Glutathione Depletion Down-regulates Tumor Necrosis Factor α-induced NF-κB Activity via IκB Kinase-dependent and -independent Mechanisms*

Reduced glutathione (GSH) plays a crucial role in hepatocyte function, and GSH depletion by diethyl maleate was shown previously to inhibit expression of NF-κB target genes induced by tumor necrosis factor α (TNFα) and sensitize primary cultured mouse hepatocytes to TNF-mediated apoptotic killing. Here we demonstrate in the same system that GSH depletion down-regulates TNF-induced NF-κB transactivation via two mechanisms, depending on the extent of the depletion. With moderate GSH depletion (∼50%), the down-regulation is IκB kinase (IKK)-independent and likely acts on NF-κB transcriptional activity because TNF-induced IKK activation, IκBα phosphorylation and degradation, NF-κB nuclear translocation, NF-κB DNA binding in vitro, and NF-κB subunit RelA(p65) recruitment to κB sites of target gene promoters all appear unaltered. On the other hand, with profound GSH depletion (∼80%), the down-regulation also is IKK-dependent, and a timeline is established linking the inhibition of polyubiquitination of receptor-interacting protein 1 in TNF receptor 1 complex to partial blockage of IKK activation, IκBα phosphorylation and degradation, and NF-κB nuclear translocation. Of note, pretreatment with antioxidant trolox protects against the inhibitory effect of profound GSH depletion on IKK activation and NF-κB nuclear translocation but fails to restore expression of NF-κB target genes, revealing both IKK-dependent and -independent inhibition. These findings provide new insights into the complex effects of oxidative stress and redox perturbations on the NF-κB pathway.

Reduced glutathione (GSH) plays a crucial role in hepatocyte function, and GSH depletion by diethyl maleate was shown previously to inhibit expression of NF-B target genes induced by tumor necrosis factor ␣ (TNF␣) and sensitize primary cultured mouse hepatocytes to TNF-mediated apoptotic killing. Here we demonstrate in the same system that GSH depletion down-regulates TNF-induced NF-B transactivation via two mechanisms, depending on the extent of the depletion. With moderate GSH depletion (ϳ50%), the down-regulation is IB kinase (IKK)-independent and likely acts on NF-B transcriptional activity because TNF-induced IKK activation, IB␣ phosphorylation and degradation, NF-B nuclear translocation, NF-B DNA binding in vitro, and NF-B subunit RelA(p65) recruitment to B sites of target gene promoters all appear unaltered. On the other hand, with profound GSH depletion (ϳ80%), the downregulation also is IKK-dependent, and a timeline is established linking the inhibition of polyubiquitination of receptor-interacting protein 1 in TNF receptor 1 complex to partial blockage of IKK activation, IB␣ phosphorylation and degradation, and NF-B nuclear translocation. Of note, pretreatment with antioxidant trolox protects against the inhibitory effect of profound GSH depletion on IKK activation and NF-B nuclear translocation but fails to restore expression of NF-B target genes, revealing both IKK-dependent and -independent inhibition. These findings provide new insights into the complex effects of oxidative stress and redox perturbations on the NF-B pathway.
TNF␣ 2 plays an important role in liver injury in many animal models and has been implicated in the progression of human alcoholic liver disease and hepatic viral infections (1)(2)(3)(4). TNF␣ exerts diverse biological effects by activation of multiple pathways promoting inflammation and both cell survival and death (5). Like many cell types, normal hepatocytes do not undergo apoptosis in response to TNF␣ because successful activation of TNF-induced NF-B pathway blocks the TNF-induced apoptotic pathway (5,6).
It is reasonable to speculate that in the context of specific liver injury models or disease conditions, hepatocytes become unable to respond normally to the diverse effects of TNF␣ and thereby become sensitized to TNF-induced death. GSH, a cysteine containing tripeptide, as the single most abundant antioxidant for detoxifying enzymes and a determinant of the thioldisulfide state, is crucial to hepatocyte function (19). Previous work in our laboratory demonstrated that GSH depletion by DEM (0.25-0.5 mM) inhibited TNF-induced expression of NF-B-responsive survival genes and sensitized primary cultured mouse hepatocytes to TNF-induced apoptosis (20,21). The present studies were designed to delineate the inhibitory effect of GSH depletion on signaling events leading to inhibition of TNF-induced NF-B transactivation. By reporter gene assay we verified that GSH depletion by DEM down-regulates TNF-induced NF-B activity. By Western blotting and electrophoretic mobility shift assay (EMSA) we confirmed that GSH depletion by DEM at low concentrations (Յ0.2 mM) did not inhibit NF-B nuclear accumulation and DNA binding activity but did so partially at a higher concentration (0.5 mM). Therefore, we hypothesized that IKK-independent and -dependent mechanisms are involved and set out to prove this by evaluating DEM effect on major signaling events leading to IKK-dependent NF-B activation and relating them with the timing and degree of GSH depletion. Furthermore, the role of reactive oxygen species (ROS) was assessed indirectly by evaluating the possible protective effect of antioxidant pretreatment. Finally, nuclear p65 was analyzed for phosphorylation, acetylation, and recruitment to B sites of NF-B-target gene promoters as an initial step to define the IKK-independent mechanism.

EXPERIMENTAL PROCEDURES
Cell Culture, Treatments, Preparation of Protein Extracts, and Cellular Glutathione Determination-Primary cultured mouse hepatocytes (PMH) were used in all experiments. Hepatocytes were isolated from male C57BL/6 mice 7-9 week of age and plated at 1.2 ϫ 10 6 cell/60-mm dish as described previously (22). Three hours after plating, hepatocytes were washed and rested in phenol red-free and serum-free Dulbecco's modified Eagle's medium/F-12 medium (3-ml/dish) overnight (ϳ15 h). The incubation and following treatments were carried out in a 37°C cell culture incubator with 5% CO 2 . For drug cotreatments, working stock was added to the culture just before TNF␣ addition except when specified otherwise. After the indicated time periods, cells were washed with ice-cold Dulbecco's PBS followed by further processing. The resulting preparations were stored in aliquots at Ϫ80°C. For whole cell extracts, cells were lysed in APB buffer (20 mM Tris-HCl, pH 7.6, 1 mM EDTA, 0.5 mM EGTA, 10% glycerol, 20 mM ␤-glycerophosphate, 1.3 mM p-nitrophenol phosphate, 0.5 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, and protease inhibitor mixture (Sigma)) supplemented with 1% Triton-100, 2 mM DTT, and 300 mM NaCl. Nuclear extracts were prepared as described (23). The protein concentration of the extracts was determined by the method of Lowry. Recombinant mouse TNF␣ (20 ng/ml; EMD Biosciences) was used in the set of experiments shown in Figs. 1, A and B, 2, and 3, A-C. To study TNFR1-mediated signaling (24), we used recombinant human TNF␣ (20 ng/ml; R&D Systems) in the rest of experiments. For stock solutions, aliquots of DEM (0.5 M in Me 2 SO; Sigma), trolox (0.5 M in ethanol; EMD Biosciences), buthionine sulfoximine (BSO, 0.1 M in H 2 O, filtered; Sigma), and actinomycin D (ActD, 0.4 mg/ml in ethanol; Sigma) were stored at Ϫ20°C as recommended by the manufacturer's. To determine the cellular glutathione (GSH ϩ 2GSSG) content, cell extracts were prepared with 5% trichloroacetic acid and subjected to the GSH recycling assay (25).
Plasmids, Transient Transfection, and Reporter Gene Assays-Plasmids pNF-B-Luc, pTAL-Luc, and pAP1-Luc were from EMD Biosciences. Plasmids pTBP-Luc and c-Fos-Luc were kind gifts from Dr. D. L. Johnson (University of Southern California). Luciferase reporter plasmids were introduced into PMH by transient transfection using Targefect F-1 (Targeting system) as described previously (26). A Renilla luciferase reporter plasmid (pRL-TK; Promega) was included in all transfection experiments as an internal control for transfection efficiency. Six hours post-transfection PMH were treated with TNF␣ and/or DEM as indicated for 4 h, lysed, and assayed for luciferase activities using the dual-luciferase reporter assay system (Promega).
EMSA-The experiments were performed as described previously (27). NF-B consensus oligonucleotides (B; Promega) were labeled with [␥-32 P]ATP by T 4 polynucleotide kinase (Promega), purified by 8% PAGE, and used as a probe. Nuclear extracts (2 g) were incubated with the labeled probe (40,000 cpm/10 fmol) at 30°C for 20 min in a binding reaction (10 l) containing 10 mM Tris, pH 7.5, 100 mM KCl, 40 mM NaCl, 1 mM EDTA, 1 mM DTT, 10% glycerol, 1 g poly(dI-dC), and 5 g of bovine serum albumin. The protein-DNA complexes were fractionated on a native 4% PAGE with recirculation of TAE buffer (6.7 mM Tris, pH 7.5, 3.3.mM sodium acetate, and 1 mM EDTA). Labeled Oct-1 oligo (Promega) was used for loading control in parallel experiments. For competition or super shift experiments, nuclear extracts were incubated with unlabeled oligonucleotides (B or AP-1; Promega) or antibodies as specified at 4°C for 1 h before the addition of the labeled probe. Antibodies against RelA(p65) (100-4165; Rockland) and p50 (sc114; Santa Cruz Biotechnology) were used in a supershift assay.
Preparation of IKK Complex and Kinase Assay-Whole cell extracts (250 g) were precleared by normal rabbit IgG plus protein A-agarose followed by immunoprecipitation (IP) with IKK␣ antibody (5 l, sc7190; Santa Cruz Biotechnology) and protein A-agarose (20 l). Precipitates of IKK complex were washed three times in APB buffer supplemented with 0.25% Triton X-100, 2 mM DTT, and 150 mM NaCl and once in kinase buffer containing 20 mM Hepes, pH 7.2, 20 mM MgCl 2 , 2 mM DTT, 20 mM ␤-glycerophosphate, 0.1 mM Na 3 VO 4 , and 10 g/ml aprotinin. IKK activities of the precipitates were determined in an in vitro kinase assay (28) in the kinase buffer (20 l) supplemented with 1 g of glutathione S-transferase (GST)-IB␣ and 0.3 mM ATP. After 30 min incubation at 30°C, the reaction was stopped and analyzed by WB for the extent of phosphorylation of IB␣ with an antibody specific for phospho-IB␣ (Ser-32/-36) (#9246; Cell Signaling Technology). To determine whether DEM is a direct inhibitor of IKK activity, DEM (0.5 mM) was added to the washed immunoprecipitates in kinase buffer (10 l) in the absence of DTT. After 10 min of incubation at 30°C, the precipitates were diluted with kinase assay (KA) mixture (30 l), and kinase activities were determined. GST-IB␣ was prepared from an Escherichia coli strain (JM109) containing plasmid pGST-IB␣ using glutathione-Sepharose 4B. This plasmid encodes an amino-terminal GST tagged IB␣ (amino acids 13-54) of mouse origin (GenBank TM accession number C046754).
Preparation of TNFR1 Signaling Complex-Experiments were performed as described (11) with some modification mainly due to the relative large size and small surface area of PMH. In brief, 5 ϫ 10 6 PMH were left alone or stimulated with FLAG-tagged human TNF␣ (0.4 g/ml; Apotech) in the presence or absence of DEM (0.5 mM) for indicated times. In these experiments DEM was added 2 min before TNF␣ addition to ensure a starting point of 50% GSH depletion in all TNF␣/DEM cotreatments. PMH were then washed and lysed in APB buffer (1 ml) supplemented with 0.2% Nonidet P-40, 150 mM NaCl, 5 mM N-ethylmaleimide, and 1 mM DTT. Whole cell extracts (12 mg) were precleared and subjected to IP overnight with monoclonal anti-FLAG antibody (2 g, F3165; Sigma) and protein G-Sepharose (30 l; GE Healthcare). Precipitates of TNFR1 signaling complex were washed and then boiled in 2ϫ sample buffer (pH 8.0, 30 l) before processing for WB with antibodies specific for TNFR1 (AF-425-PB; R&D Systems), RIP1 (610459; BD Biosciences), TRADD, or TRAF2 (sc7868 and sc876; Santa Cruz Biotechnology).
Isolation and Quantitative Analysis of RNA Transcripts-Total RNA was extracted following a previously described protocol (29) from PMH either left untreated or treated with TNF␣ plus or minus DEM. Two microgram of total RNA was reversetranscribed using an Omniscript reverse transcription kit (Qiagen) supplemented with 10 M random primer (Applied Biosystems). The resulting cDNA (equivalent to 6 ng of total RNA) was subjected to quantitative real-time PCR analysis with Taq-Man Universal Master Mix (Applied Biosystems) and ABI Prism 7900HT sequence detection system (Applied Biosystems). Primer pairs and TaqMan probes (Biosearch Technologies, Inc.) used in quantification of gene expressions are summarized in Table 1 except of the set for IB␣, which was reported previously (30). Each set of these oligonucleotides was designed to encompass an exon-exon junction so that genomic DNA would not be amplified or detected. We used the standard curve method provided by the system software (SDS2.1) for the relative quantification following the manufacturer's instruction (Applied Biosystems). The standard used in construction of relative standard curves was composed of two cDNA stocks (50% of each) derived from total RNAs of PMH stimulated with TNF␣ for 30 and 180 min, respectively; as in our preliminary experiments, this mixture showed high level of the transcripts for the genes listed in Table 1. All data were normalized by 18S ribosomal RNA, and the ratios were shown directly to reflect relative abundances of the mRNA examined under different experimental conditions (see Figs. 1D and 9C) or were expressed as percentages of the relevant maximal level induced by TNF␣ (set as 100, see Figs. 5D and 8B).
Chromatin Immunoprecipitation (ChIP) and Quantification-The recruitment of RelA(p65) NF-B subunit to B sites of certain NF-B target gene promoters in intact hepatocytes was assessed by ChIP assays following a previously described protocol (31) with minor modifications. PMH were switched to phenol red-free and serum-free Dulbecco's modified Eagle's medium/F-12 medium for 15 h, then either left untreated or treated with TNF␣, TNF␣/DEM, or DEM for the indicated times. After cross-linking, cells were washed and lysed on ice in APB buffer supplemented with 150 mM NaCl, 0.5% Nonidet P-40, and 0.25% Triton X-100. Nuclei from 8 ϫ 10 6 hepatocytes were collected, lysed in 0.4 ml of nuclear lysis buffer, sonicated,

Sets of primer pairs and TaqMan probe for quantification of transcript levels of NF-B-responsive genes by quantitative real-time PCR
FAM, 6-carboxyfluorescein; BHQ, black hole quencher 1.

Transcript Forward (F) and reverse (R) primers and TaqMan probe (T)
and centrifuged. The supernatant chromatin was precleared, diluted, and divided. Two aliquots, one with 7 l of RelA(p65) antibody (Rockland) and the other without the antibody as negative control, were immunoprecipitated overnight. A third portion, equivalent to only 1% of the IP input, was kept frozen at Ϫ80°C as input chromatin. The immune complex was collected with protein G-Sepharose pre-blocked with salmon sperm DNA, washed, and eluted. Cross-links on immunoprecipitated and input chromatins were then reversed, and DNAs were purified with the QiaQuick PCR purification kit (Qiagen). Immunoprecipitated DNAs (undiluted) and input DNAs (diluted 1:10) were subjected to quantitative real-time PCR using standard curve method as described above. Primer pairs and TaqMan probes (Biosearch Technologies Inc.) used in quantification of gene specific promoter regions are summarized in Table 2. Each set of these oligonucleotides was designed to amplify and detect a specified promoter region encompassing B sites. The standard used in construction of relative standard curves for the quantification was from an input DNA stock derived from the chromatin sample of untreated PMH. All data for recruitment of RelA(p65) to specified promoters were expressed as percentages of their relevant input DNAs. Apoptosis and Necrosis Assays-Apoptosis and necrosis were measured by Hoechst 33258 and Sytox green staining as described previously (20) with Sytox green concentration decreased to 50 nM to accommodate computerized fluorescence image acquiring and analysis with MetaMorph software.

GSH Depletion by DEM Down-regulates TNF-induced NF-B Activity, but Perturbation of Nuclear Accumulation of NF-B Is
Not a Requisite-Previous work in our laboratory demonstrated that GSH depletion by DEM (0.25 and 0.5 mM) inhibits TNF-induced increase of mRNA of endogenous NF-B-responsive genes, such as IB␣, iNOS, and cIAP1, sensitizing PMH to TNF-induced apoptosis (20,21). To explore the underlying mechanism, we proceeded to determine whether B site is sufficient to confer the down-regulation using reporter gene assays. pNF-B-Luc, a plasmid with a luciferase reporter gene under the control of a B site-containing minimal promoter, was introduced into PMH by transient transfection. Transfected PMH were either untreated or treated for 4 h with DEM, TNF␣, or TNF␣ plus DEM before being assayed for luciferase activity.
As expected, stimulation with TNF␣ increased NF-B-reporter activity, whereas GSH depletion by DEM (0.1-0.25 mM) repressed the basal as well as TNF-induced activities (Fig. 1A,  left). Importantly, such an inhibitory effect was not seen with pTAL-Luc, a control plasmid lacking the B sites (Fig. 1A, right). Further experiments revealed that the inhibitory effect was dose-dependent, and a ϳ50% inhibition was observed at a  3). B and C, experiments were carried out as in A with results expressed as -fold increases relative to the untreated control (set as 1). DEM in a lower concentration range (0.025 ϳ 0.1 mM) was used in B to define the lowest effective concentration for the down-regulation. Luciferase reporter plasmids driven by different promoter elements as indicated were utilized in C to evaluate the effect of DEM on the assay. D, total RNAs were prepared from PMH either left untreated or stimulated with TNF␣ (20 ng/ml) plus or minus DEM as indicated for 1, 2, or 4 h corresponding to time of peak induction of each gene and reverse-transcribed. The resulting cDNA was analyzed by quantitative real-time PCR for the expression of IB␣, A20, cIAP1, and iNOS. Cotreatment with ActD (0.5 g/ml) was used as a positive control for inhibition of new RNA formation. All data were normalized to 18S ribosomal RNA, and the ratios are shown for relative mRNA levels. Each bar represents the mean Ϯ S.D. (n ϭ 3).  OCTOBER 1B). As additional controls, we examined the effect of GSH depletion on expression of reporter genes driven by TBP-promoter, c-Fospromoter, or AP-1 site containing promoter (Fig. 1C). In these cases, GSH depletion alone or in combination with TNF␣ either rendered an increase in reporter activity, albeit to different extent, or showed no obvious effect, ruling out a nonspecific effect on the assay. Because GSH depletion by 0.1 mM DEM down-regulated NF-Bdriven reporter gene expression, and the depletion was only moderate (see Fig. 3 and detail in later sections), we proceeded to verify the effect on endogenous gene expression. Total RNAs were isolated from PMH after treatments, reversetranscribed, and analyzed by quantitative real-time PCR for transcript levels of IB␣, A20, cIAP1, and iNOS ( Fig. 1D and data not shown). Indeed, TNF␣ stimulated transcription of these NF-B-responsive genes with the maximal induction level reached at 1 h for IB␣ and A20 and 4 h for cIAP1 and iNOS. GSH depletion by 0.1 mM DEM cotreatment inhibited the observed induction by TNF␣ for all four genes at the time points examined, although less severe inhibition for the initial induction was observed for IB␣ and A20 compared with the late-response genes (cIAP1 and iNOS), indicating the response is likely to be gene-specific. Increasing DEM to 0.5 mM further enhanced the inhibition.

GSH Depletion Represses TNF␣-induced NF-B Activity
Having demonstrated that GSH depletion down-regulates TNF-induced NF-B activity, we next determined whether the down-regulation requires corresponding alteration in TNF-induced NF-B DNA binding activity or NF-B nuclear accumulation. Nuclear extracts were prepared from PMH after 3-h treatments, analyzed by EMSA with 32 P-labeled B-oligonucleotide probes for NF-B DNA binding activity and by WB with specific antibody for NF-B subunit RelA(p65) or p50.
TNF stimulation as expected resulted in an increase in NF-B DNA binding activity as compared with the untreated control ( Fig. 2A), which was B site-specific and composed mainly of p50/p50 homodimer and p50/p65 heterodimer as confirmed by competition and supershift experiments (Fig. 2B, odd lanes). GSH depletion by 0.1 mM DEM did not alter TNFinduced NF-B DNA binding activity ( Fig. 2A). Furthermore, GSH depletion by 0.1 mM DEM exhibited no obvious effect on either the specificity or the dimer pattern of NF-B DNA complexes (Fig. 2). In contrast, GSH depletion by 0.5 mM DEM inhibited TNF-induced NF-B DNA binding activity ( Fig. 2A).  WB analysis (Fig. 2C) showed a good correlation of NF-B nuclear accumulation with the above EMSA results. Nuclear levels of RelA(p65) and p50 were clearly elevated 3-h after the start of TNF␣ stimulation. GSH depletion with DEM at 0.025-0.2 mM exerted no effect on the nuclear accumulation of NF-B, but dose-dependent inhibition occurred with DEM at 0.25-0.5 mM. From this correlation we concluded that the observed decrease in NF-B DNA binding activity at 0.5 mM DEM was largely due to decreased nuclear accumulation of NF-B.
Taken together, our data demonstrated that GSH depletion by DEM (0.1 mM) down-regulated TNF-induced NF-B activity without interfering with NF-B accumulation in the nucleus or NF-B DNA binding in vitro, suggesting an IKK-IB pathwayindependent mechanism. Alternatively, with DEM at 0.5 mM, the down-regulation occurred with concomitant decreases in both NF-B nuclear accumulation and NF-B DNA binding activity in vitro, indicating involvement of an additional mechanism either at the level of, or more upstream of IKK, which may have masked the coexistence of nuclear inhibition of NF-B activity.
GSH Depletion by DEM Is Dosedependent, Rapid, and Persistent during Initial TNF␣ Signaling-At first we examined the relationship between GSH depletion and inhibition of initial TNF␣ signaling, which is known to be very rapid, to see how the rate of GSH depletion and subsequent levels correlate with inhibition of TNF␣ signaling. DEM (32, 33) is a commonly used agent for acute GSH depletion because it diffuses freely into cells due to its hydrophobicity and forms glutathione conjugates in reactions catalyzed by glutathione S-transferase (Fig. 3A). We, therefore, performed time course experiments in PMH to examine the kinetics of acute GSH depletion by DEM in detail. As shown in Fig. 3B, the cellular GSH content was not affected initially by TNF␣ treatment but dropped immediately after cotreatment of DEM; 50% GSH depletion was observed in 2 min with 0.5 mM DEM and 40% depletion in 5 min with 0.1 mM DEM. The time required for maximal depletion depended on the initial DEM concentration; DEM at 0.5 mM led to a profound depletion of cellular GSH (ϳ80%) in 5 min, whereas 0.1 mM led to a moderate depletion (ϳ50%) in 10 min. After the initial phase, cellular GSH levels remained relatively steady in the 45-min time course. Thus, DEM-induced GSH depletion is dose-dependent, and with 0.5 mM DEM, the depletion is very rapid and marked.
Profound GSH Depletion Interferes with TNF-induced Activation of IKK Leading to Decreased Nuclear Translocation of NF-B-Next, we performed a detailed time course study in PMH to examine the possible effect of GSH depletion on TNFinduced activation of IKK-IB pathway. IKK activity was determined by immunocomplex kinase assay, phosphorylation and degradation of cellular IB␣ were examined by WB analysis of whole cell extracts, and nuclear translocation of RelA(p65) and p50 was examined by WB analysis of nuclear extracts.
After TNF␣ stimulation, IKK activity strongly increased at 5 min and reached the maximum at 10 min (Fig. 4A, top panel,  left); consistently, the level of phospho-IB␣ (Ser-32/36) in cells increased sharply at 5 min (third panel, left) followed by a rapid decline of IB␣ between 10 and 30 min (fourth panel left) and a − FIGURE 4. Effect of GSH depletion by DEM on TNF-induced IKK-IB pathway activation. A, WCE were prepared from PMH either untreated or treated over the indicated time course in a 37°C cell culture incubator with TNF␣ (20 ng/ml) or TNF␣ plus DEM (0.1 and 0.5 mM). IKK complexes were immunoprecipitated (IP) with IKK␣ antibody and assayed for KA using GST-IB␣ and cold ATP as substrates. WB was utilized to visualize the reaction product, phospho-IB␣ (Ser-32/36) (p-IB␣) with a specific antibody (top panel). In parallel, 20% of the input WCE was subjected to direct WB to show levels of IKK␣ (loading control), endogenous p-IB␣, and IB␣ (lower panels). B, nuclear extracts were prepared from PMH treated as described in A and subjected to WB to verify changes in nuclear levels of NF-B subunits p65 and p50. ‡, a nonspecific band shown as a loading control. C, first, one set of PMH was pretreated for 1 h with proteasome inhibitor MG-132 (25 M) to block TNF␣-induced IB␣ degradation, whereas the other set was left alone; then PMH from both sets were either unstimulated or stimulated for 15 min with TNF␣ (20 ng/ml) or TNF␣ plus DEM (0.5 mM). WCE were then prepared and analyzed by WB to examine endogenous p-IB␣ level. IB␣ and ␤-actin levels are shown as loading controls. D, IKK complex was immunoprecipitated from WCE prepared from untreated or TNF-stimulated PMH and washed. The active IKK complex was then incubated with DEM (0.5 mM) or N-ethylmaleimide (NEM; 2 mM) for 10 min at 30°C followed by KA assay. IKK␣ levels in precipitates are shown for equal loading. Unlike N-ethylmaleimide, DEM did not inhibit IKK activity directly. Results presented in A-D were confirmed at least in three independent experiments. NUC, nuclear extracts. concurrent increase of nuclear RelA(p65) and p50 (Fig. 4B, left). In comparison, moderate GSH depletion by cotreatment with 0.1 mM DEM exerted no effects on 1) the initial activation of IKK at 5 min, 2) phosphorylation and degradation of IB␣, and 3) nuclear translocation of RelA(p65) and p50 (Fig. 4, A and B, center panels). In contrast, profound GSH depletion by 0.5 mM DEM resulted in a marked reduction in IKK activity as early as 5 min post-TNF␣ stimulation followed by decreases in IB␣ degradation and corresponding nuclear translocation of both RelA(p65) and p50 (Fig. 4, A and B, right panels). Although under the latter condition a corresponding decrease in phospho-IB␣ at early time points was not observed directly (Fig.  4A, third panel), its existence was unmasked after degradation of IB␣ was blocked by pretreatment of PMH with M-132, an inhibitor of 26S proteasome (Fig. 4C). The observed inhibition by 0.5 mM DEM on initial IKK activation is not likely due to a direct effect of DEM on IKK activity, as incubation of immunoprecipitates of active IKK complex with DEM (0.5 mM) did not cause reduction of IKK activity in the subsequent KA assay (Fig.  4D, compare lane 4 and 5 with lane 3). On the other hand, N-ethylmaleimide (2 mM), a commonly used agent for blocking protein thiol groups, strongly inhibited IKK activity (compare lane 6 with lane 3). Thus, we have established that down-regulation of TNF-induced NF-B activity is IKK-independent when GSH is only moderately depleted (ϳ50%) but becomes also IKK-dependent when GSH is severely depleted (ϳ80%).

Antioxidant Protects against the Inhibitory Effect of GSH Depletion on TNF-induced IKK Activation and NF-B Nuclear
Translocation but Does Not Restore NF-B-responsive Gene Expression-Because the high DEM dose (0.5 mM) used to produce profound cellular GSH depletion (ϳ80%) also significantly depletes mitochondrial GSH (Ͼ55%) (21) and severe mitochondrial GSH depletion is associated with increased generation of ROS (34), it is possible that ROS may play a role in the inhibitory effect of profound GSH depletion on TNF-induced IKK activation. We, therefore, examined whether antioxidant pretreatment protects against the inhibitory effect. Trolox (0.5 mM), a potent water-soluble vitamin E analog and a chain-breaking antioxidant (35), was used in the experiments.
As shown in Fig. 5A, GSH depletion by DEM at 0.5 mM inhibited TNF-induced IKK activity; however, with trolox pretreatment, the inhibition was no longer observed, indicating ROS plays a major role in the inhibition process. As additional controls, we examined cellular glutathione levels and found that treatment with trolox neither increased the cellular glutathione level nor prevented the depletion of GSH by DEM (Fig. 5B). Furthermore, the inhibition of TNFinduced NF-B nuclear translocation was prevented by trolox pretreatment (Fig. 5C, lanes 6 and 7), but the protection was not sufficient to prevent inhibition of TNF-induced expression of NF-B target genes, such as A20 and iNOS, at mRNA levels ( Fig. 5D), unmasking the coexistence of nuclear inhibition of NF-B activity. Thus, the further inhibition of TNF-induced expression of IB␣ and A20 by increasing DEM concentration to 0.5 mM (Fig. 1D) is likely the result of IKK-dependent inhibition of NF-B nuclear translocation.
Profound GSH Depletion Interferes with TNF-induced Polyubiquitination of RIP1 in TNFR1 Complex and Subsequent Activating Phosphorylation of IKK-Because the initial IKK activation at 5 min, not the maximal activation at 10 min, was most severely affected by profound GSH depletion (Fig. 4A) and the inhibition could be prevented by antioxidant pretreatments (Fig. 5A), we speculated that the targeted signaling step(s) is likely upstream of IKK. Thus, we examined the effect of profound GSH depletion on TNFR1 complex during initial TNF␣ signaling following a previously published protocol (11). PMH were either unstimulated or stimulated with FLAG-tagged human TNF␣ in the presence or absence of 0.5 mM DEM, and cell extracts were prepared 2 and 5 min post-stimulation. TNF␣-engaged TNFR1 was separated by IP with a FLAG-specific antibody and further analyzed by WB for levels of TNFR1, TRADD, RIP1, and TRAF2 in the complex.

FIGURE 5. Protective effect of antioxidant trolox on GSH depletion-triggered suppression of IKK-NF-B activation.
A, PMH were pretreated with or without trolox (0.5 mM) for 1 h followed by TNF␣ stimulation. TNF treatment was performed in a 37°C water bath for only 5 and 10 min in the presence or absence of DEM (0.5 mM). IKK-KA was then performed as described in Fig. 4A (top panel) with IKK␣ in precipitates shown for equal loading. Rel. activity, IKK activities were normalized by corresponding IKK␣ levels then expressed as percentages relative to the relevant maximal activity induced by TNF␣. B, cellular glutathione levels were examined for PMH exposed to DEM (0.5 mM) for the indicated times with or without trolox pretreatment as in A. Results are presented as percentages of glutathione in the untreated control. Each bar represents the mean Ϯ S.D. (n ϭ 3). C, PMH were treated with or without trolox as described in A before 1 h of TNF␣ stimulation in the presence and absence of DEM (0.5 mM). Nuclear extracts (NUC) were then prepared and subjected to WB to verify nuclear translocation of p65 with histone H3 as a loading control. D, PMH were treated with or without trolox as described in A before 1 and 4 h of TNF␣ stimulation in the presence and absence of DEM (0.5 mM), and total RNAs were prepared. The mRNA levels for A20 at 1 h and iNOS at 4 h were determined by quantitative real-time PCR as described in the Fig. 1D legend and expressed as percentages of the maximal level induced by TNF␣ alone. Each bar represents the mean Ϯ S.D. (n ϭ 3).
As expected, TNF␣ stimulation led to engagement of TNFR1 and the recruitment of TRADD, RIP1, and TRAF2 to the signaling complex as early as 2 min post-stimulation (Fig. 6A). Note that TNF␣ induced an ongoing process of polyubiquitination of RIP1 in TNFR1 complex; polyubiquitinated RIP1 (polyUb-RIP1) appeared at 2 min as a ladder of slow migrating bands (80 -200 kDa) with non-ubiquitinated RIP1 (ϳ75 kDa, pointed by the arrowhead) being the predominant form; by 5 min, the relative amount of polyUb-RIP1 (mainly 150 -250 kDa) predominated over non-ubiquitinated RIP1 (Fig. 6, A,  third panel, and C). These changes in RIP1 were not observed in cell extracts (Fig. 6B, third panel). Unlike RIP1, TNFR1-associated polyubiquitination of TRAF2 was not detectable within 5 min of stimulation, but a TRAF2 immuno-reactive band (ϳ65 kDa) that migrated slightly slower than the unmodified form of TRAF2 (56 kDa) was readily visible (Fig. 6A, fourth panel).
In comparison, profound GSH depletion showed no inhibitory effect at 2 min on either receptor engagement or the recruitment of TRADD, RIP1, and TRAF2 (arrowheads in Fig. 6A), but the level of TNF-induced polyUb-RIP1 was reduced, especially at 5 min (Fig. 6,  A, third panel, and C). Interestingly, the decrease in polyUb-RIP1 was associated with a relatively higher level of non-ubiquitinated RIP1 at 2 min but was no longer so at 5 min, suggesting the involvement of either decreased polyubiquitination of RIP1 or increased deubiquitination of polyUb-RIP1 at both 2 and 5 min and increased degradation of RIP1 at 5 min. The larger TRAF2 form noticed above was also decreased by profound GSH depletion (Fig. 6A,  bottom panel), although the significance is unknown. As additional controls, we examined cell extracts directly by WB analysis and found that with or without profound GSH depletion, TNF␣ stimulation did not affect total protein levels of TNFR1, TRADD, RIP1, and TRAF2 in PMH (Fig. 6B). Thus, the observed decrease in polyUb-RIP1 level in TNFR1 complex was likely due to changes in polyubiquitination or deubiquitination process rather than the recruitment of RIP1.
RIP1 is an essential mediator for TNF-induced IKK-NF-B activation, and recent studies have shown increasing evidence suggesting a crucial role of polyUb-RIP1 in the process. We reasoned that if the role of polyUb-RIP1 in TNF␣ signaling holds true in our model, a decrease in TNFR1 associated polyUb-RIP1 should result in decreased signaling downstream, such as activating phosphorylation of IKK. TNF␣ induces IKK activity by phosphorylation of serines 177 and 181 in the activation loop of IKK␤ (Ser-176/Ser-180 in IKK␣), which can be detected by WB analysis using a specific phospho-IKK antibody. As shown in Fig. 6D, TNF␣ induced a transient activation of IKK by phosphorylation on these sites, which appeared at 5 min, peaked at 10 min, and started to decline at 15 min. This activating phosphorylation was indeed decreased at both 5 and 10 min under the condition of profound GSH depletion, which correlated well with the decreased IKK activity at these time points (Fig. 4A, top panel), whereas total IKK␤ levels remained stable.
Because in TNF␣-TNFR1-mediated signaling both p38 and JNK pathways are downstream of RIP1, we examined their acti-FIGURE 6. Effect of profound GSH depletion on polyubiquitination of RIP1 in TNFR1 complex and activating phosphorylation of IKK, p38, and JNK. A-C, PMH were stimulated with FLAG-tagged TNF␣ (0.4 g/ml) for the indicated times in the presence or absence of DEM (0.5 mM) and collected for cell extracts. Engaged TNFR1 complexes were brought down by immunoprecipitation (IP) with anti-FLAG antibody (␣-FLAG) and subjected to WB with specific antibodies to detect TNFR1, TRADD, RIP1, and TRAF2 as shown in A. IgG (L), IgG light chain of ␣-FLAG detected by anti-TNFR1 antibody. PolyUb-RIP1, polyubiquitinated RIP1. Arrowheads point to non-ubiquitinated forms. In parallel, cell extracts were subjected to direct WB to examine levels of total TNFR1, TRADD, RIP1, TRAF2, and ␤-actin (loading control) as shown in B. Identical experiments were performed three times, and changes of TNFR1-associated RIP1 in both polyubiquitinated (gray bar) and non-ubiquitinated (open bar) forms were summarized in C. All data were expressed relative to the non-ubiquitinated RIP1 recruited at 2 min after TNF␣. D, PMH were treated similarly as in A over a 15-min time course, and then cell extracts were analyzed by WB to evaluate TNF␣-induced activating phosphorylation of IKK (p-IKK) with an antibody specific for phospho-IKK␣/␤ (Ser-176/180). IKK␤ is shown as a loading control. Results were verified in at least three independent experiments. E, PMH were treated as in D. WB for phospho-p38 (Thr-180/Tyr-182) (p-p38), phospho-JNK (Thr-183/Tyr-185) (p-JNK), phospho-c-Jun (Ser-63) (p-c-Jun) and ␤-actin in cell extracts are shown. vation by WB with antibodies specific for phospho-p38 and phospho-JNK (Fig. 6E). In response to TNF␣, phosphorylation of p38 and JNK peaked at 5 and 15 min, respectively. However, peak phosphorylation of both was inhibited with profound GSH depletion. In addition, TNF-induced phosphorylation of c-Jun by JNK was decreased (Fig. 6E, third panel).
So far our data established a time line linking profound GSH depletion by 0.5 mM DEM to the decreased polyubiquitination of RIP1 in TNF␣-TNFR1 signaling complex and further to the downstream events in IKK-IB pathway. In supporting this view, TNF-induced transient activation of p38 and JNK signaling pathways were found to be inhibited by profound GSH depletion, suggesting the targeting step is most likely shared by the three and located upstream.

Moderate GSH Depletion by 0.1 mM DEM Does Not Reduce TNF-induced Nuclear Levels of Phospho-p65 (Ser-536) and
Acetyl-p65 (Lys-310) or Interfere with TNF-induced Recruitment of p65 to Promoters of NF-B Target Genes-Because post-translational modification of p65 affects its transcriptional activities, we next performed time course studies to determine TNF-induced phosphorylation of p65 at Ser-276, Ser-468, and Ser-536 as well as acetylation of p65 at Lys-310 by WB using specific phospho-p65 antibodies and a specific acetyl-p65 (Lys-310) antibody.
We observed that in PMH, p65 was phosphorylated at Ser-536 in response to TNF but not at Ser-276 or Ser-468 ( Fig. 7A  and data not shown). The phosphorylation peaked 5 min after TNF exposure, but its level in nuclei quickly declined afterward (Fig. 7A). Cotreatment with 0.1 mM DEM showed no clear effect on the process (Fig. 7A). The inhibitory effect of cotreatment with 0.5 mM DEM is likely due to decreased nuclear translocation of p65, as both reductions were prevented by trolox pretreatment (Fig. 7A, top and middle  panels).
On the other hand, we observed that nuclear level of acetyl-p65 (Lys-310) was clearly elevated 1 h after TNF stimulation (Fig. 7B, top  We then determined NF-B DNA binding activity in intact cells by ChIP assay. PMH were treated, fixed, and used for nuclei isolation. Chromatins were then released from nuclei, fragmented, and immunoprecipitated with an antibody raised against the carboxyl terminus of p65 protein. The recruitment of p65 to B sites of IB␣, A20, and cIAP1 promoters was analyzed by quantitative real-time PCR. As expected, stimulation with TNF␣ resulted in increased recruitment of p65 to these promoters in a 3-h time course study (Fig. 7C). For the immediate-early response genes IB␣ and A20, the first peak of p65 recruitment between 30 and 60 min correlated well with the sharp increase of their mRNA levels during the same period (Fig. 1D, left two panels), whereas the recruitment for late response gene cIAP1 (Fig. 7C, right panel) occurred much earlier than the significant increase of mRNA (Fig. 1D, top right  panel). Cotreatment with 0.1 mM DEM did not inhibit TNFinduced recruitment of p65 for all three genes (Fig. 7C), suggesting that transcriptional activity of p65, not DNA binding activity of p65, is inhibited by moderate GSH depletion.
Taken together our data indicated that moderate GSH depletion does not disrupt TNF-induced steady state nuclear phospho-p65 (Ser-536) and acetyl-p65 (Lys-310) or TNF-induced binding of p65 to B sites of NF-B target gene promoters. To further delineate the inhibition of NF-B transcriptional activity in our model, detailed analysis is required to elucidate the state of phosphorylation and acetylation of promoter-recruited p65 and its association and dissociation with coactivator and corepressor complexes but is beyond scope of this paper.
Moderate GSH Depletion Sensitizes PMH to TNF-induced Apoptosis-An alternative approach for GSH depletion is pretreatment of PMH with BSO, an inhibitor of GSH synthesis. A moderate GSH depletion (30%) was achieved by BSO pretreat- FIGURE 7. Effect of moderate GSH depletion on TNF-induced p65 phosphorylation, acetylation, and recruitment to promoters of endogenous NF-B-responsive genes. A, PMH were pretreated with or without trolox (0.5 mM) for 1 h followed by incubation with TNF, TNF plus DEM, or DEM alone for the time indicated. Nuclear extracts (NUC) were then prepared and examined by WB for phospho-p65 at serine 536 (p65-pS536), p65, and histone H3 (17 kDa) as a loading control. p65 was phosphorylated after TNF stimulation, but the levels of p65-pSer-536 quickly declined in nuclei. B, nuclear extracts were prepared from PMH treated with TNF in the presence or absence of 0.1 mM DEM for the indicated times. WB detected p65, acetyl-p65 at Lys-310 (p65-Ac-K310, pointed by the arrowhead), and HDA3 (50-kDa) as a loading control. ‡, a band migrates faster than p65 in a 10% gel but not so in a 10.5-14% gel as in A, although its identity is currently unknown. C, PMH were either left untreated or stimulated with TNF (20 ng/ml) plus (F) or minus (E) DEM (0.1 mM) over the indicated time course. ChIP assays were performed using an antibody raised against the carboxyl terminus of p65 protein. Recruitment of p65 to the promoters of IB␣, A20, and cIAP1 genes was detected by real-time quantitative PCR. All data were expressed as the percentage of the relevant DNA input (equivalent to 1% of total IP input). As negative controls, ChIP assays were also performed in the absence of antibody and analyzed similarly; the non-specifically precipitated promoter for each gene analyzed was usually below the detection level (data not shown). ment (33 M) for 16 h without interfering with cell viability. BSO pretreatment showed no effect on TNF-induced IKK activation or phosphorylation and degradation of IB␣ (Fig. 8A) but inhibited TNF-induced expression of NF-B-responsive genes such as IB␣ and A20 at mRNA levels (Fig. 8B). Of note, TNF␣ causes a further decrease of GSH level (50% after a 2-h exposure) in PMH pretreated with BSO (25 M). Thus, we have verified that moderate GSH depletion in PMH, either acutely or slowly, inhibits TNF-induced NF-B activity downstream of IKK-IB pathway.
Profound GSH depletion by 0.5 mM DEM causes necrosis and in the presence of TNF, the death of PMH is a combination of apoptosis and necrosis with sustained JNK activation (20). In contrast, moderate GSH depletion by 0.1 mM DEM caused neither JNK activation (data not shown) nor an increase in necrotic death as compared with the control (0.6 Ϯ 0.6 versus 0.4 Ϯ 0.6% by 24 h). However, it sensitized PMH to TNF-induced apoptosis (Fig. 8C) with preceding cleavage of caspase-8 and caspase-3 (Fig. 8D), as did pretreatment with BSO (25 M; Fig. 8C and data not shown). These are expected, as moderate GSH depletion inhibits TNF-induced NF-B transactivation and, therefore, the expression of its targeting anti-apoptotic and survival genes, such as A20, cIAP1, iNOS, and cFLIP L (Fig. 1D and 9C). In addition, the JNK activation caused by 0.1 mM DEM cotreatment was greater and lasted longer than that by TNF␣ alone (Fig. 9A). Consistently, the cellular level of cFLIP L protein, an endogenous inhibitor of caspase 8, dropped considerably (Fig. 9B), as JNK activation promotes cFLIP L degradation (36).
Notice that the apoptotic death was slow and moderate (for 0.  (Fig. 8D, top panels), but the latter showed strong amplification in the subsequent caspase-3 activation (middle panels). At present the cause of the difference has not been defined, but it is clear that ActD causes more complete inhibition of the NF-B pathway (Figs. 1D and 9C) and stronger sustained JNK activation (Fig. 9A) than seen with 0.1 mM DEM in response to TNF␣.

DISCUSSION
Previous work in our laboratory demonstrated that GSH depletion by DEM inhibits TNF␣-induced transcription of endogenous NF-B-responsive genes, sensitizing PMH to TNF␣-induced apoptosis (20,21). This phenomenon has broad potential implications in relation to the role of TNF␣ in liver diseases associated with GSH depletion or oxidative redox perturbations and represents a unique example of sensitization to the lethal effect of TNF␣. Therefore, to fully understand the mechanism for this effect, we performed the current studies. Using the same model, we demonstrate systematically that GSH depletion down-regulates TNF␣-induced NF-B transactivation via IKK-independent and -dependent mechanisms.
Our data establish that moderately depleting cell GSH (ϳ50%) by 0.1 mM DEM (Fig. 3B) represses expression of NF-B reporter gene in a B site-dependent manner (Fig. 1, A-C) and inhibits expression of endogenous NF-B-responsive anti-apoptotic genes (Figs. 1D and 9C), which contributes to the sensitization of PMH to TNF-induced apoptosis (Fig. 8, C  and D). Similar results were obtained when moderate GSH depletion was achieved via an alternative approach, inhibition of GSH synthesis by BSO pretreatment (Fig. 8, B and C). The down-regulation is independent of the IKK-IB pathway, as it does not affect TNF-induced IKK activation, IB␣ degradation, and NF-B nuclear translocation (Figs. 4 and 8A). Furthermore, it does not interfere with NF-B DNA binding activity in vitro (Fig. 2) or p65-DNA binding in vivo (Fig. 7C), indicating repression of NF-B transcriptional activity.
Transcriptional activity of NF-B can be regulated by posttranslational modifications, such as phosphorylation and acetylation, which in many cases facilitate binding of transcriptional coactivators and enhance the transcription of target genes (9,18). In PMH, we observed TNF-induced nuclear p65 phosphorylation at Ser-536 and acetylation at Lys-310, and the two processes were not reduced by cotreatment with DEM (0.1 mM) (Fig. 7, A and B, and data not shown). Further investigation is necessary to evaluate the state of phosphorylation and acetylation of promoter-recruited p65 and the effect of moderate GSH depletion. Moreover, redox regulation of NF-B on a conserved RXXRXRXXC motif has been reported. For p50, redox regulation involves its Cys-62 and affects primarily its DNA binding affinity (38), whereas for c-Rel, oxidation of Cys-27 prevents its phosphorylation and reduces its transcriptional activity (39) though the affected phosphorylation site, and the underlying mechanism remain unknown. C-Rel is a member of Rel family with transcriptional activity, forms a dimer with p50, p65, or itself, and is present in PMH. Therefore, it will be interesting to know if moderate GSH depletion interferes with any of these events induced by TNF␣.
On the other hand, our data also demonstrate that severely depleting cell GSH (ϳ80%) by 0.5 mM DEM (Fig. 3B) represses expression of NF-B-responsive genes below basal levels (21), at least partially through blocking TNF-induced nuclear translocation of NF-B ( Fig.  2C and 4B) in an IKK-IB pathwaydependent manner (Fig. 4, A and C), although DEM at 0.5 mM does not function as a direct inhibitor of IKK activity (Fig. 4D). Interestingly, the targeting step is likely at the level of polyUb-RIP1 in TNF-engaged TNFR1 signaling complex, as a decrease in polyUb-RIP1 (Fig. 6, A and C) preceded the decrease in activating phosphorylation of IKK (Fig. 6D) and subsequent decrease in IKK activity, IB␣ phosphorylation, and degradation (Fig. 4, A and C).
It is well known that RIP1 is an essential mediator of TNF␣induced IKK-NF-B activation. Oxidants such as H 2 O 2 and diamide (40,41) or agents leading to increased H 2 O 2 production such as L-mimosine (42) have been reported to block the TNF␣-induced IKK/IB pathway. For L-mimosine, it is likely due to decreased recruitment of RIP1 to TNFR1 complex, as levels of RIP1 in both receptor complex and cell are reduced by its pretreatment (42). Similarly, H 2 O 2 pretreatment destabilizes cellular RIP1 and is able to completely block TNF␣-induced IKK activation (40). Unlike the case of L-mimosine, profound GSH depletion in our model system does not affect recruitment and cellular levels of RIP1 during initial TNF␣ signaling, indicating differences between their underlying mechanisms. Considering the critical roles of polyUb-RIP1 in recruitment of IKK complex to TNFR1 signaling complex, the decrease in polyUb-RIP1 in the receptor complex in our studies is likely to contribute to the inhibition of TNF␣ signaling down the IKK-IB pathway. This view is supported by our observation that TNF-induced activation of p38 and JNK pathways was simultaneously inhibited by profound GSH depletion (Fig. 6E), both of which are downstream of RIP1 in TNFR1-mediated TNF␣ signaling. Because TRAF2 is the ubiquitin ligase for TNF␣-induced polyubiquitination of RIP1 and A20 is the deubiquitinase that reverses the reaction, effects of GSH depletion on the activity of TRAF2 and the recruitment and activity of A20 need to be explored in future work.
Additionally, we have provided evidence suggesting that ROS play a major role in the inhibition of initial TNF␣ signaling in our model, as pretreatment with antioxidant trolox afforded protection against the effects of profound GSH depletion by DEM on TNF-induced IKK activation (Fig. 5A) without pre- FIGURE 9. Effect of moderate GSH depletion on JNK activation in TNF-treated PMH. A and B, WCE prepared from PMH treated with TNF, TNF/ActD (0.5 g/ml), or TNF/DEM (0.1 mM) were examined by WB for phospho-JNK (Thr-183/Tyr-185) (p-JNK) and JNK (A) and FLIP L as well as ␤-actin as a loading control (B). C, total RNA was prepared from PMH 1 and 2 h after the same treatments as in A and B and analyzed for the relative level of c-FLIP L mRNA as described in Fig. 1D legend.
serving GSH (Fig. 5B). The source and critical ROS species remain to be explored. Importantly, the protective effect of trolox, which restores NF-B nuclear translocation as well (Fig.  5C), is still not sufficient to prevent the inhibition of expression of NF-B target genes (Fig. 5D), revealing that profound GSH depletion not only interferes with NF-B nuclear translocation but also inhibits nuclear NF-B activity. Thus, GSH depletion can inhibit TNF-induced NF-B transactivation via IKK-dependent and -independent mechanisms; with profound GSH depletion both mechanisms are in operation, whereas for moderate GSH depletion the inhibition occurs independent of IKK pathway.
It is of interest that a major distinction between high and low concentration of DEM is depletion of mitochondrial GSH (21), suggesting that this may be the source of oxidative inhibition of IKK activation. Because mild GSH depletion, which spares mitochondrial GSH (21), is not expected to promote oxidative stress, our findings suggest thiol-disulfide redox balance itself influences the assembly and activity of the NF-B transcriptional apparatus. Even if moderate GSH depletion is unaccompanied by increased oxidized glutathione (GSSG), the Nernst equation predicts that the redox potential is related to [GSH] 2 / [GSSG]; therefore, decreased GSH without a change in GSH/ GSSG can exert significant effects on redox susceptible thiols on proteins.
Finally, the strongly sustained JNK activation seen in ActD cotreatment was not observed in PMH treated with TNF␣ plus 0.1 mM DEM (Fig. 9A). This is likely due to incomplete inhibition of NF-B activity by 0.1 mM DEM as compared with ActD (Figs. 1D and 9C) and probably accounts for the less striking sensitization to TNF-induced apoptosis compared with ActD.
In summary, we demonstrate in PMH that GSH depletion down-regulates TNF-induced NF-B transactivation via IKKdependent and -independent mechanisms. Although more work is required to assess the source and species of ROS and mechanism of decreased polyUb-RIP1 with profound GSH depletion and mechanism of inhibition of transcription by moderate GSH depletion, the findings underscore the potential for oxidative stress and redox perturbations to impair cell survival pathways in response to TNF␣.