Molecular mechanism of transforming growth factor (TGF)-beta1-induced glutathione depletion in alveolar epithelial cells. Involvement of AP-1/ARE and Fra-1.

Glutathione (GSH) is a ubiquitous antioxidant in lung epithelial cells and lung lining fluid. Transforming growth factor beta1 (TGF-beta1) is a pleiotropic cytokine involved in cellular proliferation and differentiation. The level of TGF-beta1 is elevated in many chronic inflammatory lung disorders associated with oxidant/antioxidant imbalance. In this study, we show that TGF-beta1 depletes GSH by down-regulating expression of the enzyme responsible for its formation, gamma-glutamylcysteine synthetase (gamma-GCS) and induces reactive oxygen species production in type II alveolar epithelial cells (A549). To investigate the molecular mechanisms of inhibition of glutathione synthesis, we employed reporters containing fragments from the promoter region of the gamma-GCS heavy subunit (h), the gene that encodes the catalytic subunit of gamma-GCS. We found that TGF-beta1 reduced the expression of the long gamma-GCSh construct (-3802/GCSh-5'-Luc), suggesting that an antioxidant response element (ARE) may be responsible for mediating the TGF-beta1 effect. Interestingly, the electrophoretic mobility shift assay revealed that the DNA binding activity of both activator protein-1 (AP-1) and ARE was increased in TGF-beta1-treated epithelial cells. The gamma-GCSh ARE contains a perfect AP-1 site embedded within it, and mutation of this internal AP-1 sequence, but not the surrounding ARE, prevented DNA binding. Further studies revealed that c-Jun and Fra-1 dimers, members of the AP-1 family previously shown to exert a negative effect on phase II gene expression, bound to the ARE sequence. We propose a novel mechanism of gamma-GCSh down-regulation by TGF-beta1 that involves the binding of c-Jun and Fra-1 dimers to the distal promoter. The findings of this study provide important information, which may be used for the modulation of glutathione biosynthesis in inflammation.

Glutathione, or L-␥-glutamyl-L-cysteinylglycine (GSH), is a ubiquitous non-protein thiol that can eliminate reactive oxygen species and free radicals, such as hydrogen peroxide, superoxide, and the hydroxyl radical by sacrificing its sulfhydryl group (1). GSH is required for signal transduction, immune modulation, maintenance of surfactant, remodeling of extracellular matrix, regulation of apoptosis, proliferation, mitochondrial respiration, and control of pro-inflammatory processes in the lungs (2). This tripeptide is the principal antioxidant in the lung and is present in large quantities in the epithelial lining fluid, presumably due to release from type II cells (3). Several disorders, such as acute respiratory distress syndrome (4), cystic fibrosis (5), and idiopathic pulmonary fibrosis (IPF) 1 (6), are characterized by a depletion of this essential antioxidant in the airways, suggesting a role for oxidative stress in the pathogenesis of these chronic inflammatory lung diseases. The underlying causes of these diseases, in particular IPF, are unknown, and an effective antioxidant/anti-inflammatory treatment strategy remains to be developed. Recent studies have suggested that the depletion of the antioxidant defense shield in the lungs of IPF patients may leave the respiratory epithelium more susceptible to oxidant-mediated damage and subsequent fibrosis (7,8). The mechanism of glutathione depletion in patients with IPF remains to be elucidated; however, there is now evidence to indicate that the cytokine transforming growth factor-␤ 1 (TGF-␤ 1 ) may be involved (9). TGF-␤ 1 , which is elevated in IPF, mediates fibrosis by inducing fibroblast proliferation, differentiation, and extracellular matrix production (10). Previous studies have shown that TGF-␤ 1 can deplete glutathione levels in alveolar epithelial cells in vitro, although the molecular mechanisms that regulate this process have not yet been investigated. A greater understanding of the mechanism that leads to the depletion of glutathione in the lungs of IPF patients may aid the development of effective antioxidant treatment strategies. GSH formation is controlled by the actions of the enzymes ␥-glutamylcysteine synthetase (␥-GCS) and glutathione synthetase, with the former enzyme catalyzing the rate-limiting step (1). ␥-GCS is a holoenzyme comprised of a heavy chain (␥-GCSh, 73 kDa) and a light subunit (␥-GCSl, 28 kDa), which functions to stabilize the enzyme and is therefore termed the regulatory chain (11). As the entire catalytic activity of the enzyme resides within the heavy chain, regulation of glutathione synthesis is routinely considered in terms of ␥-GCSh gene expression.
GSH and ␥-GCS expression are induced by a variety of * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  agents, such as oxidants, phenolic antioxidants, and inflammatory mediators (12). The molecular mechanism of ␥-GCSh upregulation has been extensively studied. We have previously shown that basal and inducible ␥-GCSh expression in alveolar epithelial cells is controlled by a TPA (12-O-tetradecanoylphorbol-13-acetate)-responsive element (TRE) situated between Ϫ269 and Ϫ263 bp in the 5Ј-flanking region of the promoter (13). Agents that induce intracellular oxidative stress, such as tumor necrosis factor-␣, hydrogen peroxide, menadione, cigarette smoke (13)(14)(15) and okadaic acid (16) have been shown to cause an initial depletion of GSH and a reduction in ␥-GCSh gene expression. This early response is followed by an increase in the DNA binding activity of the transcription factor activator protein 1 (AP-1), which binds to the TRE and induces transcription of the ␥-GCSh gene, resulting in elevated intracellular GSH concentrations (13,15,17). The transcriptional regulation of ␥-GCSh appears to be dependent on the stimulus and the cell type (12,18). In an elegant study, Mulcahy et al. (19) have suggested that basal ␥-GCSh expression in liver HepG2 cells is controlled by an antioxidant response element (ARE), which lies 3.1 kb upstream from the start of the gene. This ARE (denoted ARE4) contains an AP-1 binding site embedded within it, and subsequent studies showed that basal ␥-GCSh gene expression was mediated by the TRE, whereas transcription induced by the phenolic antioxidant ␤-naphthoflavone was regulated by the surrounding ARE4 (20). All of these studies showed up-regulation of ␥-GCSh expression at the transcriptional level in various cells. However, the molecular mechanism of ␥-GCS down-regulation has not been investigated so far. Identification and characterization of the regulatory elements controlling transcription of ␥-GCSh will provide information for the modulation of GSH synthesis in pathophysiology.
The aim of this study was to determine the molecular mechanism of TGF-␤ 1 -induced glutathione depletion in human alveolar epithelial cells (A549). TGF-␤ 1 caused a dose-and timedependent decrease in GSH and ␥-GCS activity, reduced ␥-GCSh mRNA expression, and caused a rise in intracellular ROS levels. We hypothesized that TGF-␤ 1 depletes GSH by down-regulating either AP-1-or ARE4-mediated ␥-GCSh gene expression. To investigate this, we employed both the short ␥-GCSh 1.1-kb construct and the longer ␥-GCSh 3.8-kb reporter systems, which have been used to study the regulation of ␥-GCSh in various cells (13,19). We show that TGF-␤ 1 enhanced expression of the short ␥-GCSh 1.1-kb reporter, but down-regulated the longer ␥-GCSh 3.8-kb construct, indicating that the ARE4 consensus site may play a role in the inhibition of ␥-GCSh by TGF-␤ 1 . Analysis of the DNA binding properties of the ARE4 consensus sequence showed a dramatic increase in binding activity and that the embedded AP-1 site, but not the surrounding ARE, was the critical response element. Supershift analysis revealed that the AP-1/ARE4 DNA binding complex was composed of c-Jun and Fra-1 dimers, AP-1 family members that have previously been shown to exert a negative effect on phase II gene expression (21). We propose a novel mechanism of ␥-GCSh down-regulation, where recruitment of c-Jun/Fra-1 proteins plays a critical role in controlling GSH levels in alveolar epithelial cells.

EXPERIMENTAL PROCEDURES
Materials-All reagents were purchased from Sigma unless stated otherwise.
Cell Culture-A549 type II lung epithelial cells were obtained from the European Collection for Animal Cell Cultures (ECACC number 86012804) and were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Labtech International), 2 mM L-glutamine, 100 g/ml penicillin, and 100 units/ml streptomycin (Invitrogen Life Technologies). The cells were grown to confluence at 37°C in a humidified atmosphere containing 5% CO 2 , washed with Ca 2ϩ /Mg 2ϩ -free PBS (PBS-CMF), and harvested with 0.25% trypsin, 1 mM EDTA in HBSS (Invitrogen). Following passage, the cells were seeded at a density of 2.2 ϫ 10 4 cells/cm 2 and cultured overnight until monolayers of 60 -80% confluency were formed. The cells were then treated with TGF-␤ 1 (R&D Systems, Oxon, UK) or solvent control (2 M HCl with 0.5 g/ml bovine serum albumin) for the indicated times.
Assessment of Total Cellular Glutathione Concentration (GSH ϩ GSSG)-Monolayers of A549 cells were treated with TGF-␤ 1 (1-5 ng/ ml) or solvent control for various times (1-72 h). The cells were then washed with PBS-CMF, harvested with trypsin/EDTA, and then washed again in PBS-CMF. Total intracellular glutathione was determined according to the method of Tietze (22) using dithiobis[2-nitrobenzoic acid]-GSSG/glutathione reductase recycling, modified for a 96well plate (23). The actual total concentration of glutathione in the samples was determined using linear regression to calculate the values obtained from a standard curve and expressed as nmol per mg of protein.
␥-GCS Activity Assay-␥-GCS activity was assessed using a coupled assay with pyruvate kinase and lactate dehydrogenase as described previously (11). The rate of decrease in absorbance at 340 nm was followed at 37°C. Enzyme specific activity was measured as mmol of NADH oxidized/min/mg protein, which is equal to 1 international unit (IU).
Assessment of ␥-GCSh mRNA by RT-PCR-Cells treated with TGF-␤ 1 or solvent control for the indicated times were washed with PBS-CMF, and total cellular RNA was isolated using TRIzol Reagent (Invitrogen) based on the method of Chomczynski and Sacchi (24). The RNA was reverse-transcribed, using M-MLV-RT (Promega) according to the manufacturer's instructions, and the resultant cDNA was stored at Ϫ20°C until required. The polymerase chain reaction (PCR) was performed using oligonucleotide primers chosen from the published sequences for human ␥-GCSh cDNA (25) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (26), and the reaction was conducted using the conditions described previously (14). The sequences of the primers used in PCR were ␥-GCSh: sense, 5Ј-GTG GTA CTG CTC ACC AGA GTG ATC CT-3Ј) and antisense, 5Ј-TGA TCC AAG TAA CTC TGG ACA TTC ACA-3Ј); GAPDH: sense, 5Ј-CC ACC CAT GGC AAA TTC CAT GGC A-3Ј and antisense, 5Ј-TC TAG ACG GCA GGT CAG GTC AAC C-3Ј. The RT-PCR product was then resolved on a 1.5% agarose gel, and the bands were visualized using a UVP High Performance Ultraviolet Transilluminator (Laboratory Products) and Grab-IT Version 2.5 software. The intensity of the ␥-GCSh mRNA bands (531 bp) was expressed as a percentage of the intensity of the GAPDH bands (600 bp) with the aid of GelBase/GelBlot software. The pKS-hGCS plasmid (American Type Culture Collection, Manassas, VA; I.M.A.G.E clone ID 79023) was used as a positive control for ␥-GCSh PCR specificity.
Assessment of Intracellular Reactive Oxygen Species Production-Cells exposed to TGF-␤ 1 or solvent control were incubated with 40 M of the fluorescent probe dichlorodihydrofluorescein diacetate (H 2 DCFDA) for 30 min at 37°C, washed with PBS-CMF, harvested, and then washed again. The degree of fluorescence, which correlated to the level of intracellular ROS, was determined using a FACSCalibur flow cytometer (BD PharMingen) with an excitation wavelength of 530 nm (emission, 488 nm). The proportion of fluorescent cells was determined using CellQuest software (BD PharMingen) on a G3 workstation (Apple MacIntosh).
Generation of Reporter Constructs-The reporter construct Ϫ1050/ GCSh-5Ј-CAT (pCBGCS) was created by cloning the fragment from Ϫ1050 to ϩ82 bp of the human ␥-GCSh promoter into the plasmid pCAT Basic Vector (Promega), which has CAT activity and has been previously described (13). The recombinant plasmid Ϫ3802/GCSh-5Ј-Luc was a kind gift from Professor R. T. Mulcahy (University of Wisconsin), which was generated by cloning a 4.2-kb sequence of the 5Јflanking region of the human ␥-GCSh promoter into the pGL3 basic vector (Promega), as described elsewhere (19).
Transient Transfection and Assessment of CAT and Luciferase Activities-A549 cells were grown until they reached 60 -70% confluence and were then transiently transfected with 2 g of plasmid DNA using LipofectAMINE Reagent (Invitrogen) for 20 h. The cells were then allowed to recover for 24 h prior to treatment with 3 ng/ml TGF-␤ 1 or solvent control for 24 h. Following treatment, the monolayers were washed with PBS-CMF, and cellular extracts were prepared using reporter lysis buffer (Promega) for the CAT-transfected cells or luciferase lysis buffer (25 mM Tris phosphate buffer, pH 7.8, 8 mM MgCl 2 , 1 mM dithiothreitol, 1% Triton X-100, 15% glycerol) for the luciferase-transfected cells. Luciferase activity was assessed immediately by adding the extracts to luciferin reagent (0.25 mM luciferin, 1% bovine serum albu-min, and 1 mM ATP in luciferase lysis buffer) and measuring the degree of light intensity generated with a Biomac 2500 Luminometer. CAT activity was determined using a CAT enzyme-linked immunosorbent assay (ELISA) kit (Roche Molecular Biochemicals). Transfection efficiency was monitored by co-transfecting the cells with a ␤-galactosidase expression vector (PSVgal) (Promega), and ␤-galactosidase activity was measured in the extracts using an ELISA kit. In all experiments empty vectors were used as negative controls.
Electrophoretic Mobility Shift Assay (EMSA) and Supershift-Nuclear extracts were prepared on ice according to the method of Staal et al. (27). The EMSAs were conducted using a commercially available AP-1 oligonucleotide (5Ј-CGC TTG ATG AGT CAG CCG GAA-3Ј, obtained from Promega) and oligonucleotides for ARE4 consensus and mutant sequences (20) (Table II) and ␥-GCSh AP-1 consensus and mutant sequences (13) that were specifically synthesized (MWG Biotech, Ebersberg, Germany). Electrophoresis was conducted on 10 g of nuclear protein as previously described (13). A549 cells treated with 10 ng/ml tumor necrosis factor-␣ (R&D Systems) for 24 h (15) or HeLa nuclear extract (Promega) served as a positive control for DNA binding, and a negative control was established by substituting the nuclear extracts for distilled water. To prove specificity of binding, the nuclear extracts were preincubated for 10 min with a 100-fold molar excess of either non-labeled AP-1, ARE4. or NF-B oligonucleotides (Promega) prior to electrophoresis, which acted as cold-and non-competitors, respectively. To characterize the particular DNA-protein complexes, the reaction mixture described above was preincubated with 2 l of various antibodies or appropriate preimmune sera for 2 h prior to running the gel overnight at 4°C; the antibodies used were anti-c-Jun (KM-1), anti-c-Fos (4), anti-Fra-1 (R-20), and anti-Nrf2 (C-20) (Santa Cruz Biotechnology, Santa Cruz, CA).
Statistical Analysis-Individual experiments were conducted in triplicate, and the data represent the mean Ϯ S.E. (n ϭ 3) unless stated otherwise. Statistical significance was determined using one way analysis of variance with post-hoc Tukey's pairwise comparison (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001).

TGF-␤ 1 Depletes Glutathione Concentration in Alveolar Ep-
ithelial Cells-Treatment of A549 epithelial cells with various amounts of TGF-␤ 1 for 72 h caused a dose-dependent depletion of GSH (p Ͻ 0.001) (Fig. 1A). Exposure of the epithelial cells to 3 ng/ml TGF-␤ 1 for 6 h or longer also caused a decrease in intracellular glutathione concentrations (p Ͻ 0.01 and p Ͻ 0.001) (Fig. 1B). In preliminary studies, treatment of primary rat type II epithelial cells with 5 ng/ml TGF-␤ 1 for 72 h also depleted intracellular GSH from 0.99 Ϯ 0.01 to 0.34 Ϯ 0.01 nmol/mg protein, a decrease of 66%.
TGF-␤ 1 Decreases ␥-GCS Activity in Alveolar Epithelial Cells-We also investigated the effects of TGF-␤ 1 on the activity of ␥-GCS, the enzyme responsible for catalyzing the ratelimiting step in glutathione formation (Fig. 2). The growth factor caused a dose-dependent reduction in ␥-GCS activity when the epithelial cells were exposed for 72 h ( Fig. 2A). TGF-␤ 1 (3 ng/ml) also induced a significant decline in ␥-GCS activity when the cells were treated for 24, 48, and 72 h (Fig. 2B).
TGF-␤ 1 Reduces ␥-GCSh mRNA-We examined the effects of TGF-␤ 1 on ␥-GCSh mRNA expression by RT-PCR (Fig. 3). A549 cells treated with 3 ng/ml TGF-␤ 1 for 24, 48, and 72 h had significantly less ␥-GCSh mRNA than control cells. Similar results were found in preliminary studies using primary rat type II cells that were exposed to 5 ng/ml TGF-␤ 1 for 72 h, as ␥-GCSh mRNA was reduced to 76 Ϯ 1% of the control value.
TGF-␤ 1 Induces Intracellular ROS Production in Alveolar Epithelial Cells-To determine the effects of GSH depletion on intracellular redox status, we measured ROS levels in epithelial cells using the fluorescent probe H 2 DCFDA and flow cytometry (Fig. 4A). TGF-␤ 1 caused a dose-dependent accumulation of ROS in epithelial cells treated for 72 h (Fig. 4B). When epithelial cells were exposed to TGF-␤ 1 (3 ng/ml) for 24, 48, and 72 h, there was also a significant increase in oxidative stress (Fig. 4C). This suggests that TGF-␤ 1 is having a profound effect on the redox balance and that intracellular ROS may be re-sponsible for initiating additional signaling cascades within the epithelial cells.
Effects of TGF-␤ 1 in Separate ␥-GCSh Reporter Systems-We hypothesized that TGF-␤ 1 inhibits ␥-GCSh gene expression by exerting a negative effect on the promoter. To investigate this possibility, we employed two reporter constructs, Ϫ1050/GCSh-5Ј-CAT and Ϫ3802/GCSh-5Ј-Luc, which have previously been used to study ␥-GCSh gene expression. Fig. 5A shows a schematic drawing of the two reporters, detailing the relative base pair positions of the transcription factor binding sites. In each experiment a promoter-less vector control was used, but did not yield any significant reporter gene expression. TGF-␤ 1 stimulated CAT activity, but repressed luciferase expression in alveolar epithelial cells at 24 h (Fig. 5B). This suggests a role for the ARE4 site in controlling TGF-␤ 1 -mediated down-regulation of ␥-GCSh.
TGF-␤ 1 Increases the DNA Binding Activity of AP-1 and ARE4 in A549 Cells-We hypothesized that TGF-␤ 1 may stimulate Ϫ1050/GCSh-5Ј-CAT by increasing AP-1 DNA binding and down-regulate Ϫ3802/GCSh-5Ј-Luc by reducing ARE4 activity. Surprisingly, TGF-␤ 1 actually enhanced both AP-1 (Fig.  6, A and B) and ARE4 activity (Fig. 6, C and D), with a noticeable difference after 8 h of treatment. The DNA binding activity of both responsive elements increased with time and was maximal after 72 h of treatment. TGF-␤ 1 also increased the DNA binding activity of the native ␥-GCSh AP-1-like sequence to a similar extent as commercial AP-1 (data not shown).
Unlike the majority of AREs identified to date, the ␥-GCSh ARE4 nucleotide sequence contains a perfect AP-1 site within it (some examples are given in Table I). We synthesized a series of ARE4 oligonucleotide mutants (20) in which the consensus AP-1 and ARE4 sites were changed, to determine which particular bases were required for the TGF-␤ 1 -induced effect. Table II describes the following consensus and mutant (mut) sequences: in ARE4 mut1, both the AP-1 and the ARE4 consensus sites were affected; in ARE4 mut2, only the AP-1 binding site was disrupted as these bases are not required for ARE binding (28,29); in ARE4 mut3, only the ARE4 sequence was mutated as these bases extend beyond the AP-1 site.
Exposure of A549 cells to 3 ng/ml TGF-␤ 1 for 72 h caused a substantial increase in the DNA binding activity of ARE4 (Fig.  7A, compare lanes 3 and 4). This DNA binding was completely eliminated by mutating both the core ARE4 and AP-1 sites (ARE4 mut1) (Fig. 7A, lanes 5 and 6) or the consensus AP-1 site alone (ARE4 mut2) (Fig. 7A, lanes 7 and 8). Mutation of the consensus ARE4 sequence (ARE4 mut3) (Fig. 7A, lanes 9 and 10) did not prevent DNA binding; however, the intensity of the bands for both the control and TGF-␤ 1 -treated extracts is slightly diminished.
To confirm these findings, the commercial AP-1 and mutated ARE4 oligonucleotides were used as cold competitors to establish whether or not they could compete with the consensus ARE4 oligonucleotide for DNA binding. The nuclear extracts prepared from A549 cells treated with 3 ng/ml TGF-␤ 1 or solvent control for 72 h were incubated with a 100-fold molar excess of non-labeled oligonucleotide prior to addition of radiolabeled ARE4 (Fig. 7B). Both the consensus ARE4 (Fig. 7B,  lanes 5 and 6) and the commercial AP-1 oligonucleotides (Fig.  7B, lanes 13 and 14) were able to effectively compete with radiolabeled ARE4 for DNA binding, indicated by the absence of bands in these lanes. ARE4 mut1 (Fig. 7B, lanes 7 and 8) and ARE4 mut2 (Fig. 7B, lanes 9 and 10), however, were unable to prevent the association of ARE4 with the nuclear extracts because the AP-1 site was disrupted. Mutation of only the core ARE4 sequence (ARE4 mut3) resulted in a probe that was able to compete for DNA binding (Fig. 7B, lanes 11 and 12). These data strongly suggest that AP-1 proteins are binding to the ARE4 sequence in response to TGF-␤ 1 treatment.

FIG. 2. TGF-␤ 1 decreases ␥-GCS activity in alveolar epithelial cells.
A, cells were treated with TGF-␤ 1 (1-5 ng/ml) or solvent control (0 ng/ml) for 72 h. B, cells were exposed to 3 ng/ml TGF-␤ 1 (filled histograms) or control (open histograms) for 24, 48, and 72 h. ␥-GCS enzyme activity was assessed and expressed as a percentage of the control value. Each graph represents the mean of four experiments conducted in duplicate, and the bars represent the S.E. **, p Ͻ 0.01 and ***, p Ͻ 0.001, compared with control.

Characterization of the AP-1 Complex Induced by TGF-␤ 1 -
The particular AP-1 complexes that were being induced by TGF-␤ 1 were investigated by EMSA supershift analysis using anti-c-Jun, -c-Fos, -Fra-1, and -Nrf2 antibodies (Fig. 8). Of the antibodies tested, only c-Jun and Fra-1 formed complexes with the ARE4 oligonucleotide, which subsequently migrated more slowly through the gel (Fig. 8A, lanes 7-10). Identical results were obtained with the consensus AP-1 oligonucleotide (Fig.  8B, lanes 5-8). This indicates the recruitment of a c-Jun and  Fra-1 heterocomplex into the active AP-1/ARE4 region of the ␥-GCSh gene by TGF-␤ 1 in A549 cells. DISCUSSION Idiopathic pulmonary fibrosis and chronic obstructive pulmonary disease are characterized by elevated concentrations of TGF-␤ 1 and depleted levels of GSH in the lungs of patients (6,10,30,31), and in vitro studies have suggested that this growth factor has a direct effect in lowering the antioxidant capacity of alveolar epithelial cells (9). We attempted to elucidate the molecular mechanisms that mediate such an effect, in the hope of gaining a greater understanding of the pathogenesis of chronic inflammatory lung diseases, in particular IPF.
Previous studies have shown that exposure of alveolar epithelial and endothelial cells to TGF-␤ 1 depletes intracellular GSH in a dose-and time-dependent fashion, and our results were consistent with these findings (9,32,33). Our results also support earlier studies suggesting that the decline of GSH in TGF-␤ 1 -treated epithelial cells is caused by a reduction in ␥-GCS activity (9). To firmly establish that our findings mirrored those previously published, we investigated the effects of TGF-␤ 1 on ␥-GCSh mRNA expression and proved that TGF-␤ 1 is not only causing a decrease in ␥-GCS activity, but is also reducing the expression of the ␥-GCSh gene.
Exposure of epithelial cells to TGF-␤ 1 induced intracellular ROS production, measured by the fluorescent probe H 2 DCFDA. We attempted to establish a link between TGF-␤ 1 -mediated GSH depletion and ROS production by the use of ␥-GCS overexpression vectors, which have been described previously (34). Unfortunately, we were unable to increase GSH in cells overexpressing ␥-GCS (data not shown). This may be because of the constitutively high levels of ␥-GCS in A549 cells or the fact we used transiently transfected cells where others have used stable transfectants (34,35). It is therefore unclear at this time whether or not the rise in ROS production occurs because of direct induction of ROS-generating machinery, as seen in fibroblasts (36 -38), endothelial cells (39), or macrophages (40) or if ROS accumulation occurs as an indirect effect of GSH depletion. Furthermore, the effects of TGF-␤ 1 on GSH depletion could not be attenuated by the addition of N-acetylcysteine, catalase, or superoxide dismutase, indicating that the repression of ␥-GCS is not dependent on oxidative stress (data not shown). Additional studies are ongoing in our laboratory to identify the source of these ROS, which probably contribute to additional signaling cascades within these cells.
We have previously shown that an increase in AP-1 DNA binding activity is associated with an increase in Ϫ1050/GCSh-5Ј-CAT reporter activity, ␥-GCSh expression, and elevated glutathione levels in epithelial cells exposed to oxidative stress (15). Other investigators have shown that the critical response element for induction of Ϫ3802/GCSh-5Ј-Luc following xenobiotic treatment in hepatocytes is an antioxidant response element, denoted ARE4 (19). It was therefore hypothesized that TGF-␤ 1 may deplete intracellular GSH by inhibiting AP-1 or ARE4 DNA binding to the ␥-GCSh promoter, and subsequently this was investigated by reporter assay. We found that TGF-␤ 1 enhanced the activity of the short Ϫ1050/GCSh-5Ј-CAT reporter but inhibited expression of the long Ϫ3802/GCSh-5Ј-Luc construct. This decrease in reporter gene expression is not as great as the repression of ␥-GCSh mRNA that is observed in epithelial cells exposed to TGF-␤ 1 . This may be due to the fact that inhibitory transcription factors will bind to the endogenous promoter in addition to the reporter promoter, thereby limiting the repressive effect. It is also possible that additional, as yet unidentified, elements may play a role in mediating TGF-␤ 1 -induced repression of ␥-GCSh. However, these findings indicate that the ARE4 element present in the long Ϫ30802/ GCSh construct but not the AP-1 sequence present in the short Ϫ1050/GCSh promoter, could play a role in mediating the TGF-␤ 1 -induced effect. When we studied the activity of AP-1 and ARE4 in alveolar epithelial cells, we discovered that nuclear protein binding to both elements was increased following exposure to TGF-␤ 1 . There is extensive evidence showing that TGF-␤ 1 activates AP-1 in a number of different cell types (41)(42)(43)(44) and emerging data to suggest that TGF-␤ 1 may play a role in ARE activation (45). However, we were surprised that the ARE4 appeared to be activated to the same degree as AP-1, and when the two oligonucleotide probes were resolved on one gel, the complexes migrated to the same position, indicating that the DNA-binding proteins were of similar size (data not shown).
To establish which particular bases within the ARE4 sequence were important for DNA binding in alveolar epithelial cells, we synthesized a series of oligonucleotide mutants in which the consensus AP-1 and ARE4 sites were mutated (20). Simultaneous disruption of the AP-1 and ARE4 sequences, or the AP-1 site alone, resulted in a probe that was unable to associate with the nuclear extracts from either control or TGF-␤ 1 -treated cells. These findings were confirmed by using both the consensus and mutant sequences as competitors for DNA binding. These data suggest that AP-1 is the critical component of ARE4, as opposed to the ARE itself. Mutation of only the consensus ARE4 sequence (ARE4 mut3) did not prevent DNA binding; however, the intensity of the bands for both the control and TGF-␤ 1 -treated extracts was slightly reduced. These terminal GC residues, although not actually part of the AP-1 binding site, are required for maximal gene expression mediated by the human collagenase TRE (29), indicating that these residues may function to stabilize AP-1-DNA binding. Previous studies have shown that AP-1 is not the major ARE-binding protein; however, the ARE sequence used in these experiments was cloned from the mouse GST-Ya subunit gene, which does not contain a perfect AP-1 binding domain (see Table I) (46).
Having firmly established that AP-1 is the major DNA binding factor of interest, we proceeded to use EMSA supershift analysis to investigate which particular AP-1 complexes (Fos/ Jun) were being induced by TGF-␤ 1 . Nrf2, a transcription factor previously shown to stimulate ARE4-mediated ␥-GCSh upregulation in response to phenolic antioxidants and xenobiotics in hepatocytes (47), did not form a complex with ARE4 in alveolar epithelial cells. Of the antibodies tested, only c-Jun and Fra-1 formed complexes with the AP-1 or ARE4 oligonucleotide. AP-1 complexes containing Fra-1 have previously been shown to exhibit low transactivational potential (46), suggesting that the induction of these complexes are restricting ␥-GCSh expression. Other investigators have found that overexpression of Fra-1 repressed ARE-mediated induction of the human NAD(P)H:quinone oxidoreductase (NQO1) gene (21). Like ␥-GCSh ARE4, the ARE from this gene also contains a perfect AP-1 site embedded within it (summarized in Table I). The rat ␥-GCSh gene is also controlled by an antioxidant response element containing an AP-1 binding site, and overexpression of c-Jun repressed transcription (48). We have previ-ously shown that up-regulation of Ϫ1050/GCSh-5Ј-CAT is associated with an increase in c-Jun binding to the ␥-GCSh AP-1-like sequence. It is therefore probable that the elevated levels of c-Jun in TGF-␤ 1 -treated alveolar epithelial cells accounts for the stimulation of this reporter. This study supports earlier findings that regulation of ␥-GCSh gene expression is dependent on cell type (18), indicating that epithelial cells and hepatocytes probably differ slightly in their composition of transcription factors. We suggest a model where the composition of AP-1 transcription factors determines whether the gene will be induced or repressed, which is outlined in Fig. 9.
We propose a novel mechanism of ␥-GCSh gene down-regulation where recruitment of Fra-1 (a negative modulator of phase II genes) into the active AP-1⅐ARE4 complex occurs in response to TGF-␤ 1 in epithelial cells. Although the studies presented here have been solely conducted in an epithelial cell line, preliminary studies in primary rat type II epithelial cells support our findings. As TGF-␤ 1 is a critical mediator of pulmonary fibrosis (32,49,50), which is also characterized by depleted GSH (51,7,8), these findings can be assumed to have general implications for inflammatory lung diseases. However, this study cannot rule out the possibility of an additional, but as yet unidentified, transcription factor that binds to the ARE4 in cells exposed to TGF-␤ 1 . Effectors of the TGF-␤ 1 -mediated signaling cascades are capable of recruiting a large variety of such negative elements, for example c-Ski (52), SnoN (53), and 5Ј-TG-3Ј interacting factor (TGIF) (54), which can interact directly with Smad proteins and indirectly with other transcription factors, such as AP-1. It is possible that the products of such protein-protein interactions could bind to the ARE4 or other sites such as a Smad binding element, present in the distal ␥-GCSh promoter. This new concept involved in the FIG. 9. Working model for regulation of ␥-GCSh in different cell types. A, basal ␥-GCSh expression in alveolar epithelial cells is regulated by c-Jun dimers, which bind to an AP-1-like responsive element in the proximal promoter. B, oxidative stress, from tumor necrosis factor-␣ for example, increases the levels of c-Jun and stimulates ␥-GCSh gene expression in epithelial cells. C, TGF-␤ 1 induces the formation of c-Jun and Fra-1 heterodimers, which bind to the AP-1 site embedded within the ARE4 sequence in the distal promoter and down-regulate ␥-GCSh gene expression. D, basal ␥-GCSh gene expression in hepatocytes is controlled by Nrf2/bZIP dimers binding to the AP-1 sequence and the initial portion of ARE4 in the distal promoter. E, phenolic antioxidants and xenobiotics induce the formation of Nrf2 and either JunD or small Maf heterodimers, which bind to the complete ARE4 sequence in the distal ␥-GCSh promoter. regulation of glutathione synthesis requires further experimentation, which is currently ongoing in our laboratory. Nevertheless, it is tempting to hypothesize that a variety of phase II AP-1-dependent genes, which are modulated by TGF-␤ 1 during inflammatory responses (55), may be regulated by such protein-protein interactions (e.g. formation of a c-Jun⅐Fra-1 complex). Studies on the molecular regulation of those genes may provide a novel mechanism where specific therapeutic strategies can be made.
In conclusion, these studies show that TGF-␤ 1 imposes an oxidant/antioxidant imbalance in alveolar epithelial cells, which is a hallmark of various chronic inflammatory lung diseases. TGF-␤ 1 down-regulates the expression of ␥-GCSh at the transcriptional level, which is associated with activation of AP-1 and ARE. Supershift experiments revealed that TGF-␤ 1mediated down-regulation of the ␥-GCSh gene was associated with recruitment of a c-Jun⅐Fra-1 complex to the distal ARE4 responsive element, which exerts a negative effect on ␥-GCSh gene expression in alveolar epithelial cells. These data indicate a novel molecular mechanism of ␥-GCSh down-regulation by TGF-␤ 1 , which may be used for the modulation of glutathione biosynthesis in chronic inflammatory lung diseases such as IPF.