Oxidants Positively or Negatively Regulate Nuclear Factor κB in a Context-dependent Manner*

Redox-based mechanisms play critical roles in the regulation of multiple cellular functions. NF-κB, a master regulator of inflammation, is an inducible transcription factor generally considered to be redox-sensitive, but the modes of interactions between oxidant stress and NF-κB are incompletely defined. Here, we show that oxidants can either amplify or suppress NF-κB activation in vitro by interfering both with positive and negative signals in the NF-κB pathway. NF-κB activation was evaluated in lung A549 epithelial cells stimulated with tumor necrosis factor α (TNFα), either alone or in combination with various oxidant species, including hydrogen peroxide or peroxynitrite. Exposure to oxidants after TNFα stimulation produced a robust and long lasting hyperactivation of NF-κB by preventing resynthesis of the NF-κB inhibitor IκB, thereby abrogating the major negative feedback loop of NF-κB. This effect was related to continuous activation of inhibitor of κB kinase (IKK), due to persistent IKK phosphorylation consecutive to oxidant-mediated inactivation of protein phosphatase 2A. In contrast, exposure to oxidants before TNFα stimulation impaired IKK phosphorylation and activation, leading to complete prevention of NF-κB activation. Comparable effects were obtained when interleukin-1β was used instead of TNFα as the NF-κB activator. This study demonstrates that the influence of oxidants on NF-κB is entirely context-dependent, and that the final outcome (activation versus inhibition) depends on a balanced inhibition of protein phosphatase 2A and IKK by oxidant species. Our findings provide a new conceptual framework to understand the role of oxidant stress during inflammatory processes.

Oxidant stress is a critical pathophysiological mechanism that stands at the foreground of a number of inflammatory diseases. In such conditions, highly reactive oxygen and nitrogen species exert their biological activity by inflicting various oxidative damages to biomolecules and by modulating the activity of redox-sensitive signal transduction pathways (1). The tran-scription factor nuclear factor B (NF-B) 2 is a master regulator of inflammation and apoptosis, which is considered a prototypical example of such sensitivity to oxidant stress (2). NF-B is a family of dimeric proteins normally retained in the cytoplasm of nonstimulated cells, bound to inhibitory proteins, the IBs (3). The critical step in NF-B activation relies on its dissociation from the IB protein, resulting from stimulus-induced phosphorylation of IB, followed by its polyubiquitination and proteasomal degradation. IB itself is phosphorylated by IB kinase (IKK), composed of a heterodimer of two catalytic subunits, IKK␣/␤, and a regulatory subunit, IKK␥ (4). A considerable variety of stimuli lead to IKK activation and downstream NF-B signaling, comprising inflammatory cytokines, various microbial components, as well as genotoxic, physical, or chemical stress factors (5).
Since the first report by Schreck et al. (6) that NF-B could be activated directly by H 2 O 2 in a subclone of Jurkat cells, a number of experimental studies have lent support to the concept that NF-B activation is mediated via a redox-based mechanism (for review, see Ref. 2). The underlying molecular mechanisms were shown to involve either IKK-dependent IB␣ serine phosphorylation (7) or tyrosine phosphorylation of IB␣ at Tyr 42 (8). Recently however, the sensitivity of NF-B to oxidant stress has been brought into question by a series of investigations showing that oxidant stress activates NF-B only in some, but not all cell types (for review, see Ref. 9). Furthermore, several recent studies, including our own, have indicated that oxidants such as H 2 O 2 (10 -12) or peroxynitrite (PN) (13) repressed NF-B activation by inflammatory cytokines in several cell systems in vitro, through a mechanism involving oxidative modification of the upstream kinase IKK. Similar inhibition of IKK has also been observed with various biochemical agents triggering oxidation of cysteine 179 in IKK␤ (14 -16). Thus, NF-B modulation by oxidants appears largely unpredictable and dependent on the cell type studied and the kind of redox stimulus.
An essential aspect of NF-B signaling is its transient nature, which requires short term activation and tight regulation by negative feedback loops (17). Given the multiplicity of their biological targets, oxidants might be able to modulate both pos-itive and negative signals in the NF-B pathway, and the balance between these contrasted influences might dictate the final outcome (activation or inhibition) on NF-B. To address this hypothesis, lung A549 cells were stimulated with TNF␣ before or after a short exposure to oxidant stress conditions. Preexposure to oxidant stress prevented TNF␣-mediated NF-B activation by impairing IKK phosphorylation, whereas postexposure to oxidant stress markedly increased and prolonged TNF␣-mediated NF-B activation by suppressing phosphatasedependent IKK dephosphorylation. Thus, oxidants can inactivate both kinases and phosphatases in the NF-B signaling pathway and may thereby either suppress or amplify TNF␣dependent NF-B activation. These findings provide novel insights into the mechanisms of NF-B regulation in conditions of oxidant stress.

EXPERIMENTAL PROCEDURES
Cell Culture and Stimulation-Human lung carcinoma epithelial A549 cells and murine fibrosarcoma L929 cells were grown (5% CO 2 , 37°C) in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 0.1 mg/ml streptomycin. Oxidant stress conditions were realized by exposing cells to the potent oxidant PN (1). In some experiments, the reactive oxygen species hydrogen peroxide (H 2 O 2 ) was used instead of PN. PN (Calbiochem) was synthesized as described (18) and used along a previously established protocol (13,19). Briefly, PN was delivered to the cells (in PBS-glucose medium) at a final concentration of 10 -500 M. In the preexposure protocol, cells were exposed to PN (or H 2 O 2 ) for the indicated times, washed in PBS-glucose, and replaced in culture medium for stimulation with 10 ng/ml human TNF␣ (Pierce). In some experiments, human IL-1␤ (PreproTech, London, UK) was used (1 ng/ml) instead of TNF␣. In the postexposure protocol, cells were first stimulated with TNF␣ for the indicated times, washed in PBS-glucose, and then treated with PN (or H 2 O 2 ) for various periods, as indicated. In specified protocols, the protein phosphatase inhibitor okadaic acid (OA; 100 nM) was used in place of the oxidants, as indicated.
To mimic the postexposure protocol described above in a physiologically relevant condition, separate experiments were conducted in L929 cells stimulated with TNF␣. Indeed, these cells are known to produce large amounts of reactive oxygen species upon TNF␣ stimulation (20,21). L929 cells were stimulated with 10 ng/ml TNF␣ for 1 h. The role of endogenously produced oxidants was evaluated by pretreating cells with the antioxidant N-acetylcysteine (25 mM) to prevent the accumulation of reactive oxygen species.
Preparation of Protein Extracts, SDS-PAGE, and Western Immunoblotting-Cytoplasmic and nuclear extracts were obtained as described (13). Thirty g of cytoplasmic or 10 g of nuclear proteins were separated by SDS-PAGE, followed by standard immunoblotting protocol, using the following primary antibodies: anti-NF-B RelA/p65, anti-IB␣ (both from Santa Cruz Biotechnology), anti-IKK␣, anti-IKK␤, anti-phospho-IKK␣/␤, and anti-␣-tubulin (Cell Signaling, Beverly, MA), and followed by incubation with the appropriate horseradish peroxidase-conjugated secondary antibody (Bio-Rad).
NF-B Gene Reporter Assay-Cells were transiently transfected with 1 g of a multimeric NF-B pGL2 luciferase vector and 0.1 g of the Renilla pRL-TK vector. Luciferase activity was determined and normalized as detailed in Ref. 13.
Immunoprecipitation and IKK Assay-IKK was immunoprecipitated, and the kinase reaction was performed by adding 1 g of recombinant glutathione S-transferase-IB␣ and 5 Ci of [␥-32 P]ATP for 30 min at 30°C, according to our previously published procedures (13). Proteins were separated on a 12% polyacrylamide gel, and gels were fixed, dried, and examined by autoradiography.
Quantification of IL-8 production-The concentration of IL-8 was determined in the medium of A549 cells stimulated with TNF␣ for 1 h followed by a 4-h posttreatment with H 2 O 2 , using a commercially available enzyme-linked immunosorbent assay (R&D Systems) according to the manufacturer's protocol.
Protein Phosphatase 2A Assay-PP2A activity was determined with the serine-threonine phosphatase assay system obtained from Promega, using the PP2A-specific reaction buffer. A549 cells in 6-well plates were left untreated or were treated with PN (500 M) or the PP2A inhibitor OA (100 nM) for 30, 60, or 90 min. Cells were then lysed, cleared by centrifugation (100,000 ϫ g, 4°C, 1 h), and supernatants were passed through Sephadex G-25 spin columns to remove free phos- phate. 35 l of the phosphate-free lysate was then incubated on a 96-well plate together with 5 l of the phosphopeptide substrate RRA(pT)VA and 10 l of PP2A-specific buffer for 30 min at 30°C. After addition of a molybdate complex dye, formation of molybdate malachite green-phosphate complex was monitored at 600 nm (pmol of phosphate/g of protein/min), and PP2A activity was expressed as a percentage of control conditions. In some experiments, the reaction was done directly in crude cell extracts obtained from untreated cells, in the presence of PN (500 M) or OA (100 nM).
Statistical Analysis-Densitometric analyses of blots are shown as mean Ϯ S.E. Comparisons were done by analysis of variance and Bonferroni adjustment. A p Ͻ 0.05 was considered significant.

Oxidants and TNF␣ Activate NF-B with Different Time
Courses-Stimulation of A549 cells with TNF␣ triggered the early (10 min) nuclear translocation of NF-B p65 and rapidly enhanced NF-B-DNA binding activity (Fig. 1, A and B). In agreement with the classical mechanism of NF-B activation, these effects were correlated with the early degradation of IB␣ after 10 and 20 min, followed by a secondary increase of the IB␣ signal starting at 30 min (Fig.  1C), consistent with IB␣ resynthesis for negative feedback (13). PN also activated NF-B, albeit with different intensity and time course than TNF␣. Indeed, p65 nuclear translocation and NF-B-DNA binding were weaker and occurred later after PN, starting only after 20 -30 min, and were associated with a delayed degradation of IB␣ (Fig. 1, D-F). In contrast to TNF␣, there was no detectable IB␣ resynthesis up to 90 min after PN.
Given that the signal for IB␣ degradation is triggered by the activity of the upstream kinase IKK, we explored the influence of TNF␣ and PN on IKK activity. As illustrated in Fig. 1G, TNF␣ promoted rapid and robust activation of IKK that was sustained for 20 min followed by return to baseline activity. In contrast, PN induced a much weaker activation of IKK, which did not dissipate over time and remained sustained for 60 min. The initiation of IKK activity being mediated by its phosphorylation at 2 serine residues, we determined whether the different time courses of IKK activation by PN and TNF␣ were associated with differences in IKK phosphorylation. As shown in Fig. 1H, IKK phosphorylation triggered by TNF␣ started after 5 min, peaked after 10 and 20 min, and then progressively returned to baseline levels, consistent with phosphatase-dependent IKK dephosphorylation (22). IKK was also phosphorylated in response to PN after 5-10 min, but, contrary to TNF␣, such phosphorylation failed to be reversed and was maintained over 60 min.
Postexposure to Oxidant Stress Conditions Amplifies TNF␣dependent NF-B Activation-Cells were first stimulated for 30 min with TNF␣ to induce NF-B activation and were subsequently treated with PN for 1 h. As indicated in Fig. 2A, such a sequence led to a significant enhancement of NF-B-DNA binding activity, at concentrations of PN of 250 and 500 M for 1 h. Control experiments using decomposed PN showed no effect. A comparable increase in NF-B-DNA binding activity was observed when H 2 O 2 was used instead of PN (supplemental Fig. S1). Similarly, p65 nuclear translocation was markedly enhanced by 500 M PN for 1 h compared with a posttreatment with the PBS-glucose buffer (Fig. 2B). This was associated with an increased Ser 536 phosphorylation of nuclear p65, an effect that was detected as early as 10 min after the addition of PN in time course experiments (Fig. 2C). Furthermore, the expression of NF-B-dependent genes was also affected by postexposure to oxidants. As indicated in Fig. 2D, NF-B-depen- dent luciferase activity induced by TNF␣ was significantly enhanced by postexposure to 500 M PN (Fig. 2D). In addition, the secretion of IL-8 (Fig. 2E) induced by TNF␣ was significantly greater when cells were exposed to H 2 O 2 (500 M) after TNF␣. Finally, the role of endogenous oxidants on the process of NF-B activation was also evaluated in L929 cells, which generate reactive oxygen species when stimulated with TNF␣. As shown in Fig. 2F, NF-B activation triggered by TNF␣ in L929 cells was largely reduced by the antioxidant N-acetylcysteine, indicating that oxidants produced in response to TNF␣ amplify the process of NF-B activation in these cells.
Postexposure to Oxidant Stress Conditions Prevents IKK Dephosphorylation, Leading to Prolonged Activation of IKK and Sustained IkB␣ Degradation-To explore the mechanisms of oxidant stress-dependent amplification of NF-B activation by TNF␣, we determined the influence of PN on the upstream activation of IKK and degradation of IB␣. Cells were first stimulated with TNF␣ for 10 min and then with PN (500 M) or PBS-glucose buffer for 10 -60 min. As shown in Fig. 3A, IKK phosphorylation triggered by TNF␣ was transient, with rapid dephosphorylation when cells were exposed to PBS-glucose buffer after TNF␣. The concomitant degradation of IB␣ was followed by robust IB␣ resynthesis. In marked contrast, stimulation with PN after TNF␣ prevented IKK dephosphorylation up to 60 min, and this was associated with complete and persistent degradation of IB␣ without any signs of IB␣ resynthesis. The kinase activity of IKK triggered by TNF␣ was also profoundly enhanced by PN, as shown in Fig. 3B. This effect was evident as soon as 5 min after the addition of PN, as indicated by a 20-fold increase of IKK activity (Fig. 3, C and D). PP2A Activity Is Inhibited by Oxidant Stress or OA, Leading to Prolonged IKK Phosphorylation-In a first set of experiments, total protein extracts from untreated cells were incubated with PN or OA for 30 min, followed by measurements of PP2A activity. As shown in Fig. 4A, both PN and OA produced a comparably potent inhibition of PP2A activity. In a second set of experiments, cells were first incubated with PN or OA for 30 -90 min before measurements of PP2A activity. In such conditions, both PN and OA elicited a time-dependent significant reduction of PP2A activity (Fig. 4B). In a third set of experiments, cells were first activated with TNF␣ for 5 min and then with PBS-glucose buffer, PN, or OA for 1 h. As indicated in Fig.  4C, the phosphorylation of IKK after TNF␣ decreased rapidly in cells incubated with PBS-glucose, whereas it was sustained at a high level in cells incubated wit PN or OA.
Preexposure to Oxidant Stress Conditions Prevents NF-B Activation by TNF␣-A549 cells were treated with concentrations of PN from 10 to 500 M for 1 h and then with 10 ng/ml TNF␣ for 30 min. At concentrations of 50 M and above, PN produced a concentration-dependent inhibition of NF-B activation by TNF␣, as evidenced by reduced degradation of IB␣ (Fig. 5A) and NF-B p65 nuclear translocation (Fig. 5B). Similarly, PN dose-dependently inhibited NF-B-DNA binding (Fig. 5C), an effect that was similarly observed when H 2 O 2 was used in place of PN (supplemental Fig. S1). These effects correlated with PN-dependent inhibition of IKK kinase, as indicated in Fig. 5D. IKK activity measured 10 min after TNF␣ was abro-gated following preexposure (1 h) to PN. Furthermore, the induced luciferase expression following TNF␣ was abrogated when cells were preexposed to PN (1 h, 500 M,  Fig. 5E). Thus, transient exposure of A549 cells to oxidant stress conditions prevents subsequent IKK activation and downstream NF-B signaling in response to TNF␣, which agrees with our previous observations made in unrelated cell lines (13).
Regulation of NF-B by Oxidants Is Independent of the Nature of the Immune Stimulus-Cells were activated with IL-1␤ (30 min) instead of TNF␣, and PN or PBS-glucose buffer was added for 1 h, either before or after IL-1␤. As shown in Fig. 6A, NF-B-DNA binding activity was greatly prevented by preexposure to PN, whereas it was reinforced by postexposure to PN. Similar observations were obtained in luciferase reporter experiments (Fig. 6B, upper panel). Preexposure to PN abrogated luciferase activity measured after 4-h stimulation with IL-1␤, whereas postexposure enhanced such activity (Fig. 6B,  lower panel).

DISCUSSION
Defining the mechanisms of NF-B regulation by oxidant stress is of major importance to understand the role of oxidants and free radicals in the pathophysiology of acute and chronic inflammation. Although it has generally been assumed that oxidants are direct NF-B activators (for review, see Ref. 2), recent findings have put this paradigm into question, by showing that oxidants rather inhibit NF-B via oxidative inactivation of IKK␤ (for review, see Ref. 9). In the current study, we hypothesized that such contrasted effects might reflect the ability of oxidants to interfere with both positive and negative signals in the NF-B pathway so that the final outcome (activation versus inhibition) would depend on the balance between these two opposite signals. This issue was addressed by comparing the effects of the oxidants PN or H 2 O 2 , given alone or in various combinations with TNF␣, in an in vitro cellular model.
Oxidant stress on its own elicited a delayed activation of NF-B compared with TNF␣. This effect was associated with a continuous degradation of IB␣, contrasting with the early (within 30 min) resynthesis of IB␣ observed after TNF␣, indicating that this essential negative feedback loop was impaired in conditions of oxidant stress. The critical mechanism preventing continuous IB␣ degradation is the transient nature of IKK activation, related to a well regulated balance between phosphorylation and dephosphorylation of critical serine residues within the activation loop of IKK␣ and IKK␤ (5). Our finding of transient IKK phosphorylation and activation in response to TNF␣ is consistent with this concept. In contrast, oxidants triggered persistent IKK phosphorylation associated with long lasting kinase activity, indicating that proper dephosphorylation and inactivation of IKK were prevented in conditions of oxidant stress.
Several recent lines of evidence support a critical role of the serine-threonine phosphatase PP2A in catalyzing IKK dephosphorylation leading to termination of IKK activity. First, PP2A is constitutively recruited to IKK␤ (17,23), where it appears to control dephosphorylation of randomly activated IKK (23). Second, inactivation of PP2A with OA (22), UVB (17,23), or RNA interference (24) has been associated with persistent IKK activation and NF-B activation. Therefore, our findings of sustained IKK phosphorylation and permanent IB␣ degradation under oxidative conditions pointed to inactivation of PP2A as the most likely underlying mechanism, the more so that PP2A is especially sensitive to inactivation by various oxidant species (25)(26)(27).
This hypothesis was further supported by the striking enhancement of TNF␣-dependent NF-B activation observed when oxidants were applied after TNF␣. In these conditions, IKK phosphorylation and activity persisted over time, and IB␣ resynthesis was completely abrogated. This translated into a significant enhancement of the expression of NF-B-dependent genes, as shown both by increased luciferase activity and IL-8 secretion when oxidants were applied after TNF␣. To evaluate the biological significance of these findings, we determined whether endogenously produced reactive oxygen species would reproduce the effects of exogenously added oxidants. For this purpose, murine fibrosarcoma L929 cells, which produce H 2 O 2 after TNF␣ (20,21), were stimulated with TNF␣ in the presence of the antioxidant N-acetylcysteine. In such conditions, the activation of NF-B by TNF␣ was markedly attenuated, indicating that oxidants produced after TNF␣ in these cells served to amplify the process of NF-B activation. Overall, these findings support the concept that oxidants promote a prolonged activation of NF-B following TNF␣ stimulation via a mechanism involving the prevention of phosphatasedependent IKK dephosphorylation.
We therefore directly evaluated the sensitivity of PP2A to oxidants in our experimental conditions and found that the latter was inhibited either when the oxidant was applied directly on isolated cellular proteins or when the assay was performed in cellular proteins isolated after stimulation with the oxidant. Importantly, comparable effects were obtained when PN was replaced by the PP2A inhibitor okadaic acid: inhibition of PP2A activity with OA promoted persistent IKK phosphorylation when applied after TNF␣.
In addition to the sustained phosphorylation of IKK, we also noticed that TNF␣-induced phosphorylation of Ser 536 on nuclear p65 was enhanced after exposure to oxidants. This posttranslational modification of p65 has been shown to be essential for proper transcriptional activity of NF-B (28). Recent findings have indicated that PP2A physically associates with p65 (29) and that inhibition of PP2A with OA (29) or PP2A RNA interference (24) blocked p65-Ser 536 dephosphorylation. Thus, the marked increase of Ser 536 phosphorylation upon PN treatment is totally consistent with its ability to inhibit PP2A activity.
In striking contrast with the above discussed results, preexposure to oxidant stress markedly inhibited NF-B activation by TNF␣, as indicated by reduced IB␣ phosphorylation and degradation, impaired NF-B DNA binding, and suppressed transcriptional activity, an effect related to the inhibition of TNF␣-dependent activation of the kinase IKK (Fig. 4). We did not explore the mechanism underlying such inhibition, as this result confirmed our previous studies performed in unrelated cell lines (13). Furthermore, these findings are consistent with the oxidant-mediated abrogation of IKK activation and downstream signaling in response to cytokines reported in other cell systems in vitro (10 -12), an effect related to impaired phosphorylation of serines 177/181 in IKK␤, consecutive to redox modifications (S-glutathionylation) of interjacent cysteine 179 (11,30).
Based on our current findings, we therefore propose a new scheme of interactions among oxidant stress, IKK, and NF-B, based on oxidative inhibition of both IKK and PP2A, as depicted in Fig. 7. In such a model, oxidant-mediated inactivation of PP2A promotes a slow accumulation of randomly phosphorylated IKK, resulting in low grade, delayed activation of NF-B. In the presence of an activating stimulus such as TNF␣, the influence of oxidants will depend entirely on the initial phosphorylation status of IKK. If not yet phosphorylated at the time of oxidant addition, oxidized IKK will remain unresponsive to the activating stimulus, and NF-B will not be activated. Alternatively, if already phosphorylated at the time of oxidant addition, IKK will not be dephosphorylated due to impaired PP2A activity, resulting in the amplification and prolongation of NF-B activation. We believe that this model may explain the controversy surrounding the issue of the redox regulation of NF-B.
Because oxidant stress plays key roles in many chronic inflammatory diseases, antioxidant therapies are frequently considered as potentially suitable agents to treat or prevent such diseases, but large scale studies have failed to confirm such theoretical benefits (31). Our observations help explain these disappointing results by showing that the influence of oxidants on NF-B signaling is entirely context-dependent. Understand-ing the complex interactions between oxidant stress and NF-B should be an essential prerequisite to design efficient antiinflammatory therapies and to avoid unexpected side effects.