCAGCTGCTT-3′, and R, 5′-TTCCGGAACCTTTTACCAAC-3′, for the −3872 ARE, and F, 5′-GGGCAGTCTTAAGCAATCCA-3′, and R, 5′-AAGGGTTCAGTCTGGAGCAA-3′, for the −9801 ARE) and Nqo1 (F, 5′-GCAGTTTCTAAGAGCAGAACG-3′, and R, 5′-GTAGATTAGTCCTCACTCAGCCG-3′) promoters. The primers located downstream of transcriptional start sites of Hmox1 (F, 5′-GGCAGGTATGGACCTTCAAA-3′ and R, 5′-AAAGGAGTCAGGGAGGGAGA-3′]) and Nqo1 (F, 5′-GGGCGCACTATTGTCATCTT-3′, and R, 5′-AGGAGATGGAAGGCAGGAAG-3′) were used to determine the specificity of Nrf2 binding in ChIP assays.

Determination of protein turnover rate.

MEFs were treated with cycloheximide (25 μg/ml) for the indicated time durations, and total cell lysates were subjected to Western blot analysis with the indicated antibodies. Data were normalized to the internal control beta-actin protein. Band intensities from multiple blots were densitometrically quantified and plotted.

Immunocytochemistry.

Fra1^+/+ and Fra-1^Δ/Δ MEFs were grown on coverslips and treated with 200 μM H₂O₂ for 30 min. The MEFs were washed with phosphate-buffered saline (PBS) and fixed in cold methanol for 10 min at −20°C. The cells were permeabilized with 0.1% Triton X-100, blocked with 5% bovine serum albumin (BSA) and 1% normal donkey serum, and incubated with anti-Nrf2 antibodies (SC-13032; Santa Cruz Biotechnology, Santa Cruz, CA) for 16 h. The cells were washed and further incubated with peroxidase-conjugated anti-rabbit IgG for 1 h. Immunostains were developed using a Liquid Dab-Plus substrate kit per the instructions of the manufacturer (Invitrogen, Carlsbad, CA) and counterstained with hematoxylin. Cells were mounted using Cytoseal. An Aperio (Vista, CA) Scanscope CS slide scanner system was used to capture whole-slide digital images using a 20× objective and the built-in IHC parameter set. Using ImageScope release 11.1.2.752, a region of interest was drawn within each coverslip. The Aperio Nuclear version 9 and Colocalization version 9 algorithms were utilized in the analysis. These algorithms provide the percentages and numbers of negative, weak, moderate, and strongly staining nuclei in the image. For this analysis, we looked at the percentage of positive nuclei in each sample and performed a chi-square analysis. The output of the colocalization algorithm gave the percentages of pixels that were nucleus negative, nucleus positive, and cytoplasm positive. Cytoplasm-negative pixels were not evaluated by the algorithm and display as white space.

Statistical analyses.

Data are expressed as means ± standard deviations (SD) (n = 3 to 5 for each treatment). The statistical significance of the differences between samples was determined using Student's t test. Differences were considered significant for samples with P values of less than 0.05.

RESULTS

Fra-1 deficiency protects from oxidant-induced cell death.

To determine the role of Fra-1 in mediating oxidant-induced cell death, we treated isogenic Fra-1^+/+ and Fra-1^Δ/Δ mouse embryonic fibroblasts (MEFs) with either 500 μM H₂O₂ or 200 μM diquat for 8 h, following which cell survival and death were assessed. Fra-1^Δ/Δ cells were found to be resistant to H₂O₂- and diquat-induced death compared to isogenic Fra-1^+/+ cells (Fig. 1A, B, and C). Microscopic evaluation (Fig. 1A) revealed either a total lack or an extremely reduced level of death in Fra-1^Δ/Δ compared to Fra-1^+/+ cells in response to H₂O₂ and diquat treatment. Quantification of cell survival (Fig. 1B) and death (Fig. 1C) showed a decrease (∼50%) in viability of Fra-1^+/+ cells after treatment with both oxidants. However, Fra-1^Δ/Δ cells were strikingly resistant to both H₂O₂- and diquat-induced cell death. The percentage of survival of Fra-1^Δ/Δ cells was significantly higher than that of the Fra-1^+/+ MEFs after treatment with either oxidant. There was no significant difference in cell death data between Fra-1^+/+ and Fra-1^Δ/Δ genotypes in response to other death-promoting stimuli such as etoposide (Fig. 1F) or puromycin (Fig. 1G). Thus, the loss of Fra-1 appears to specifically decrease susceptibility to oxidant-induced cell death. We next determined the effects of H₂O₂ and diquat on Fra-1 expression in Fra-1^+/+ and Fra-1^Δ/Δ cells. Expression of mRNA (Fig. 1D) and protein (Fig. 1E) was strongly induced in the Fra-1^+/+ cells following treatment with either H₂O₂ or diquat. As expected, Fra-1 mRNA expression was not detected in the Fra-1^Δ/Δ MEFs (Fig. 1D).

Fig 1.

Effects of oxidants on cell death in Fra-1^+/+ and Fra-1^Δ/Δ MEFs. (A) Photomicrographs of cells after treatment with H₂O₂ or diquat for 8 h. (B and C) Quantitation of cell viability (B) and cell death (C) after treatment with H₂O₂ or diquat. (D and E) H₂O₂- and diquat-induced upregulation of Fra-1 mRNA (D) and protein (E). For mRNA expression analysis, cells were treated, RNA was extracted, cDNA was prepared, and Fra-1 expression was quantitated by qRT-PCR using β-actin as a reference. Data are expressed as fold change relative to the control (Con) group. For protein analysis, cells were treated and total protein lysates prepared and analyzed by immunoblotting using anti-Fra-1 antibodies. β-Actin was used as a loading control. ND, not detectable. (F and G) Quantitation of cell viability following treatment with etoposide (50 μM) (F) and puromycin (4 μg/ml) (G). Data shown represent averages of the results of three independent experiments. *, P < 0.05, control versus treatment group. †, P < 0.05, Fra-1^+/+ versus Fra-1^Δ/Δ.

Fra-1 deficiency blocks the oxidant-induced activation of cell death effectors.

Oxidants are known to cause cell death through the activation of proapoptotic caspases, such as caspase 3 and caspase 9, and cleavage of their downstream substrate, poly (ADP-ribose) polymerase (PARP). We therefore compared levels of activation of these death effectors in Fra-1^+/+ and Fra-1^Δ/Δ cells. As expected, H₂O₂ (Fig. 2A) and diquat (Fig. 2B) induced the proteolytic activation of caspase 3, caspase 9, and PARP in Fra-1^+/+ cells but not in Fra-1^Δ/Δ cells. Thus, a defective activation of caspases and its downstream substrates is consistent with the lack of oxidant-induced cell death seen in Fra-1^Δ/Δ cells.

Fig 2.

H₂O₂- and diquat-induced activation of proapoptotic proteins in Fra-1^+/+ and Fra-1^Δ/Δ MEFs. Data represent expression of proapoptotic proteins in cells treated with (A) H₂O₂ (500 μM) or (B) diquat (200 μM). Cells were treated for the indicated time periods, following which total protein lysates were prepared and analyzed by immunoblotting using antibodies specific for caspase 3, caspase 9, and cleaved PARP. β-Actin was used as a loading control. Blots shown are representative of the results of three independent experiments.

Fra-1 deficiency prevents oxidant-induced mitochondrial membrane depolarization.

Reactive oxygen species (ROS) generated following exposure to oxidants are known to initiate the apoptotic pathway by directly disrupting the mitochondrial membrane potential (Δψm). Given the failure of Fra-1^Δ/Δ MEFs to activate caspases 3 and 9 following oxidant treatment, we asked whether such a failure to respond had occurred because the mitochondria had not been damaged. Mitochondrial dysfunction (i.e., loss of Δψm) causes the release of cytochrome c, which activates caspase 9 via the formation of apoptosomes and then caspase 3. Therefore, we investigated changes in Δψm during oxidative stress. Fra-1^+/+ and Fra-1^Δ/Δ cells were treated with oxidants, and the changes in Δψm were evaluated using JC-1 dye (Mitotracker). As expected, treatment of the Fra-1^+/+ cells with H₂O₂ or diquat led to a decrease in Δψm, as indicated by an increase in the monomeric form of JC-1, which emits green fluorescence, in contrast to the red fluorescence observed with intact mitochondria of control cells (Fig. 3, left panels). In Fra-1^Δ/Δ cells, treatment with H₂O₂ or diquat did not cause any decrease in Δψm (Fig. 3, right panels). Thus, Fra-1 is required for the oxidant-induced apoptosis that occurs as a result of a decrease in Δψm.

Fig 3.

Alterations in mitochondrial membrane potential (Δψm) in Fra-1^+/+ and Fra-1^Δ/Δ MEFs following oxidant exposure. Fluorescence microscopy imaging of mitochondria in Fra-1^+/+ and Fra-1^Δ/Δ MEFs stained with JC-1 following treatment with vehicle control, H₂O₂ (500 μM), or diquat (200 μM) is shown. In channels 1 (CH1), red indicates hyperpolarized, J aggregates; in channels 2 (CH2), green indicates the monomer form of JC-1. The final column in each panel shows merged fluorescence images. Δψm was disrupted in the Fra-1^+/+ MEFs after oxidant treatment, as indicated by the loss of red and gain in green fluorescence. Fra-1^Δ/Δ MEFs do not exhibit a major change in Δψm, as indicated by the persistence of red fluorescent staining in the mitochondria even after oxidant treatment. Data shown are representative of the results of three independent experiments.

Fra-1^Δ/Δ cells express elevated levels of antioxidant enzymes and GSH.

Cells employ a number of mechanisms to help them cope with oxidative stress, one of which involves the induction of a host of antioxidant enzymes and proteins (32). To identify the basis for the failure of Fra-1^Δ/Δ cells to undergo oxidant-induced cell death, we next analyzed the expression of several antioxidant-responsive genes. The levels of Nqo1, Hmox1, Gclc, Gclm, Gpx2, and Scl7a11 expression were compared between Fra-1^+/+ and Fra-1^Δ/Δ MEFs following exposure to H₂O₂ and diquat. We chose these genes as they are upregulated by cellular stress, mainly by the Nrf2 transcription factor, and are known to play key roles in the cellular detoxification process (reviewed in reference 32). Nqo1 reduces quinones to hydroquinones, Hmox1 cleaves heme to biliverdin, Gclc and Gclm are required for GSH biosynthesis, Gpx2 scavenges hydrogen peroxide, and Slc7a11 transports cysteine and glutamate, which are precursors of GSH. A significant rise in the steady-state levels of Nqo1 (7-fold), Hmox1 (4-fold), Gclm (1.7-fold), and Slc7a11 (3-fold) mRNAs was observed in Fra-1^Δ/Δ MEFs (Fig. 4A, filled bars) compared with the wild-type cells (open bars). Further, H₂O₂ (bars 3 and 4) or diquat (bars 5 and 6) markedly induced Nqo1, Hmox1, and Gclc gene transcripts in Fra-1^Δ/Δ cells (filled bars, Fig. 4A) but only modestly in Fra-1^+/+ cells (open bars, Fig. 4A). While the levels of induction of Gclm and Slc7a11 of the two cell types by H₂O₂ were comparable, their induction by diquat was markedly higher in Fra-1^Δ/Δ cells than Fra-1^+/+ cells. Although H₂O₂- and diquat-induced Gpx2 expression was higher in Fra-1^Δ/Δ cells than Fra-1^+/+ cells, this difference was not statistically significant. We performed immunoblot analyses to confirm the increased expression levels of Nqo1, Hmox1, and Gclc in Fra-1^Δ/Δ cells. These studies revealed increased levels of Hmox1 (4-fold), Nqo1 (5-fold), and Gclc (4.5-fold) expression in Fra-1^Δ/Δ MEFs compared to Fra-1^+/+ cells under basal conditions (Fig. 4B and C). H₂O₂ and diquat stimulated Hmox1 protein levels in both cell types. However, the expression level of Hmox1 was higher in Fra-1^Δ/Δ MEFs than in the wild-type counterparts (Fig. 4C). There was a significant increase in Nqo1 and Gclc expression in wild-type cells at 3 h post-H₂O₂ treatment, while their induction by diquat was observed only at 6 h posttreatment. In Fra-1^Δ/Δ cells, the expression levels of Nqo1 and Gclc, which were markedly higher under basal conditions, were not further increased following H₂O₂ or diquat treatment (Fig. 4B and C). It is possible that a lag between Gclc mRNA induction and protein synthesis could be attributed to this condition in our experimental settings. Alternatively, as Fra-1^Δ/Δ cells express increased levels of Gclc, a feedback loop may limit a further rise in Gclc protein levels in these cells.

Fig 4.

The effects of Fra-1 deficiency on H₂O₂- and diquat-induced antioxidant enzyme expression and GSH levels. (A) qRT-PCR analysis of Nqo1, Hmox1, Gclc, Gpx2, Gclm, and Scl7a11 expression. Cells were treated with H₂O₂ or diquat for 6 h, and total RNA was extracted and analyzed. Data are expressed as fold changes relative to vehicle-treated wild-type cells. (B) Immunoblot analysis of Hmox1, Nqo1, and Gclc expression following treatment with H₂O₂ or diquat. β-Actin was used as a loading control. (C) Densitometric analysis of Hmox1, Nqo1, and Gclc expression normalized to β-actin. The ratios of the absolute densitometric values determined for the respective genes versus β-actin have been plotted. (D) Quantitation of GSH levels in MEFs before and after treatment with H₂O₂ or diquat. Data shown represent averages of the results of two independent experiments performed in triplicate. *, P < 0.05, control versus treatment group. †, P < 0.05, Fra-1^+/+ versus Fra-1^Δ/Δ.

In addition to the aforementioned enzymes, GSH is essential for the cellular defense against oxidative stress. Reduced GSH levels maintain the intracellular redox state by scavenging ROS and protect against the induction of apoptosis. We therefore analyzed the levels of GSH in the Fra-1^+/+ and Fra-1^Δ/Δ MEFs following treatment with H₂O₂ or diquat (Fig. 4D). GSH levels decreased in Fra-1^+/+ cells after treatment with diquat, although no significant difference was observed after treatment with H₂O₂. In contrast, the Fra-1^Δ/Δ MEFs expressed significantly higher basal levels of GSH than the Fra-1^+/+ cells. These levels remained unchanged after treatment with H₂O₂ but increased following treatment with diquat. Thus, Fra-1 appears to maintain the cellular redox state and the expression of antioxidative genes in these cells.

Depletion of GSH or inhibition of antioxidant enzymes sensitizes Fra-1^Δ/Δ cells to oxidant-induced cell death.

To determine whether the elevated levels of GSH and antioxidant genes in the Fra-1^Δ/Δ MEFs contribute to their resistance to oxidant-induced cell death, we depleted GSH levels by treating the cells with buthionine sulfoximine (BSO), a specific inhibitor of de novo GSH synthesis, and then assessed the ability of H₂O₂ and diquat to induce cell death (Fig. 5A). As expected, BSO treatment caused an increase in oxidant-induced cell death in the Fra-1^+/+ cells. Importantly, BSO also sensitized the Fra-1^Δ/Δ MEFs to H₂O₂- and diquat-induced cell death. In the presence of BSO, H₂O₂ and diquat caused a significant decline in the number of viable Fra-1^Δ/Δ MEFs compared to the controls (Fig. 5A). These results were consistent with our earlier finding that Fra-1^Δ/Δ cells showed very little change in mitochondrial membrane potential in response to oxidants (Fig. 3) and contained elevated levels of GSH (Fig. 4D). Since GSH prevents oxidant-mediated apoptosis by maintaining the Δψm levels (1), we asked next whether depletion of GSH (by BSO pretreatment) in Fra-1^Δ/Δ cells would allow the mitochondrial damage (Δψm) to occur in response to H₂O₂ or diquat. As shown in Fig. 5B, pretreatment with BSO led to a significant decrease in Δψm in the Fra-1^Δ/Δ cells after exposure to H₂O₂ or diquat (right panel), in a manner similar to that seen in wild-type Fra-1^+/+ cells (left panel). Thus, the elevated GSH levels appear to protect Fra-1^Δ/Δ cells from oxidant-induced cell death.

Fig 5.

Effect of inhibition of antioxidant enzyme activity or depletion of GSH on oxidant-induced cell death in Fra-1^Δ/Δ MEFs. (A) Effect of GSH depletion on cell viability following treatment with H₂O₂ or diquat. Fra-1^+/+ and Fra-1^Δ/Δ cells were treated with the respective oxidants in the presence or absence of BSO (100 μM) for 8 h, following which cell viability was assessed. (B) Effect of GSH depletion on oxidant-induced change in Δψm. Fluorescence microscopy imaging of mitochondria in Fra-1^+/+ and Fra-1^Δ/Δ MEFs stained with JC-1 following treatment with H₂O₂ or diquat in the presence of BSO (100 μM) is shown. In channels 1 (CH1), red indicates hyperpolarized, J aggregates; in channels 2 (CH2), green indicates the monomer form of JC-1. The final column in each panel shows merged fluorescence images. Mitochondrial membrane potential was disrupted in the Fra-1^Δ/Δ MEFs following treatment with oxidants in the presence of BSO, as indicated by the loss of red and gain in green fluorescence. (C) Immunoblot analysis showing the expression of Hmox1 and Nqo1 in cells transfected with either a control nontargeting siRNA pool (C) or Hmox1 (H1)- or Nqo1 (N1)-specific siRNA following treatment with vehicle control (Con), H₂O₂, or diquat. (D) Effect of H₂O₂ or diquat on cell survival following siRNA-mediated knockdown of Nqo1 and Hmox1 expression. A nontargeting siRNA pool was used as a control. Data shown represent averages of the results of two independent experiments performed in triplicate. *, P < 0.05, control versus treatment group.

In addition to GSH, antioxidant enzymes Nqo1 and Hmox1 are also critical for protecting cells against oxidative stress (52, 57). Therefore, we knocked down the expression of Nqo1 and Hmox1 by the use of RNA interference (RNAi) (Fig. 5C) and assessed the impact on H₂O₂- and diquat-induced cell death (Fig. 5D). Knocking down Nqo1 or Hmox1 expression further increased the susceptibility of Fra-1^+/+ cells to H₂O₂ or diquat treatment (Fig. 5D, left panel). Conversely, knockdown of Nqo1 or Hmox1 expression restored the susceptibility of Fra-1^Δ/Δ cells to either H₂O₂ or diquat treatment (Fig. 5D, right panel).

Effect of oxidant exposure on primary lung fibroblasts.

In order to demonstrate the relevance of our results obtained with MEFs in a physiologically relevant model, we isolated primary fibroblasts (PLFs) from the lungs of Fra-1^+/+ and Fra-1^Δ/Δ mice, exposed them to either H₂O₂ or diquat for 8 h, and determined their viability. Consistent with the results obtained with MEFs, PLFs lacking Fra-1 showed significantly lower susceptibility to oxidant-induced cell death (Fig. 6A). Fra-1^Δ/Δ PLFs also contained elevated levels of GSH (Fig. 6B), and BSO treatment sensitized them to H₂O₂- and diquat-induced cell death (Fig. 6C). Analysis of antioxidant gene expression also revealed constitutively elevated expression of the antioxidant genes Nqo1 and Hmox1 but not Gclc and Gpx2 in Fra-1^Δ/Δ PLFs, which was further increased in the presence of H₂O₂ or diquat (Fig. 6D). Thus, the effects of oxidant exposure in PLFs were similar to those observed in MEFs.

Fig 6.

Effects of H₂O₂ and diquat on cell death and antioxidative gene expression in Fra-1^+/+ and Fra-1^Δ/Δ primary mouse lung fibroblasts (PLFs). (A) Quantitation of cell survival after treatment with H₂O₂ or diquat for 8 h. (B) GSH levels before and after treatment with H₂O₂ or diquat. (C) Effects of H₂O₂ and diquat on cell survival in the presence of BSO (100 μM). Cells were treated with the respective oxidants in the presence or absence of BSO for 8 h, and cell viability was estimated. (D) qRT-PCR analysis of Nqo1, Hmox1, Gclc, and Gpx2 expression induced by H₂O₂ and diquat. Data are expressed as fold changes relative to the wild-type vehicle-treated control results. Data shown represent averages of the results of two independent experiments performed in triplicate. *, P < 0.05, control versus treatment group. †, P < 0.05, Fra-1^+/+ versus Fra-1^Δ/Δ.

Reconstitution of Fra-1 in Fra-1^Δ/Δ MEFs restored their sensitivity to oxidants, whereas Fra-1 knockdown in Fra-1^+/+ MEFs had an opposite effect.

Based on the observations presented above, we next asked whether reconstitution of Fra-1^Δ/Δ cells with Fra-1 could restore their sensitivity to oxidant-induced cell death. Fra-1^Δ/Δ MEFs were transfected either with a plasmid vector expressing Fra-1 under the control of the cytomegalovirus (CMV) promoter (Fra-1^{Δ/Δ + Fra-1}) or with an empty CMV vector (Fra-1^{Δ/Δ + CMV}). Stably transfected clones were generated, pooled, and used for experiments. Fra-1^Δ/Δ cells transfected with an empty CMV vector were used as an appropriate control to rule out potential problems associated with stable transfection or the expression vector. We have verified the restoration of Fra-1 in Fra-1^Δ/Δ cells by RT-PCR analysis (Fig. 7A) and also confirmed that the reexpressed Fra-1 was induced following oxidant treatment (Fig. 7B). The transfected cells were then exposed to H₂O₂ or diquat, following which cell viability was determined. Reexpression of Fra-1 restored the sensitivity of these cells to oxidant-induced cell death (Fig. 7C). Analysis of antioxidant gene expression revealed that these cells also had reduced basal levels of Hmox1 (Fig. 7D). Furthermore, restoration of Fra-1 significantly suppressed the increased expression levels of Hmox1 following treatment with H₂O₂, while it had a modest effect on diquat-induced Hmox1 expression. Ectopic expression of Fra-1 in Fra-1^+/+ cells neither caused cell death under basal conditions nor increased their sensitivity to oxidants (data not shown). However, knockdown of Fra-1 expression by the use of an RNAi approach caused increased expression levels of Hmox1 and Nqo1 in Fra-1^+/+ cells under basal conditions and in response to oxidant treatment compared to the results seen with control siRNA-transfected cells (Fig. 7E). Immunoblot analysis also confirmed the decreased levels of Fra-1 expression and upregulation of Hmox1 and Nqo1 expression in cells transfected with Fra-1 siRNA (Fig. 7F). The knockdown of Fra-1 in wild-type cells also increased the resistance of these cells to H₂O₂- or diquat-induced cell death compared to their control siRNA-transfected Fra-1^+/+ counterparts (Fig. 7G).

Fig 7.

The effects of Fra-1 reexpression in Fra-1^Δ/Δ MEFs and Fra-1 knockdown in wild-type cells on oxidant-induced cell death and antioxidant expression. (A) qRT-PCR analysis of Fra-1 mRNA expression in Fra-1^Δ/Δ cells transfected with a Fra-1-expressing vector (Fra-1^{Δ/Δ + Fra-1}) or an empty vector (Fra-1^{Δ/Δ + CMV}). Expression of Fra-1 is calculated relative to Fra-1 expression in Fra-1^+/+ cells, which was assigned a value of 1. (B) qRT-PCR analysis of Fra-1 expression in Fra-1^{Δ/Δ + CMV} and Fra-1^{Δ/Δ + Fra-1} cells following treatment with H₂O₂ or diquat. (C) Quantitation of cell survival in Fra-1^{Δ/Δ + CMV} and Fra-1^{Δ/Δ + Fra-1} cells after treatment with H₂O₂ or diquat. (D) Quantitation of Hmox-1 mRNA in Fra-1^{Δ/Δ + Fra-1} cells following exposure to H₂O₂ or diquat. (E) qRT-PCR analysis of expression of Fra-1 and antioxidant genes following transfection of Fra-1^+/+ cells with either control nontargeting siRNA or Fra-1 siRNA. (F) Immunoblot analysis of expression of Fra-1, Nqo1, and Hmox1 in Fra-1^+/+ cells transfected with either a control nontargeting siRNA (C) or Fra-1 siRNA (F1). (G) Quantitation of cell survival in Fra-1^+/+ cells transfected with control or Fra-1 siRNA. Data shown represent averages of the results of two independent experiments performed in triplicate. *, P < 0.05, control versus treatment group. †, P < 0.05, Fra-1^+/+ versus Fra-1^Δ/Δ.

Fra-1 deficiency increases the binding of Nrf2 to the promoters of antioxidant enzymes following oxidant treatment.

We had earlier observed that RNAi-mediated inhibition of Nqo1 and Hmox1 expression drastically increased the sensitivity of the Fra-1^Δ/Δ cells to diquat-induced cell death (Fig. 5D). It is well established that the Nrf2 transcription factor plays a crucial role in regulating the expression of antioxidant genes, including Nqo1 and Hmox1 (reviewed in reference 32). Thus, we hypothesized that the elevated Hmox1 and Nqo1 expression in Fra-1^Δ/Δ cells (Fig. 4A and B) could have been due to an increased recruitment of Nrf2 to their respective enhancers. We therefore examined the binding of Nrf2 to the antioxidant response elements (AREs) within the Hmox1 and Nqo1 promoters after treatment with diquat by the use of chromatin immunoprecipitation (ChIP) assays. These assays revealed that binding of Nrf2 to the Hmox1 and Nqo1 promoters was significantly augmented in Fra-1^Δ/Δ cells following oxidant treatment (Fig. 8A and B). Moreover, this binding was persistent in Fra-1^Δ/Δ cells even at 6 h post-diquat treatment, whereas it returned to baseline in the Fra-1^+/+ cells. PCR analysis of the same ChIP samples with primers located downstream of Hmox1 and Nqo1 promoters barely yielded any products in both cell types, suggesting a lack of Nrf2 binding to these nonspecific regions (Fig. 8C). We also examined the levels of Nrf2 to determine whether its expression was elevated in Fra-1^Δ/Δ cells. For this purpose, cells were treated with diquat or H₂O₂ for 6 h, after which Nrf2 mRNA levels were determined using qRT-PCR (Fig. 8D). There was no significant difference between the Fra-1^+/+ cells and Fra-1^Δ/Δ cells in terms of Nrf2 mRNA expression (Fig. 8D). However, total cell lysates prepared from H₂O₂- or diquat-treated cells showed elevated Nrf2 expression in Fra-1^Δ/Δ cells at 3 and 6 h posttreatment (Fig. 8E, top panel). Similarly, nuclear extracts prepared from these cells also showed enhanced Nrf2 nuclear accumulation in both Fra-1^+/+ and Fra-1^Δ/Δ cells at 3 and 6 h posttreatment (Fig. 8E, bottom panel). However, high nuclear levels of Nrf2 persisted in Fra-1^Δ/Δ cells. To confirm the Western blot data in situ, we have performed a quantitative analysis of immunocytochemical staining of Nrf2 by the use of an Aperio Scanscope whole-slide scanner system (see Materials and Methods for details) in Fra-1^+/+ and Fra-1^Δ/Δ MEFs treated with H₂O₂ or left untreated (Fig. 8F, G, and H). The results, in agreement with the Western blot data, revealed a significant increase in Nrf2 staining in Fra-1^Δ/Δ cells compared to Fra-1^+/+ cells under basal conditions. Although both nuclear and cytoplasmic levels of Nrf2 were increased in the two cell types following H₂O₂ treatment compared to untreated cells, Nrf2 signal intensity was significantly (2.25-fold; P < 0.0001) higher in Fra-1^Δ/Δ than Fra-1^+/+cells. The observed differences in fold change in the Nrf2 levels between Western blot and immunocytochemistry analysis could have been due to the different methods employed. It is possible that the Nrf2 epitopes are accessible to the antibody to a lower extent in the intact cells compared to the conditions of Western blot analysis, which denatures the protein. Nonetheless, results obtained from these complementary approaches showed a significantly elevated nuclear Nrf2 level in Fra-1^Δ/Δ cells under basal and treatment conditions.

Fig 8.

Analysis of Nrf2 expression and binding following oxidant treatment. (A and B) ChIP assays were performed to determine the binding of Nrf2 to the promoters of Hmox1 (A) and Nqo1 (B) encompassing the functional AREs after diquat treatment. MEFs were treated with diquat for the indicated time periods, and ChIP was performed. Input DNA, chromatin immunoprecipitants with immunoglobulin G (IgG), and anti-Nrf2 antibodies were used for PCR amplification of each promoter. DNA was analyzed by qRT-PCR. Values on each bar represent fold increases compared to the respective untreated controls obtained from two independent experiments. (C) ChIP analysis showing amplification of regions of DNA located downstream from the Hmox1 and Nqo1 promoters by the use of control nonspecific primers. (D) qRT-PCR analysis of Nrf2 expression following oxidant treatment. *, P < 0.05, control versus treatment group. (E) Immunoblot analysis of Nrf2 and Keap1 expression in total cell lysates (top) and Nrf2 expression in nuclear extracts (bottom) prepared from cells treated with H₂O₂ or diquat. (F) Immunocytochemical analysis of Nrf2 localization in Fra-1^+/+ and Fra-1^Δ/Δ MEFs. Pseudo-color “markup” images of Fra-1^+/+ and Fra-1^Δ/Δ MEFs immunostained with Nrf2 antibody following treatment with vehicle control or H₂O₂ (200 μM) for 30 min are shown. Images were developed using an Aperio Scanscope CS slide scanner system with the built-in IHC parameter set. The colors in the top panel indicate locations (green = cytoplasm and cyan = nucleus), and the colors in the bottom panel (nucleus) indicate the expression levels of Nrf2 in Fra-1^+/+ and Fra-1^Δ/Δ MEFs (blue = negative, yellow = low, orange = moderate, and red = high). (G and H) Percent Nrf2-positive nuclei (G) and total intensity of the Nrf2 immunostaining (H) quantified with Aperio Nuclear version 9 algorithms and analyzed by chi-square test. *, P < 0.0001, control versus treatment group. †, P < 0.0001, Fra-1^+/+ versus Fra-1^Δ/Δ MEFs.

Fra-1 deficiency alters the turnover of Nrf2.

We next determined if the increased levels of Nrf2 in the Fra-1^Δ/Δ cells were due to a change in its protein turnover rate (Fig. 9A). New protein synthesis was blocked with cycloheximide, and the rate of decline in preexisting Nrf2 levels was determined using Western blot analysis. These studies revealed a longer half-life of 79.3 min for Nrf2 in the Fra-1^Δ/Δ cells versus 30.8 min in Fra-1^+/+ cells. As we had observed increased levels of Gclc in the Fra-1^Δ/Δ cells under basal conditions (Fig. 4B), the turnover rate of Gclc was also examined. We observed that the Gclc half-life in Fra-1^Δ/Δ cells was comparable to that of Fra-1^+/+ cells. The rates of turnover of Gclc in Fra-1^Δ/Δ cells and Fra-1^+/+ cells were 5.1 h and 4.7 h, respectively, and the difference between the two genotypes was not statistically significant (Fig. 9B).

Fig 9.

Analysis of Nrf2 and Gclc turnover rates in Fra-1^+/+ and Fra-1^Δ/Δ MEFs. Data represent protein turnover rates for Nrf2 (A) and Gclc (B) in Fra-1^+/+ and Fra-1^Δ/Δ cells. Cells were treated with cycloheximide for the indicated time periods, following which cell lysates were prepared and immunoblot analysis was carried out for detection of Nrf2 and Gclc. β-Actin was used as a loading control. The bands were quantified by densitometry and normalized to β-actin levels. The arbitrary values were plotted in a log scale. The experiment was repeated once to obtain reproducible results.

To further confirm that Nrf2 mediates the protective responses of Fra-1^Δ/Δ cells, we transfected Fra-1^+/+ and Fra-1^Δ/Δ cells with an Nrf2-specific siRNA and then assessed the impact on oxidant-induced cell death (Fig. 10). As anticipated, the viability of the Fra-1^+/+ cells was significantly reduced in response to H₂O₂ or diquat treatment after RNAi-mediated knockdown of Nrf2 compared to that of cells transfected with a pool of nontargeting control siRNAs (Fig. 10A, left panel). Knockdown of Nrf2 expression in Fra-1^Δ/Δ cells also enhanced the susceptibility of these cells to both H₂O₂ and diquat (Fig. 10A, right panel). Diquat caused the death of ∼90% of the Fra-1^Δ/Δ cells after Nrf2 knockdown. To verify the knockdown of Nrf2 and the expression of its target genes, RNA was extracted from Fra-1^+/+ and Fra-1^Δ/Δ cells transfected with either an Nrf2-specific siRNA or nontargeting control siRNA followed by H₂O₂ or diquat treatment (Fig. 10B). Nrf2 knockdown markedly reduced the expression levels of Nrf2 and its target genes Hmox1 and Nqo1 in both Fra-1^+/+ and Fra-1^Δ/Δ cells compared to their control siRNA-transfected counterparts. We also confirmed the knockdown of Nrf2 and its target gene expression at the protein level. Immunoblot analysis was performed using anti-Nrf2 (Fig. 10C) and anti-Hmox1 (Fig. 10D) antibodies. As anticipated, the knockdown of Nrf2 expression markedly suppressed Hmox1 induction by H₂O₂ and diquat compared to their counterparts transfected with the control siRNA (Fig. 10D). As shown earlier (Fig. 2), we did not observe activation of proapoptotic (caspase 3 and 9) proteins in Fra-1-deficient cells following oxidant treatment. To determine if the increased levels of Nrf2 were responsible for this resistance to apoptosis, we examined the activation of these apoptotic proteins in Fra-1^Δ/Δ cells with Nrf2 knockdown following diquat treatment. As anticipated, immunoblot analysis did not reveal the activation of either caspase 3 or caspase 9 by diquat in Fra-1^Δ/Δ cells transfected with the nontargeting siRNA (Fig. 10E). However, we observed activation of these two caspases in Fra-1^Δ/Δ cells transfected with the Nrf2-siRNA (Fig. 10E) following diquat treatment.

Fig 10.

The effects of Nrf2 knockdown on oxidant-induced cell death in wild-type and Fra-1-deleted cells. (A) Fra-1^+/+ and Fra-1^Δ/Δ cells were transfected either with a nontargeting control siRNA pool or with Nrf2 siRNA and treated with H₂O₂ or diquat, and cell survival was measured. (B) qRT-PCR analysis of Nrf2, Nqo1, and Hmox1 expression following transfection with control or Nrf2 siRNA. Data shown represent averages of the results of two independent experiments performed in triplicate. *, P < 0.05, control versus treatment group. †, P < 0.05, Fra-1^+/+ versus Fra-1^Δ/Δ. (C) Immunoblot showing expression of Nrf2 in cells transfected with either control or Nrf2 siRNA. (D) Immunoblot showing Hmox1 expression in Fra-1^+/+ and Fra-1^Δ/Δ cells transfected with Nrf2 or control siRNA for 48 h followed by treatment with vehicle (Con), H₂O₂, or diquat for 6 h. β-Actin was used as a loading control. (E) Activation of caspase 3 and 9 following treatment of control and Nrf2 siRNA-transfected Fra-1^Δ/Δ cells with diquat.

Analysis of c-Jun and Jun-D binding in Fra-1-deficient cells.

Nrf2 has been shown to associate with Jun proteins to upregulate ARE-mediated induction of antioxidant genes (55). Therefore, it is possible that in the absence of Fra-1, an enhanced cooperative and synergistic interaction between Jun and Nrf2 proteins favors strong and persistent ARE-mediated gene transcription. Hence, we examined the binding of c-Jun and Jun-D to the AREs of the Hmox1 and Nqo1 promoters, following diquat treatment, using ChIP assays. We noted reduced binding of c-Jun and Jun-D to the Hmox1 and Nqo1 ARE sites in the Fra-1^Δ/Δ cells both before and following diquat treatment (Fig. 11). The binding of c-Jun at the Hmox1 ARE was maximum at 6 h in both Fra-1^+/+ and Fra-1^Δ/Δ cells following diquat treatment compared to the corresponding untreated cells. However, no significant increase in c-Jun binding was seen at the Nqo1 ARE (Fig. 11A). Similarly, Jun-D also showed maximum binding to the Hmox1 ARE at 6 h, with no increase in binding to the Nqo1 ARE, in the Fra-1^+/+ cells. However, a modest increase in Jun-D binding to the Nqo1 ARE was seen in the Fra-1^Δ/Δ cells at 1 h and 6 h following oxidant exposure compared to the vehicle-treated control Fra-1^Δ/Δ cells (Fig. 11B). This increase in binding in Fra-1^Δ/Δ cells was not greater than what was observed in Fra-1^+/+ cells. We observed no differences in the basal expression of c-Jun and Jun-D in the Fra-1^+/+ and Fra-1^Δ/Δ cells (data not shown).

Fig 11.

Analysis of Jun protein binding to the Hmox1 and Nqo1 promoters following oxidant treatment. (A and B) ChIP assays were performed to determine the binding of c-Jun (A) and Jun-D (B) to the promoters of Hmox1 (left panel) and Nqo1 (right panel) after diquat treatment. Chromatin preparations used were the same as used earlier to determine Nrf2 binding as described for Fig. 8. Values on each bar represent the fold increases compared to the respective untreated controls obtained from two independent experiments. (C) A model for the mechanisms regulating Fra-1-mediated oxidant-induced cell death. Nrf2 activation, which generally occurs early (within 30 min) in response to oxidative stress, leads to the induction of ARE-driven gene expression. Expression of Fra-1 (which peaks around 90 to 180 min), induced by oxidants, negatively controls Nrf2-dependent antioxidant gene expression, leading to cellular stress, mitochondrial membrane depolarization, and the activation of apoptotic cell machinery and ultimately resulting in cell death. Based on our current studies and those of others, we propose that Fra-1 negatively regulates Nrf2 functions by altering Nrf2 degradation and nuclear accumulation. Fra1 may limit either the Nrf2 DNA binding activity or Nrf2 association with its dimeric partners, such as small Mafs, Jun, and ATF-4 (M/J/A), which are known to positively regulate ARE-mediated transcription.

DISCUSSION

We demonstrate here a novel function for Fra-1 in promoting oxidative stress-induced cell death via an attenuation of Nrf2-dependent ARE-mediated antioxidative responses. We found that Fra-1 deficiency decreases the susceptibility to cell death induced by oxidants (Fig. 1). The requirement for Fra-1 in the induction of oxidant-driven cell death appears to be specific, because etoposide (a topoisomerase inhibitor) and puromycin (a translational inhibitor) equally induced cell death in Fra-1^+/+ and Fra-1^Δ/Δ cells. Fra-1 has been reported to bind to the promoters of genes associated with proliferation, such as CLND1 and CLNA, in malignant cell types. For example, in response to low levels of oxidative stress, Fra-1 can replace c-Fos in promoting CLND1 expression during the G₀-to-G₁ transition in malignant lung epithelial cells (3). In contrast, Fra-1 knockdown blocks proliferation and induces apoptosis in Ras-transformed thyroid cells, in part as the result of a loss of CLNA induction (4). Thus, Fra-1 appears to act as an effector of cellular functions and promotes either repair or cell death, depending on the cell type and the nature and severity of the stress involved.

The depolarization of mitochondria and activation of caspases leading to the cleavage of PARP are critical steps in promoting diquat- and H₂O₂-induced cell death (43). We found that the loss of Fra-1 prevented oxidant-induced changes in mitochondrial membrane potential and the consequent activation of an apoptotic cascade (Fig. 2 and 3). These observations suggest that changes in Fra-1-dependent gene expression alter mitochondrial membrane function. One of the major mechanisms involved in protecting cells from oxidative stress and subsequent cell death is the induction of expression of antioxidative genes (32, 33). The gene products involved include several detoxifying enzymes such as glutathione S-transferase (GST) isozymes Nqo1 and Hmox1 and the enzymes Gclc and Gclm, which are required for GSH synthesis. Indeed, our gene expression analyses revealed enhanced expression of several antioxidant enzymes and GSH in the absence of Fra-1 (Fig. 4). Consistent with this result, either a depletion of GSH levels or knockdown of Nqo1 and Hmox1 expression sensitized Fra-1^Δ/Δ cells to oxidant-induced cell death (Fig. 5). Thus, Fra-1 appears to attenuate the expression of products of antioxidative genes. Increased levels of Nqo1 and Hmox1 were also seen in several tissues, including the lung, harvested from Fra-1 knockout mice compared to those seen in wild-type mice (data not shown). We have recently shown that deletion of Fra-1 decreased susceptibility to lipopolysaccharide (LPS)-induced acute lung injury and inflammation and mortality in mice (54), while Fra-1 overexpression had an opposite effect (51). Previous studies using experimental (cell culture and animal) models have shown that antioxidant supplementation protects cells from oxidant-induced cell death (reviewed in reference 32). Based on these observations, we propose that dampening of antioxidant enzyme expression by Fra-1 is critical for promotion of cell death following exposure to oxidants. It is noteworthy that in PLFs, GSH levels in Fra-1^Δ/Δ cells were further increased following diquat but not H₂O₂ treatment (Fig. 6B), despite displaying greater resistance to H₂O₂-induced cell death than their wild-type counterparts (Fig. 6A). Gpx and catalase are two major enzymes that catalyze the breakdown of H₂O₂. It is possible that elevated levels of catalase or Gpx in Fra-1^Δ/Δ PLFs could be responsible for their improved cell survival under conditions of H₂O₂ treatment. In agreement with this notion, we found higher levels of Gpx2 expression in Fra-1^Δ/Δ PLFs than in wild-type cells (Fig. 6D).

The induction of Fra-1 expression generally peaks around 60 to 90 min in response to a wide variety of stressful stimuli (41), whereas nuclear accumulation of Nrf2, a transcription factor crucial for antioxidant gene induction, typically occurs within 30 to 60 min (26). Thus, it is possible that a regulatory feedback mechanism exists between these transcription factors to regulate ARE-mediated transcription. We speculate that, in response to low levels of oxidative stress, Nrf2 upregulates Fra-1 transcription through a feedback loop to decrease persistent Nrf2-dependent ARE-mediated transcription. On the other hand, antioxidants may activate Nrf2 to suppress Fra-1 expression, thereby promoting increased expression of Nrf2-dependent ARE-mediated antioxidant genes. Indeed, Yang et al. have reported an increased level of Fra-1 expression in response to an antioxidant, tert-butyl hydroxyquinone, in fibroblasts lacking Nrf2 compared to their isogenic wild-type counterparts, suggesting that Nrf2 suppresses Fra-1 expression in response to antioxidant treatment (60). We have also observed elevated levels of Fra-1 expression in Nrf2-deficient MEFs compared to Nrf2-suffcient (wild-type) cells following oxidant treatment (unpublished data). Interestingly, although reexpression of Fra-1 in Fra-1-deficient cells restores their sensitivity to oxidant-induced cell death, overexpression of Fra-1 alone in the wild-type (Fra-1-sufficient) cells did not cause cell death or enhanced oxidant-induced cell death (data not shown). Promoter sequence analysis indicated the presence of both the ARE and the AP-1 binding sites in the 5′-flanking regions of human and mouse Fra-1 and Nrf2. Recently, it was reported that knockdown of Fra-1 and c-Jun, but not Jun-D, decreases Nrf2 expression in MEFs bearing activated K-Ras mutations (9). Collectively, these observations suggest that Nrf2 and certain members of AP-1 family can each regulate the transcription of the others to modulate cellular responses during both acute and chronic oxidative stress and oncogenic transformation.

Our studies have revealed an important role for Fra-1 in the negative regulation of Nrf2-dependent expression of antioxidative genes. Nrf2 binds to the ARE in the promoters of several cellular detoxifying and antioxidative genes in response to a variety of oxidative stimuli (23, 42). Genetic disruption of Nrf2 enhances the susceptibility of cells to toxicant- and oxidant-induced cell death both in vivo and in vitro. For example, Nrf2 deficiency enhances susceptibility to pro-oxidant (e.g., hyperoxia and LPS)-induced acute lung injury and inflammation in vivo (6, 53), and MEFs lacking Nrf2 exhibited enhanced susceptibility to diquat-induced cell death in vitro (39). In the present study, loss of Fra-1, which confers protection against diquat treatment, augmented Nrf2 binding at the AREs of antioxidative (Hmox1 and Nqo1) gene promoters following oxidant exposure (Fig. 8). Consistent with this result, knockdown of Nrf2 expression in Fra-1^Δ/Δ cells restored their sensitivity to oxidants accompanied by caspase activation as well as decreased antioxidant (e.g., Hmox1) gene expression (Fig. 10).

Nucleocytoplasmic trafficking of Nrf2 is a critical step in the regulation of antioxidant gene expression (38). Under oxidative conditions, Nrf2 accumulates in the nucleus following release from its retention by the cytoplasmic inhibitor Keap1, which facilitates the degradation of Nrf2 via the ubiquitin-proteasome-dependent pathway under basal conditions (24). However, few studies have reported the presence of Keap1 in the nucleus and its role in the nuclear export of Nrf2 under stressful conditions (28, 38, 48). In general, a rapid turnover (degradation) of Nrf2 occurs in the cytosol via Keap1-dependent proteasome-mediated degradation under homeostatic conditions (for a review, see reference 25). Such degradation in the nucleus occurs slowly because of Nrf2 binding to the DNA as well as its escape from proteasomal degradation (26). We found that loss of Fra-1 leads to increased levels of nuclear Nrf2 under basal conditions and that the increase was further augmented in response to cellular stress. Such a rise in Nrf2 levels in Fra-1^Δ/Δ appears to occur in part due to a decreased protein turnover rate (Fig. 9A). We found that Fra-1 deficiency does not affect Keap1 expression under basal conditions and in response to oxidant treatment (Fig. 8E), suggesting that the nuclear accumulation of Nrf2 may not have been due to a decline in Keap1 levels under our experimental conditions. Previously, using DNA binding and overexpression studies, it was shown that Fra-1 represses Nrf2-induced Nqo1 promoter activation directly by binding to the functional ARE under basal conditions in hepatoma cells (40). Thus, it is possible that the loss of Fra-1 favors the nuclear retention and increased binding of Nrf2 to the ARE, thereby impeding its degradation, leading to enhanced transcription of antioxidant genes. Indeed, the magnitude of Nrf2 binding to antioxidant genes such as Hmox1 and Nqo1 remained high in Fra-1^Δ/Δ cells even at 6 h after oxidant treatment, and this binding was accompanied by elevated levels of Nrf2 in the nucleus (Fig. 8). In contrast, in Fra-1^+/+ cells the Nrf2 binding at Hmox1 and Nqo1 promoters at 6 h post-oxidant treatment was restored to the basal level seen in untreated cells. Control experiments performed with antibodies specific for RNA polymerase II showed that the chromatin from the Fra-1^+/+ cells and the chromatin from Fra-1^Δ/Δ cells treated with diquat for 6 h were equally accessible to precipitation (data not shown), suggesting that the increased levels of Nrf2 occupancy at the Hmox1 and Nqo1 promoters in Fra-1^Δ/Δ cells may not have been due to greater accessibility of their chromatin to Nrf2 immunoprecipitation compared to that of Fra-1^+/+ cells. Based on these observations, we propose that Fra-1 counterbalances Nrf2-dependent ARE-mediated transcriptional response both by modulating expression levels and nuclear accumulation of Nrf2 and by modulating the levels of Nrf2 binding to the ARE, thereby promoting cell death induced by oxidants (see schematic in Fig. 11C).

Despite high levels of Nrf2 in the nucleus, we did not observe increased binding of Nrf2 to the promoters of Hmox1 and Nqo1 in the Fra-1^Δ/Δ cells under basal conditions, suggesting that the elevated basal expression of at least these two genes observed in these cells may be, at least in part, attributable to the binding of other factors, in addition to Nrf2, that can potentiate Nrf2-dependent ARE-mediated transcriptional activation. For example, it was shown that small Mafs (Maf F/G/K), which are essential dimeric partners of Nrf2, differentially contribute to antioxidant gene expression (31). Knockdown of Maf G leads to an attenuation of both basal and electrophile-mediated induction of several antioxidant gene products, including Nqo1, whereas it derepresses Hmox1 expression under basal conditions (31). In addition, the transcription factor Bach1, when localized to the nucleus, represses Nrf2-dependent ARE transcriptional response by heterodimerizing with the small Mafs and also by facilitating nuclear export of Nrf2 and subsequent degradation (10, 40, 47). The nuclear export of Bach1, mediated by exportin CRM1, favors the Maf-Nrf2 interactions and subsequently promotes ARE-mediated gene expression (49, 50). ATF4 also heterodimerizes with Nrf2 and positively regulates ARE-mediated gene transcription (21). Because ATF4 is also a dimeric partner of Fra-1 (19), it is possible that loss of Fra-1 could favor Nrf2:ATF4 interactions and/or possibly Nrf2:c-Jun/Jun-D interactions, which are known to potentiate ARE-mediated gene transcription (5). Further studies are warranted to define how Fra-1 loss leads to a decline in Nrf2 turnover, its increased nuclear accumulation, and binding to the ARE and to examine the exact roles of other dimeric partners of Nrf2 (e.g., small Mafs, Bach1, and ATF4).

Nrf2 on its own does not bind to the DNA at the AREs (37). However, after dimerization with the MAF family of proteins, Nrf2 occupies the AREs and drives transcription (22). The Jun family of proteins, c-Jun and Jun-D, can dimerize with Nrf2 and upregulate antioxidant gene expression (27, 55). Jun^−/− and Jun-D^−/− MEFs exhibit decreased levels of antioxidant enzymes and enhanced oxidative stress (17, 35, 36, 58). In our studies, ChIP analysis revealed a decreased binding of c-Jun and Jun-D to the functional AREs of Hmox1 and Nqo1 in the absence of Fra-1 (Fig. 11). This result rules out the possibility that an increase in c-Jun or Jun-D binding at the AREs is a factor contributing to the increased resistance of the Fra-1^Δ/Δ cells to oxidant-induced cell death. This implies that the resistance to oxidant-induced cell death in cells lacking Fra-1 is due to increased binding of Nrf2, not c-Jun/Jun-D, at the AREs of the Nrf2 putative target genes Hmox1 and Nqo1.

In summary, the present report documents for the first time a hitherto-undescribed regulatory effect of Fra-1 on the Nrf2-mediated transcriptional response during oxidant-induced cell death. We have demonstrated a crucial role for Fra-1 in modulating cellular stress by altering intracellular antioxidant enzyme expression and GSH levels during exposure to oxidants. Given that the development of various acute and chronic diseases has been attributed to dysfunctional signaling induced by oxidative stress, targeting (suppressing) Fra-1 function may prove a promising approach to the treatment of diseases linked to oxidant exposure.