Delayed mechanism for induction of gamma-glutamylcysteine synthetase heavy subunit mRNA stability by oxidative stress involving p38 mitogen-activated protein kinase signaling.

Expression of the gamma-glutamylcysteine synthetase heavy subunit (gamma-GCSh), which encodes the rate-limiting enzymes for glutathione biosynthesis, is regulated by many cytotoxic agents. Moreover, gamma-GCSh mRNA expression is elevated in colorectal cancer, but how gamma-GCSh expression is regulated is not completely understood. By using actinomycin D, which inhibits new RNA synthesis, we showed that treatment of human colorectal cancer cells with the prooxidant sulindac increased the half-life of gamma-GCSh mRNA. By using a tetracycline-regulated gamma-GCSh mRNA assay system, we systematically dissected the cis-acting sequence and trans-acting factors that regulate the stability of gamma-GCSh by cytotoxic prooxidants. We demonstrated that a HuR recognition sequence, AUUUA, in the 3'-untranslated region is responsible for the decay of gamma-GCSh mRNA. Oxidative stress enhanced cytoplasmic content of HuR. Overexpression of HuR by transfection stabilized gamma-GCSh mRNA, whereas overexpression of a dominant-negative HuR mutant suppressed the induced stability. Furthermore, prooxidant-induced gamma-GCSh mRNA stabilization and HuR binding were blocked by p38 mitogen-activated protein kinase inhibitors. We provide the first evidence that reduction-oxidation regulation of gamma-GCSh expression, itself a reduction-oxidation sensor and regulator, is mediated at least in part by the p38 mitogen-activated protein kinase signaling through the HuR RNA-binding protein.

production of reactive oxygen species exerts oxidative stress, leading to a host of pathologic consequences. One of the ways a cell regulates redox homeostasis is through the glutathione system that exists in both the reduced form (GSH) and the oxidized form (GSSG) (1). Under oxidative stress, GSH is oxidized by GSH peroxidase to GSSG that is eliminated by the MRP1 (multidrug resistance associated protein) efflux pump or is catalytically reduced back to GSH by the NADPH-dependent GSH reductase. GSH also regulates the activities and biosynthesis of other redox-regulating enzymes, such as superoxide dismutase and DT-diphorases (NADPH quinine oxidoreductases 1 and 2). Because of the intracellular abundance of glutathione (1-10 mM) (2,3), the GSH/GSSG system is the main redox regulator of cells.
We demonstrated previously that expression of ␥-GCSh could be induced by many cytotoxic agents, including antitumor agents (10,11), heavy metals (11), carcinogens (12), and prooxidants (13)(14)(15)(16). All of these treatments, to various extents, induce intracellular reactive oxygen species imbalance. Moreover, ␥-GCSh mRNA is frequently overexpressed in human colon cancers (17), which are associated with redox imbalance (18), suggesting that the GSH/␥-GCS system is a molecular sensor of intracellular redox homeostasis. Although elevated expression of ␥-GCSh catalyzes enhanced expression of GSH, we observed that increased GSH expression provides feedback for the down-regulation of steady-state ␥-GCSh mRNA expression (16). In addition, high GSH expression has been reported previously to suppress ␥-GCSh enzymatic activities (19). This feedback mechanism underscores the importance of ␥-GCSh as a redox regulator.
We reported previously that steady-state expression of ␥-GCSh mRNA increases by severalfold in human glioma cells treated with the alkylating antitumor agent 1-(4-amino-2methyl-5-pyrimidinyl) methyl-3-(2-chloroethyl)-3-nitrosourea (10). However, nuclear run-on assays revealed less than a 0.5-fold increase in the transcriptional rate. These results suggest that post-transcriptional regulation is also involved in the increased steady-state expression, perhaps depending upon prooxidants and cell sources. The mechanism involved in this post-transcriptional regulation has yet to be elucidated. The objective of the current study was to gain insight into the post-transcriptional mechanism of redox-induced ␥-GCSh upregulation in a cultured cell system. We found that oxidative stress-induced ␥-GCSh expression is mediated by the mRNA-stabilizing protein HuR, which interacts with an AUrich sequence in the 3Ј-untranslated region (UTR) of ␥-GCSh mRNA through the p38 mitogen-activated protein kinase (MAPK) signal transduction pathway.

MATERIALS AND METHODS
Plasmid DNA and Construction of Deletion Mutants-The Ϫ3802/ GCSh5Ј-luc recombinant DNA (20), which contains nucleotides (nt) Ϫ3802 to ϩ465 of the ␥-GCSh sequence in the luciferase reporter vector pGL3 basic (Promega, Madison, WI), was obtained from R. T. Mulcahy (University of Wisconsin Medical School, Madison, WI). A series of progressively deleted flanking sequences was created by PCR by using Ϫ3802/GCSh5Ј-luc DNA as a template and appropriate primer sets. The PCR products were cloned into the pGL3 basic vector, generating Ϫ814/GCSh5Ј-luc, Ϫ202/GCSh5Ј-luc, Ϫ149/GCSh5Ј-luc, and Ϫ22/ GCSh5Ј-luc recombinants.
Wild-type and antisense HuR (35) were obtained from M. Gorospe (National Institute on Aging, Bethesda). Plasmids encoding the constitutively active and dominant-negative forms of apoptosis signal-regulating kinase 1 (ASK1), HA-ASK1-⌬N and HA-ASK1-KM (36), respectively, were obtained from H. Ichijo (The university of Tokyo, Tokyo, Japan), and the constitutively active and dominant-negative forms of MAPK-activated protein kinase 2 (MAPKAPK2), Aspx3, and Ala-207 (37), respectively, were obtained from C. Marshall (Institute of Cancer Research, London, UK). Plasmids encoding the wild-type and dominant-negative forms of p38 MAPK were provided by U. Rapp (Institut fü r Medizinische Strahlenkunde und Zellforschung, Universität Wü rzburg, Wü rzburg, Germany) and P. Scherer (Albert Einstein College of Medicine of Yeshiva University, New York). Plasmids encoding the wild-type and dominant-negative forms of MAPK kinase 3 (MKK3) were obtained from P. Chiao (University of Texas M. D. Anderson Cancer Center, Houston).
To generate site-directed mutations in the AU-rich sites, we used the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions. Briefly, pTRE-␥-GCSh(2983) DNA was used as the template in PCRs using primers containing ϩ2785 AGGGA and ϩ2595 AGGGGA to replace ϩ2785 AUUUA (site II) and ϩ2595 AUUUUA (site I), respectively. The methylated, wildtype DNA (template) was digested by DpnI, and the nonmethylated, mutated DNA (PCR product) was transformed into XL1-Blue supercompetent cells (Stratagene). All the plasmid DNAs were confirmed by sequencing.
Cell Culture and Treatments with Prooxidants-HT-29 and HCT-15 human colorectal cancer cells were purchased from the American Type Culture Collection (Manassas, VA). The cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum at 37°C in a 5% CO 2 atmosphere. Cells at exponential growth conditions were treated with prooxidants at the following concentrations: 100 M 2-acetylaminofluorene (2-AAF; Sigma), 50 M menadione (Sigma), 100 M pyrrolidine dithiocarbamate (PDTC; Sigma), 100 M tert-butylhydroquinone (tBHQ; Sigma), and 800 M sulindac (Sigma). In experiments where inhibitors were used, cells were treated with 25 mM N-acetylcysteine (Sigma) 2 h before the addition of sulindac, whereas 20 M SB203580 (Calbiochem) and 20 M PD98059 (Calbiochem) were added simultaneously with sulindac. Total RNA was extracted from the cells, and levels of ␥-GCSh mRNA were determined by the RNase protection assay.
Transient Transfection-Transient transfection was performed by using Lipofectamine (Invitrogen) according to the manufacturer's instructions. Briefly, HEK293T cells (1 ϫ 10 6 cells) were seeded into 60-mm Petri dishes; 24 h later, recombinant pTRE-␥-GCSh plasmid DNA and EC1214A plasmid, which expresses tTA regulation factor, were introduced into cells by using 8 g of Lipofectamine (the amount of DNA was kept constant at 4.5 g within experiments by adding empty vector as required). After 5 h of incubation, the medium was replaced with Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. To shut off mRNA synthesis, tetracycline (2 g/ml, Sigma) was added 24 h after transfection, and cells were harvested 0, 2, 4, 6, 9, and 12 h later in guanidine thiocyanate lysis buffer (STAT-60; Tel-Test, Friendswood, TX). RNA was extracted through phenol/chloroform extraction and isopropyl alcohol precipitation.
Luciferase Assay-HEK293T cells (2 ϫ 10 5 cells) were seeded in 24-well plates, and 24 h later, 0.2 g of ␥-GCSh reporter recombinant constructs and 10 ng of pRL-SV40 Renilla luciferase vector were cotranfected into cells using 4 g of Lipofectamine. After 5 h of incubation, the medium was replaced with Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. After 24 h of transfection, cells were treated with 100 M tBHQ for 12 h, and the cells were washed twice with 1ϫ phosphate-buffered saline and lysed with passive lysis buffer (Promega). Firefly and Renilla luciferase activities in aliquots (10 l) of cell lysates were measured according to the manufacturer's instructions.
RNase Protection Assay-RNase protection was carried out as described previously (10). In brief, the template DNA was linearized by digestion with appropriate restriction enzymes and purified by phenol/ chloroform extraction and ethanol precipitation. Antisense riboprobes were synthesized using T7 or Sp6 polymerase (Roche Applied Science) in the presence of 500 Ci of [␣-32 P]UTP (3,000 Ci/mmol; MP Biomedicals, Irvine, CA). The hybridization mixtures contained 20 g of RNA and 2 ϫ 10 5 cpm of riboprobe. Unhybridized RNA was degraded by incubation with RNase A (12 g; Roche Applied Science) and RNase T (4 g; Roche Applied Science). Protected RNA fragments were resolved by electrophoresis on denaturing 8% acrylamide, 8 M urea gels and were visualized by autoradiography. To determine the endogenous ␥-GCSh mRNA, we used a probe that generated a protected fragment of 268 nt.
To determine exogenous ␥-GCSh transcripts from the transfected DNA, we constructed a plasmid that contains ␥-GCSh cDNA (ϩ220 to 470 bp) and used it as a template. The riboprobe synthesized from this template annealed to the transfected (exogenous) ␥-GCSh mRNA and gave rise to a protected fragment of 70 nt. Under these conditions, this probe did not detect endogenous ␥-GCSh mRNA (data not shown). In all cases, the autoradiographic signals were quantified with an image analyzer (ImageQuant 5.0). The half-life (t1 ⁄2 ) of the ␥-GCSh mRNA was calculated from the exponential phase of the plot of remaining mRNA versus time. In all experiments, hybridization of 18 S rRNA was used as an internal control for sample loading. The intensities of the 18 S rRNA and ␥-GCSh mRNA bands were adjusted independently within the linear range of the signals.
Western Blot Analysis-HEK293T cells (2 ϫ 10 6 cells) were seeded onto a 100-mm Petri dish. Twelve h later, cells were treated with sulindac (800 M), PDTC (100 M), or tBHQ (100 M) alone or with SB203580 (20 M) or PD98059 (20 M) for 12 h. Cells were washed three times with phosphate-buffered saline and harvested. Cell were suspended in 200 l of phosphate-buffered saline containing 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, and protease inhibitors (Complete; Roche Applied Science) and centrifuged at 12,000 rpm for 10 min at 4°C. Protein concentrations were determined using a protein assay kit (Bio-Rad). Aliquots (40 g) of protein were resolved by SDS-PAGE and transferred to polyvinylidene difluoride membranes (Millipore, Bedford, MA) by using standard procedures. The membranes were then subjected to Western blotting, and the blots were developed with the enhanced chemiluminescence system (Amersham Biosciences).
To investigate the effect of prooxidant on intracellular distribution of HuR between the cytoplasm and nucleus, we treated the cells with 800 M sulindac or 100 M tBHQ in the presence or absence of the inhibitors SB203580 or PD98059 for 12 h. Cytoplasmic and nucleic fractions were prepared from the treated cells using the NE-PER extraction kit (Pierce) and were processed by a standard Western blot using mouse monoclonal antibodies against HuR (Santa Cruz Biotechnology, 1:1,000), ␣-tubulin (Santa Cruz Biotechnology, 1:1,000), and lamin B (Calbiochem, 1:200).
To deplete HuR from the cell extracts, 10 g of extract was incubated with 200 ng of anti-HuR antibody followed by an addition of 5 l of protein A/G plus-agarose (Santa Cruz Biotechnology) for 2 h at 4°C. The reaction mixture was centrifuged for 2 min at 14,000 rpm at 4°C. The supernatant was incubated with radiolabeled site II wild-type RNA oligonucleotide and assayed by gel mobility shift assay. As a control, the cytoplasmic fraction was incubated with protein A/G plus-agarose without anti-HuR antibody.
In Vivo Cross-linking-RNA and protein were cross-linked by the method of Cook et al. (66) with minor modifications. HEK293T cells (2 ϫ 10 6 cells) were seeded onto a 100-mm Petri dish. Twelve h later, cells were treated with sulindac (800 M) alone or with SB203580 (20 M) or PD98059 (20 M) for 12 h. Cells were washed twice with ice-cold phosphate-buffered saline and resuspended in 2 ml of phosphatebuffered saline in a 100-mm Petri dish. Cells were irradiated on ice with 254-nm UV light using a transilluminator (UVP, San Gabriel, CA) without filter at a distance of 4 cm for 15 min. The cell pellets were resuspended by vortexing in 200 l of lysis buffer (0.5% SDS, 50 mM Tris-Cl (pH 8), 1 mM EDTA, and 1 mM dithiothreitol) and boiled at 95°C for 5 min. After dilution with 800 l of correction buffer (1.25% Nonidet P-40, 0.625% sodium deoxycholate, 62.5 mM Tris-Cl (pH 8), 1.75 mM EDTA, 187.5 mM NaCl) containing protease inhibitor mixture (Roche Applied Science) and 150 units of RNase inhibitor (Roche Applied Science), the sample was gently sonicated two or three times and centrifuged for 90 min. The supernatants (1 ml total) were incubated with anti-HuR antibody (2 g) followed by 50 l of protein A/G plus-agarose (Santa Cruz Biotechnology) for 2 h at 4°C. The samples were divided into 2 aliquots; 1 aliquot of 800 l was centrifuged for 2 min at 14,000 rpm at 4°C. Pellets were washed four times with NET-2 and digested with proteinase K with tumbling at 37°C for 1 h. Following guanidine isothiocyanate extraction and ethanol precipitation, mRNA was analyzed by the RNase protection assay with probes for ␥-GCSh described above. The other aliquot of 200 l was washed four times with NET-2, suspended in electrophoresis loading buffer (15 l), boiled at 95°C for 5 min, and centrifuged. The supernatants were subjected to Western blotting using anti-HuR antibody as described above. Fig. 1A shows the time course induction of ␥-GCSh expression by various prooxidants, including sulindac, 2-AAF, tBHQ, and PDTC. Increased steady-state levels of ␥-GCSh mRNA were observed 4 -6 h and continued throughout the 10 -24-h treatment time. Levels of induction ranged from 3-to 8.5-fold, depending upon the prooxidants. These results are consistent with our previous reports (14 -16) that induction of ␥-GCSh expression by prooxidants was not an early event.

Post-transcriptional Regulation of ␥-GCSh mRNA Expression Is Induced by Prooxidants-
To investigate whether the induced up-regulation of ␥-GCSh by prooxidants is related to the transcriptional control, we analyzed the promoter activity of ␥-GCSh using a Ϫ3802/ GCSh5Ј-luc reporter construct in response to the prooxidant tBHQ. Ϫ3802/GCSh5Ј-luc plasmid contains ORE (20,21) and AP-1 (23)(24). Fig. 1B shows that treatment of 100 M tBHQ failed to increase an appreciable amount of luciferase activities. Further analyses using recombinants with progressive deletions showed that treatment with tBHQ enhanced luciferase activity by no more than 0.5-fold. Similar results were obtained with sulindac (data not shown). Given the severalfold induction of steady-state ␥-GCSh mRNA by tBHQ in cultured cells ( 1A), we conclude that transcriptional regulation by the proximal region of ␥-GCSh is relatively minimal.
To investigate whether a post-transcriptional regulation mechanism is involved in ␥-GCSh expression, we used actinomycin D at the concentration (5 g/ml) that inhibits RNA polymerase II activities (mRNA synthesis), but not RNA polymerase I activities (ribosomal RNA synthesis), to shut down new RNA synthesis, and we then measured the stability of ␥-GCSh mRNA in the presence or absence of the prooxidant sulindac. The t1 ⁄2 of ␥-GCSh mRNA in HT-29 and HCT-15 colorectal cancer cells treated by actinomycin D alone was 6.3 Ϯ 1.6 and 6.8 Ϯ 1.0 h, respectively (Fig. 2, top). In the presence of sulindac, however, the respective values increased almost 2-fold to 12 Ϯ 1.4 and 13 Ϯ 1.3 h, respectively. These results suggested that post-transcriptional regulation is involved in the up-regulation of ␥-GCSh expression induced by sulindac.
To investigate the effects of protein synthesis inhibitors on ␥-GCSh mRNA expression, we treated HT-29 and HCT-15 cells with 10 M cycloheximide with or without sulindac. In both cell lines, cycloheximide alone reduced ␥-GCSh mRNA levels over time, with t1 ⁄2 values of 6.5 Ϯ 1.2 and 8.4 Ϯ 0.7 h, respectively (Fig. 2, bottom). Simultaneous treatment with cycloheximide and sulindac initially decreased ␥-GCSh mRNA expression, but at 6 h and thereafter, the levels increased such that the t1 ⁄2 was Ͼ20 h. These results suggested a delayed mechanism of enhanced ␥-GCSh steady-state mRNA expression by sulindac.
To rule out the possibility of nonspecific inhibitor effects and to investigate post-transcriptional involvement in the regulation of ␥-GCSh mRNA expression, we constructed a tetracycline-regulated ␥-GCSh recombinant plasmid (pTRE-␥-GCSh(2983)) that produces a transcript of 2543 nt (83 nt in a 5Ј-UTR, 1914 nt in a coding region, and 588 nt in a 3Ј-UTR) (Fig. 3A). In the presence of tetracycline, binding of tTA (encoded by the cotransfected EC1214A plasmid) to the promoter is blocked, and transcription of exogenously transfected ␥-GCSh is inhibited (38,39). We transfected pTRE-␥-GCSh(2983) and EC1214 plasmid DNA into HEK293T cells. After 24 h, tetracycline alone or with sulindac was added into the culture medium. Cellular RNA was prepared at different time intervals thereafter. ␥-GCSh transcripts from the transfected DNA were determined by RNase protection assay using a probe that detected transcript from the transfected plasmid DNA but not from the endogenous counterpart. The t1 ⁄2 of the transcript was 7.8 Ϯ 0.7 and 8.4 Ϯ 1.0 h, respectively (Fig. 3, B  and C). However, when sulindac was added 12 h before the addition of tetracycline, the t1 ⁄2 increased to Ͼ20 h (Fig. 3, B and  C). We excluded the possibility that the concentration of tetracycline used here affected endogenous ␥-GCSh mRNA stability (data not shown). These results confirmed that sulindac en-hances ␥-GCSh mRNA stability through a delayed mechanism.
3Ј-UTR Sequence of ␥-GCSh mRNA Is Important for Prooxidant-induced Delayed Stabilization-Sequences that control mRNA degradation can be located at the 5Ј-UTR, coding region, and 3Ј-UTR. We investigated whether the 3Ј-UTR of ␥-GCSh mRNA contains an unstable sequence that is a target of prooxidants. To this end, we constructed a 3Ј-UTR deletion mutant, pTRE-␥-GCSh(2431), that lacked nt 2431-2983 (Fig. 4). This recombinant plasmid was transfected into HEK293T cells, and the stability of the transcript was similarly measured. The t1 ⁄2 was Ͼ20 h with or without sulindac (Fig. 4C). These results strongly suggested that the 3Ј-UTR from nt 2431 to 2983 contains unstable sequence(s).
HuR Is Responsible for Sulindac-induced mRNA Stabilization-Several trans-acting factors have been identified that control mRNA stabilization and destabilization The most no-table of these is the HuR factor, which recognizes AU-rich sequences (30, 40 -42). By examining the sequence from nt 2781 to 2983 in the 3Ј-UTR of ␥-GCSh, we found a putative HuR-binding AU-rich sequence, ϩ2785 AUUUA (site II). To investigate whether this sequence is involved in ␥-GCSh mRNA destabilization and is also the target for the sulindac-induced stabilization, we used site-directed mutagenesis in pTRE-␥-GCSh(2983) to replace this sequence with ϩ2785 AGGGA. For comparison, we also replaced a sequence outside that region, site I ( ϩ2595 AUUUUA) with ϩ2595 AGGGGA, or replaced both wild-type sites with mutant sequences (Fig. 5A). We then transfected the respective plasmids under tetracycline-regulated conditions and analyzed the stabilities of the transcripts. Mutation of site I did not enhance stabilization of the ␥-GCSh transcript, whereas mutations of site II or both sites I and II did (Fig. 5, B and C, untreated), suggesting that the site II sequence is the mRNA-degrading element. More importantly, site II was also the target of sulindac-induced ␥-GCSh mRNA stability, because only WT site II was responsive to sulindac induction (Fig. 5, B and C, sulindac). Site II contains the HuR-responsive sequence because in the cotransfection experiment with HuR expression vector, only the wild-type and site I sequence conferred ␥-GCSh mRNA stability (Fig. 5, B and C,  HuR). Taken together, these results strongly suggested that were cotransfected with EC1214A (2.25 g) into HEK293T cells, which were then treated with tetracycline alone (untreated) or with sulindac and, 12 h later, tetracycline (sulindac). Total RNA was prepared, and the stability of ␥-GCSh mRNA was measured by RNase protection assay. The expression level of 18 S RNA was also measured as a control for sample loading. C, graphical representation of the experiments shown in B. Each experiment was performed at least twice. ␥-GCSh/18 S ratios were plotted as percentage of the maximum value at the time of tetracycline addition. sulindac-induced ␥-GCSh mRNA stability is mediated by the HuR-interacting ϩ2785 AUUUA (site II) sequence.
Other Prooxidants Induce ␥-GCSh mRNA Stability-To investigate whether prooxidants other than sulindac can induce stabilization of ␥-GCSh mRNA, we used the tetracycline-regulated system to determine the t1 ⁄2 of pTRE-␥-GCSh transcripts in transfected HEK293T cells treated with 2-AAF, menadione, or t-BHQ, following the procedure as described in Fig. 3. All four prooxidants, like sulindac, induced stabilization of ␥-GCSh transcripts in delayed mechanism (Fig. 6). Because these agents can induce redox imbalance, our results are consistent with the idea of redox regulation of ␥-GCSh mRNA stability.
p38 MAPK Pathway Is Involved in Induction of ␥-GCSh mRNA Stability-To investigate the signal transduction mechanism by which redox conditions influence ␥-GCSh mRNA stability, we used an activator and several inhibitors of various signal transduction pathways (Fig. 7A). We envisioned that inhibitors of the involved pathways would suppress the induction of ␥-GCSh mRNA stability. To this end, HT-29 colorectal cancer cells were treated with a prooxidant (sulindac, PDTC, or tBHQ) and/or the activator or inhibitors. Expression levels of ␥-GCSh mRNA were determined by the RNase protection assay and quantified by densitometry, using the signal from 18 S RNA as a control for sample loading. The expression of endog-enous ␥-GCSh mRNA was increased by treatment with prooxidants alone (Fig. 7B). This increase was not suppressed by the addition of genistein, H7, or staurosporine, which inhibit protein kinases A and C, and protein kinase A, C, and G, and calmodulin kinase pathways, respectively. The induction of ␥-GCSh expression was moderately suppressed by sodium azide, an activator of AMP-activated protein kinase, and by PD98059, an inhibitor of MAPK/ERK kinase signals. SB203580, a specific inhibitor of the p38 MAPK pathway, was the most potent inhibitor overall. None of the inhibitors by themselves, at the concentrations used, enhanced the expression of ␥-GCSh. These results suggest that p38 MAPK is an important signaling pathway involved in the induction of ␥-GCSh expression by prooxidants, although the results also suggested other signaling mechanisms may be involved to a lesser degree.
To investigate whether the p38 MAPK pathway regulates ␥-GCSh mRNA stability, we again used the tetracycline-regulated system in the transfection assay. Fig. 8 shows that sulindac-induced mRNA stability was inhibited by cotreatment with SB203580 but not with PD98059. N-Acetylcysteine, an inhibitor of oxidative stress, was also a potent inhibitor of sulindacinduced ␥-GCSh mRNA stability. These results, together with those described above, strongly suggest that prooxidant-induced ␥-GCSh mRNA stability is regulated by p38 MAPK sig- FIG. 5. HuR binding controls ␥-GCSh mRNA stability. A, wild-type and three recombinant pTRE-␥-GCSh(2983) constructs. Name of construct indicates which site was mutated. B, the plasmids were cotransfected with EC1214A or empty vector (1.5 g each) into HEK293T cells, which were then treated with tetracycline alone (untreated) or with sulindac 12 h before tetracycline (sulindac). In addition, HuR plasmid (1.5 g) was cotransfected with the respective vectors (1.5 g) and EC1214A (1.5 g) into HEK 293T cells, which were then treated with tetracycline alone (HuR). Total RNA was prepared, and the stability of ␥-GCSh mRNA expression was measured by the RNase protection assay. C, graphical representation of the data shown in B. Each data point represents mean Ϯ S.D. of three independent experiments. ␥-GCSh/18 S ratios were plotted as percentage of the maximum value at the time of tetracycline addition. Because of differences in abundance, signals for ␥-GCSh and 18 S were independently adjusted to be in the linear ranges by appropriate exposure times.
Prooxidants Increase Cytoplasmic HuR and Binding to ␥-GCSh mRNA-To strengthen the involvement of HuR in prooxidant-induced, p38 MAPK-mediated stabilization of ␥-GCS mRNA, we treated HEK293T cells with sulindac or tBHQ in the presence of SB203580 or PD98059. Cytoplasmic and nuclear fractions were prepared from the treated cells and subjected to Western blotting using nuclear (lamin B) and cytoplasmic (␣-tubulin) markers. The amounts of cytoplasmic and nuclear HuR were determined by Western blotting. No detectable cytoplasmic contamination of the nuclear fraction was evident (Fig. 9). Treatments with sulindac or tBHQ increased cytoplasmic content of HuR in the absence of any inhibitors. This increase was suppressed by SB203580 but not by PD98059. These results demonstrated that the prooxidantinduced increase in cytoplasmic HuR is specifically mediated by p38 MAPK signaling.
To demonstrate that the increase in HuR was accompanied by an increase in its binding to the ␥-GCSh 3Ј-UTR sequence, we performed gel mobility shift assays by incubating various cytoplasmic extracts with 32 P-labeled probes containing site I, site II, and their mutated sequences. Fig. 10A shows that sulindac-treated extract could only support mobility shift of site II WT probe but not with probes containing site I and their mutated sequences. This mobility shift was not seen in the extracts from SB203580-treated cells. Extract from PD98059treated cells did not diminish the mobility shift. These results demonstrated that the induced binding activity to site II sequence involved p38 MAPK signaling. Binding of site II probe could be competed efficiently by its cognate ( 2785 AUUUA) sequence but not by the mutant ( 2785 AGGGA) sequence within the same concentration range, demonstrating the sequence specificity of protein binding (Fig. 10B). Moreover, depleting the cell extract by using an anti-HuR antibody diminished the binding (Fig. 10C), demonstrating the involvement of HuR in the binding to site II sequence. The in vivo interaction between ␥-GCSh mRNA and HuR protein was further confirmed by using UV cross-linking. Treatment of sulindac increased the amount of ␥-GCSh mRNA bound to HuR (Fig. 11, compare lanes 3 and lane 1), and cotreatment with SB203580 (lane 7), but not PD98059 (lane 5), reduced the amount of bound ␥-GCSh mRNA to the control level (lane 1). Only marginal amounts of ␥-GCSh mRNA were detected when UV was not used (Fig. 11, lanes 2, 4, 6, and 8).
Taken together, these results demonstrated that the prooxidant-induced increase in the cytoplasmic distribution of HuR is mediated by p38 MAPK signaling and that HuR interacts specifically with the site II sequence of ␥-GCSh mRNA.
p38 Signal Transduction Pathway in Prooxidant-induced ␥-GCSh mRNA Stability-The p38 MAPK signaling pathway is composed of three sequentially activated kinase families as follows: MAPK, MAPK kinase (MKK), and MKK kinase families. MAPK phosphorylates substrates upon activation through phosphorylation by MKKs, which are themselves activated by MKK kinases (43). To demonstrate further the involvement of the p38 pathway in prooxidant-induced ␥-GCSh mRNA stability, we investigated whether sulindac, PDTC, and tBHQ can activate p38 and MKK activities. Treating HT-29 cells with these prooxidants indeed activated p38, MKK3/6, and MAPK activating protein kinase 2 (MAPKAPK2), as evidenced by Western blotting (Fig. 12). Activation of p38 MAPK was suppressed by SB203580 but not by PD98059, consistent with the results shown above that SB203580 but not PD98059 suppressed sulindac-induced ␥-GCSh mRNA stability. Activation of MAPKAPK2 by these prooxidants was suppressed by SB203580 but not by PD98059, consistent with the results that MAPKAPK2 is a downstream effector of p38 signaling (43). SB203580 did not suppress MKK3/6 activation because this kinase is upstream of p38 MAPK.
The results presented thus far are consistent with the following signal transduction mechanisms for the induction of ␥-GCSh mRNA stability: prooxidants 3 MKK kinase (ASK1) 3 MKK (MKK3/6) 3 p38 MAPK 3 MAPKAPK2 3 HuR 3 ␥-GCSh mRNA stabilization. To provide support for this proposition, we cotransfected constitutively active and dominant-negative recombinants encoding various kinases with the tetracycline-regulated ␥-GCSh mRNA stability assay system. As shown in Fig. 13, cotransfection with plasmid DNA encoding constitutively active ASK1, MKK3, p38 MAPK, and MAPKAPK2 stabilized ␥-GCSh mRNA (Fig. 13A), whereas cotransfection with plasmid DNA encoding dominant-negative forms of these kinases suppressed the induced stabilization of ␥-GCSh mRNA (Fig. 13B). Similarly, cotransfection of HuR expression plasmid induced ␥-GCSh mRNA stability, whereas cotransfection of antisense HuR mutant suppressed the induction (Fig. 13B, bottom). Thus, we conclude that induction of ␥-GCSh mRNA stability by prooxidants is controlled by the p38 MAPK pathway through HuR, which interacts with an AU-rich sequence located at the 3Ј-UTR of the mRNA.

Post-transcriptional Regulation of ␥-GCSh Expression by
Prooxidants-Oxidative stress is known to regulate the expression of a number of genes (reviewed in Ref. 44). Recent studies demonstrated that changes in redox conditions can transcriptionally activate many cellular genes without effects on mRNA stability, including those encoding angiotensin II receptor (45), mitochondrial transporter (UCP2) (46), insulin-like growth factor binding protein-1 (47), and Rac1 GTPase (48). Moreover, redox-regulated gene expression can be controlled at the posttranscriptional levels by modulating mRNA stability. Oxidative stress induced by glucose deprivation in cultured cells increases vascular endothelial growth factor mRNA stability (49) but reduces insulin growth factor-1 mRNA stability (50). Increased stability of extracellular superoxide dismutase mRNA, but not manganese superoxide dismutase mRNA, is associated with changes of cellular redox conditions by the treatment of 17␤-estradiol (51). These observations, collectively, suggest multiple effects on mRNA stability by oxidative stress, perhaps depending upon the context of genes, cell types, and redox-modulating agents. However, mechanisms that regulate mRNA stability within these contexts are not well understood.
␥-GCSh mRNA is an important system for studying redoxregulated mRNA stability. It encodes the rate-limiting enzyme for the de novo biosynthesis of GSH. The increased GSH levels preserve cytoprotective functions under oxidative stress (52). Thus, ␥-GCSh, like glutathione S-transferase ␣2 and NADPH quinine oxidoreductase 1, can be considered as a phase II enzyme in the detoxification system of drug metabolism (33,34). Previous studies on the regulation of genes encoding phase II enzymes have been mostly focused on transcriptional mechanisms (53,54). Post-transcriptional regulation of the phase II genes has not been well studied.
The present demonstration that increased ␥-GCSh mRNA stability under oxidative stress conditions underscores the importance of the post-transcriptional mechanism in the regulation of this phase II gene expression. Actinomycin D inhibitor experiments demonstrated that sulindac treatment resulted in an ϳ2-fold increase in the t1 ⁄2 values of ␥-GCSh mRNA stability. Results from transient transfection assays using the tetracycline-regulated system are consistent with these results. More importantly, we have demonstrated that this post-transcriptional regulation was activated by many oxidative stress-inducing agents, including sulindac, PDTC, tBHQ, 2-AAF, and menadione; and some of these agents are not known for transcriptional activation of the phase II enzymes-encoded genes. Thus, the post-transcriptional regulation mechanism described here is not spurious and represents an important mechanism of ␥-GCSh gene regulation under stress conditions.
One of the important findings presented in this work is the identification of HuR as a target of prooxidant-induced ␥-GCSh mRNA stability. We first identified an AU-rich sequence ( 2785 AUUUA) located in the 3Ј-UTR of ␥-GCSh mRNA that is responsible for the instability of ␥-GCSh mRNA. Many AU-rich binding proteins have been identified, but HuR functions as an mRNA-stabilizing factor (35, 40 -42). Indeed, overexpression of HuR by transfection stabilizes ␥-GCSh mRNA degradation. HuR is a ubiquitously expressed, predominantly nucleus-located member of the elav (embryonic-lethal abnormal visual in Drosophila melanogaster) family of RNAbinding proteins (56). It has been shown that HuR translocates to the cytoplasm and stabilizes ARE-containing mRNAs (35,57). Consistent with this mechanism, we demonstrated that upon treatment of prooxidants, the cytoplasmic contents of HuR are increased. This is associated with the enhanced binding of HuR to ␥-GCSh mRNA as demonstrated by the in vivo cross-linking procedure. Another AU-rich RNA binding factor hnRNP D has also been reported to up-regulate mRNA stability in a cell typespecific manner (55). Whether hnRNP D plays a role in prooxidant-induced ␥-GCSh mRNA stabilization remains to be investigated.
Redox Regulation of p38 MAPK Kinase Signaling in the Post-transcriptional Regulation of ␥-GCSh mRNA Stability-Another important finding from the present study is the iden-tification that the p38 MAPK pathway is involved in the oxidative stress-induced ␥-GCSh mRNA stability. This was first demonstrated by using a panel of inhibitors to various signal transduction pathways, including SB203580 at concentration (20 M) that has been demonstrated to be highly selective to the p38 MAPK signaling (57)(58)(59). This initial observation led to the formulation of prooxidant-induced ␥-GCSh mRNA stabilization mechanism as depicted in Fig. 14. Fig. 14 shows that activation of p38 MAPK initiates its downstream signal, MAPKAPK2 (39,57,61), and we demonstrated that cotransfection of expression vector for MAP-KAPK2 enhanced ␥-GCSh mRNA stabilization, whereas dominant-negative recombinant suppressed prooxidant-induced stabilization. Previous studies have shown that overexpression of MAPKAPK2 increases cytoplasmic HuR accumulation, which is associated with dramatic changes in the formation of HuR and ARE complexes (42). How MAKPAPK2 regulates cytoplasmic HuR accumulation remains to be elucidated, al- FIG. 10. Gel mobility shift analysis of HuR binding to ␥-GCSh 3-UTR sequences. A, cytoplasmic proteins (10 g) prepared from sulindac-treated HEK293 T cells in the presence or absence of SB20358 or PD98059 were incubated with 32 P-labeled probes (100 fmol) containing wild-type (WT), site I, site II, and their mutated (MT) sequences. Mobility shift was determined by gel electrophoresis. B, analysis of protein binding specificity by competition assay. Cytoplasmic extracts from sulindac-treated cells were incubated with wild-type site II sequence probe in the presence of increased amounts of unlabeled oligoribonucleotides containing wild-type ( 2785 AUUUA) or mutated ( 2785 AGGGA) sequence followed by mobility shift assay. C, cytoplasmic extracts (10 g) prepared from sulindac-treated HEK293 T cells were incubated with anti-HuR antibody (200 ng) followed by 5 l of protein A/G plus-agarose to pull-down HuR. The HuR-depleted supernatants were incubated with radiolabeled wild-type site II RNA oligonucleotide and assayed by gel mobility shift assay. As a control, the cytoplasmic fraction was incubated with protein A/G plus-agarose without anti-HuR antibody.
FIG. 11. In vivo interaction between ␥-GCSh mRNA and HuR. Sulindac-treated HEK 293T cells, in the presence or absence of SB203580 or PD98059, were irradiated with 254-nm UV light to crosslink RNA and protein in vivo (lanes 1, 3, 5, and 7) or without crosslinking (lanes 2, 4, 6, and 8). Cell extracts were prepared and divided into 2 aliquots; one was used for immunoprecipitation of HuR-bound ␥-GCSh mRNA using anti-HuR antibody, and the other was used for Western blotting of HuR as reference of sample loading. ␥-GCSh mRNA levels were measured by RNase protection assay. though the Hsp20 (39) and hnRNP A0 (62) have been implicated as targets of MAPKAPK2.
It has been demonstrated that under normal physiologic conditions, the reduced form of thioredoxin is complexed with ASK1 (also known as MKK kinase 5) that inhibits ASK1 activity (63). Upon oxidative stress, thioredoxin is oxidized and dissociated from ASK1, which subsequently activates downstream signaling, by sequential phosphorylation of MKK3/6 and p38 MAPK (64). This mechanism also explains the association between cellular oxidative stress and the MAPK pathway activation as shown in Fig. 14.
This signal transduction pathway emphasizes the sequential events for redox-induced mRNA stabilization. The involvement of many steps of activation in the process may explain the delayed mechanism of induction as described here. The increased steady-state ␥-GCSh mRNA expression induced by various prooxidants is a late event; stabilization of ␥-GCSh transcripts can only be observed when the prooxidants are added before the switch to the tetracycline-off system; and induction was delayed in the time course analysis of ␥-GCSh mRNA expression in cells treated with cycloheximide and sulindac. However, this delayed effect was not seen when actinomycin D was used. This discrepancy might be attributable to unknown side effects of actinomycin D.
We demonstrated previously that expression of ␥-GCSh is itself regulated by redox conditions (16). The pathway de-scribed in Fig. 14 can also explain the redox-induced feedback inhibition of ␥-GCSh biosynthesis. Stabilization of ␥-GCSh mRNA may increase the biosynthesis of GSH, resulting in a reduced intracellular redox status. This in turn may suppress the oxidation of thioredoxin and its downstream p38 MAPK kinase pathway, thereby suppressing the induction of ␥-GCSh mRNA stability.
Post-transcriptional Regulation of ␥-GCSh Expression in Cancer-We observed previously that ␥-GCSh mRNA expression is correlated with colorectal tumor progression, being low in adenomas and high in carcinomas (15,17). The posttranscriptional regulation mechanism is likely to involve the increased ␥-GCSh mRNA levels in colon cancer from the following considerations. First, it has been suggested that increased oxidative stress is associated with colorectal carcinogenesis (18). Second, by using immunohistochemical staining, it has been shown that the cytoplasmic abundance of HuR increases with malignancy, particularly in colorectal carcinomas (65). Moreover, the development of colon cancer in experimental animals correlates with HuR content. Colon cancer cells overexpressing HuR produced significantly larger tumors than those arising from control cells, whereas those with reduced HuR levels through small interference RNA-or antisense HuR-based approaches developed significantly more slowly (65). Third, constitutively elevated p38 MAPK expression is frequently associated with cancer progression (57). These findings, together with those described in this study, suggest that the pathway elicited in Fig. 14 may be the underlying mechanism for the enhanced expression of ␥-GCSh mRNA in colon carcinomas. We have observed previously that expression of MRP1, an ATP-dependent efflux pump for eliminating diverse anticancer drugs, is tightly associated with that of ␥-GCSh (13-16), including colon cancers (15,17). It is plausible that the post-transcriptional regulation mechanism described here may be involved in the regulation of this important drug resistance transporter.
Taken together, our results have important implications in the redox regulation of colorectal carcinogenesis and the evolution of drug resistance during disease development. Further investigations are required to establish the clinical relevance of these observations. These studies may eventually lead to the development of intervention strategies for the management of colon cancer.