Interleukin-1β Induction of NFκB Is Partially Regulated by H2O2-mediated Activation of NFκB-inducing Kinase*

Reactive oxygen species (ROS) have been demonstrated to act as second messengers in a number of signal transduction pathways, including NFκB. However, the mechanism(s) by which ROS regulate NFκB remain unclear and controversial. In the present report, we describe a mechanism whereby interleukin-1β (IL-1β) stimulation of NFκB is partially regulated by H2O2-mediated activation of NIK and subsequent NIK-mediated phosphorylation of IKKα. IL-1β induced H2O2 production in MCF-7 cells and clearance of this ROS through the expression of GPx-1 reduced NFκB transcriptional activation by inhibiting NIK-mediated phosphorylation of IKKα. Although IKKα and IKKβ were both involved in IL-1β-mediated activation of NFκB, only the IKKα-dependent component was modulated by changes in H2O2 levels. Interestingly, in vitro reconstitution experiments demonstrated that NIK was activated by a very narrow range of H2O2 (1–10 μm), whereas higher concentrations (100 μm to 1 mm) inhibited NIK activity. Treatment of cells with the general Ser/Thr phosphatase inhibitor (okadaic acid) lead to activation of NFκB and enhanced NIK activity as a IKKα kinase, suggesting that ROS may directly regulate NIK through the inhibition of phosphatases. Recruitment of NIK to TRAF6 following IL-1β stimulation was inhibited by H2O2 clearance and Rac1 siRNA, suggesting that Rac-dependent NADPH oxidase may be a source of ROS required for NIK activation. In summary, our studies have demonstrated that redox regulation of NIK by H2O2 is mechanistically important in IL-1β induction of NFκB activation.

Reactive oxygen species (ROS) have been demonstrated to act as second messengers in a number of signal transduction pathways, including NFB. However, the mechanism(s) by which ROS regulate NFB remain unclear and controversial. In the present report, we describe a mechanism whereby interleukin-1␤ (IL-1␤) stimulation of NFB is partially regulated by H 2 O 2 -mediated activation of NIK and subsequent NIK-mediated phosphorylation of IKK␣.

IL-1␤ induced H 2 O 2 production in MCF-7 cells and clearance of this ROS through the expression of GPx-1 reduced NFB transcriptional activation by inhibiting NIK-mediated phosphorylation of IKK␣. Although IKK␣ and IKK␤ were both involved in IL-1␤-mediated activation of NFB, only the IKK␣-dependent component was modulated by changes in H 2 O 2 levels. Interestingly, in vitro reconstitution experiments demonstrated that NIK was activated by a very narrow range of H 2 O 2 (1-10 M), whereas higher concentrations (100 M to 1 mM) inhibited NIK activity. Treatment of cells with the general Ser/Thr phosphatase inhibitor (okadaic acid) lead to activation of NFB and enhanced NIK activity as a IKK␣ kinase, suggesting that ROS may directly regulate NIK through the inhibition of phosphatases. Recruitment of NIK to TRAF6 following IL-1␤ stimulation was inhibited by H 2 O 2 clearance and Rac1 siRNA, suggesting that Rac-dependent NADPH oxidase may be a source of ROS required for NIK activation. In summary, our studies have demonstrated that redox regulation of NIK by H 2 O 2 is mechanistically important in IL-1␤ induction of NFB activation.
Reactive oxygen species (ROS) 2 have been implicated in a number of pro-inflammatory signal transduction cascades activated by IL-1␤, TNF␣, and lipopolysaccharide (1,2). In this context, ROS have been considered second messengers. For example, ROS are generated in response to IL-1␤, TNF␣, and lipopolysaccharide (3)(4)(5), and clearance of intercellular ROS can inhibit the ability of these ligands to activate downstream signals, including NFB (6). However, despite the fact that ROS have been linked to NFB activation by certain cytokines, the molecular mechanisms remain poorly defined and controversial (7). In the present study, we sought to investigate the molecular mechanism by which ROS regulate IL-1␤ induction of NFB.
IL-1␤ is a potent pro-inflammatory cytokine that plays an important role in immune and inflammatory responses associated with sepsis (8), arthritis (9), and cancer (10). Septic shock is a systematic inflammatory response to infection that leads to the overproduction of a series of pro-inflammatory cytokines. TNF␣ and IL-1␤ are two critical cytokines produced during septic shock, which significantly contribute to morbidity associated with this disease (8,11). IL-1␤ exerts its pleiotropic effects by binding to its receptor (IL-1R1) on the plasma membrane. This initiates the IL-1␤ signaling cascade by activating structural changes in the receptor that dock cytoplasmic adaptor and effector proteins on the receptor tail (12). The first set of proteins recruited to the cytoplasmic tail of IL-1R1 following ligand binding include IL-1RacP, MyD88, and Tollip (13)(14)(15)(16). In turn, MyD88 plays an obligatory role in mediating the recruitment of interleukin-1 receptor-associated kinase family members via interactions with its N-terminal death domain (15,17). Once associated with the receptor complex, interleukin-1 receptor-associated kinase subsequently recruits TRAF6, a member of the TNF receptor-associated factor family of adaptor proteins (18). TRAF6 then recruits and activates TAK1 and/or NIK (19), two IKK kinases important in NFB activation. It has also been suggested that TAK1 and its regulators, TAB1 and -2, might act upstream of NIK (20).
The IL-1␤ signaling cascade leads to the activation of several key transcription factors that modulate the expression of genes involved in immunity and inflammation. These include NFB, AP-1, and p38 MAPK (21)(22)(23). In this context, NFB is known for its ability to induce inflammatory cytokines, inhibit apoptotic pathways, and lead to the resolution of inflammation (23)(24)(25). NFB is composed of homo-and hetero-dimers of the Rel family of transcription factors. p65 and p50 subunits represent the most common types of subunits found in the activated NFB complex. Activation of NFB is controlled by a family of IB inhibitory proteins (IB␣, IB␤, and IB⑀) that sequester the NFB complex in the cytoplasm by masking its nuclear localization signal. Pro-inflammatory stimuli that activate NFB lead to the phosphorylation of IB␣ on two N-terminal serines . This results in the ubiquitination and degradation of IB␣ and the mobilization of NFB to the nucleus where it activates transcription (26). Two IKK kinases, IKK␣ and IKK␤, have the ability to phosphorylate these two serines on IB␣ (27,28). IKK␣ and IKK␤, together with IKK␥, form a large complex within the cell (29).
Kinases known to phosphorylate IKK␣ and IKK␤ include NIK (30), MEKK1 (31), NFB-activating kinase (32), and TAK1 (19). The NFBinducing kinase (NIK) is a member of the MAPK kinase kinase family, which has Ser/Thr kinase activity. NIK was first identified as a TRAF2 interacting protein. Further studies found that it could also associate with TRAF6 and TRAF5 (33). NIK is dedicated to NFB signaling and contributes to the induction of NFB by both TNF␣ and IL-1␤ (34). NIK has been shown to interact with IKK␣ in yeast two-hybrid systems (35) and has the potential to specifically phosphorylate IKK␣ (36).
It has been suggested that ROS play an important role in the activation of NFB (37), however, this concept remains controversial (7). Various forms of ROS, including superoxide anion (O 2 . ) and hydrogen peroxide (H 2 O 2 ), have been implicated in cell signaling. ROS are tightly controlled in cells by a group of antioxidant factors, including glutathione peroxide (GPx), superoxide dismutase (SOD), catalase, peroxiredoxins, and small factors such as glutathione and thioredoxin. Among these, GPx is a group of selenoenzymes responsible for reducing various hydroperoxides in the presence of the reduced form of glutathione (38). At least four GPx isoforms have been identified (39). GPx-1 is the isoform that exists in the cytoplasm, and it uses glutathione to degrade H 2 O 2 into water.
A number of studies have revealed that ROS are involved in IL-1␤ signal transduction. For example, IL-1␤ stimulates ROS generation in a number of cell systems (40 -42). In addition, studies that have used chemical antioxidants, such as diamide, N-acetyl-L-cysteine, and phenylarsine oxide, have shown significant effects on IL-1␤-induced signaling (43,44). Although there remain disagreements over the applications of these antioxidants that might weaken this argument (7), a number of studies have also used specific ROS modulation enzymes to confirm the importance of ROS. For example, specific antioxidant enzymes, such as GPx-1 and Mn-SOD, have demonstrated strong inhibitory effects on TNF␣-or IL-1␤-induced signal transduction (6,45). In addition, factors that control ROS generation by cells have been shown to influence IL-1␤ induction of NFB and the transcriptional activation of downstream genes (46,47). However, the mechanisms of ROS action in the IL-1␤ pathway remain poorly defined. In

EXPERIMENTAL PROCEDURES
Recombinant Adenoviral Vector Infection and siRNA Transfection-Adenoviral infections were performed in serum-free medium for 2 h at an m.o.i. of 500 particles/cell, followed by the addition of an equal volume of fresh media containing 20% fetal bovine serum. Cells were fed with fresh media at 24 h and cells were analyzed at 48 h post-infection. These conditions produced Ͼ90% transduction with recombinant adenovirus, as assessed with Ad.CMV-GFP reporter gene expression. Eight different types of recombinant adenoviruses were used, including: Ad.BglII (empty control vector that does not express a transgene), Ad.NFBLuc (a NFB-responsive luciferase reporter vector) (48), Ad.GPx-1 (GPx-1 tagged with a c-Myc epitope at the N terminus) ( (49,50), Ad.IB␣AS (an adenoviral vector that expresses the antisense IB␣ cDNA and activates NFB by reducing levels of the IB␣ repressor) (51), and Ad.IB␣(S/A) (an adenoviral vector that expresses a dominant negative IB␣ mutant (S32/36A) that inhibits NFB activation by preventing IKK-mediated phosphorylation of IB␣) (52). For NFB transcriptional luciferase assays, MCF-7 cells were infected with Ad.NFBLuc at an m.o.i. of 500 particles/cell 24 h prior to experimental treatments. Human NIK siRNA and Rac1 siRNA were purchased from a library of prescreened siRNAs made by Santa Cruz Biotechnology, and transfections were performed following the manufacturer's protocol.
Cell Culture and Treatment-MCF-7 cells (a human breast cancer cell line obtained from ATCC) were chosen for studies due to their low level of endogenous GPx-1. MCF-7 cells were grown in minimal essential medium with Eagle's salts and L-glutamine, 1% minimal essential medium non-essential amino acids, 10% fetal bovine serum, 1% penicillin/streptomycin. For H 2 O 2 treatment, concentrated H 2 O 2 (30%) (Fisher Scientific, Fair Lawn, NJ) was diluted to 1 M with deionized H 2 O and added to fresh medium at a final concentration of 1 mM. The spent medium was removed from MCF-7 cells and quickly replaced with medium containing 1 mM H 2 O 2 . After incubation at 37°C for 1 h (or as otherwise indicated), the medium was changed to fresh medium without H 2 O 2 , and incubation of cells at 37°C was continued. IL-1␤ inductions were performed using 1 ng/ml human IL-1␤ (R&D Systems, Minneapolis, MN) for the indicated times. Control MCF-7 cells were also fed fresh medium but did not receive any treatment. Cells were harvested for luciferase assays at 6 h following H 2 O 2 or IL-1␤ treatments. MCF-7 cells were washed twice with ice-cold PBS and were prepared for each assay accordingly.
Western Blots-Cell lysates were prepared and normalized for protein concentrations using a Bio-Rad Kit (Bio-Rad, Philadelphia, PA). Western blotting was performed using standard protocols. In brief, 50 g of crude proteins for each condition were separated on a denaturing 10% SDS-PAGE and transferred to nitrocellulose (Hybond C, Amersham Biosciences). The membranes were then blocked and probed with primary antibody for 1 h at room temperature using dilutions suggested by the manufacturer. After being washed with blocking buffer several times, the membranes were probed with the appropriate dilution of secondary antibody. Immunoreactive proteins were detected using either an enhanced chemiluminescence ECL (Amersham Biosciences) and exposure to x-ray film, or an Odyssey Infrared Imaging System (LI-COR Biotech, Lincoln, NE). The antibodies against IKK␣, IKK␤, NIK, GST, FLAG, IL-1R1, MEKK1, and TRAF6 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Rac1 antibody was purchased from BD Transduction Laboratories (Lexington, KY).
Luciferase Assays-Luciferase activity was measured using a kit from Promega (Madison, WI) according to the manufacturer's instructions. MCF-7 cells were infected with Ad.NFBLuc 24 h prior to treatment. Ad.NFBLuc contains the luciferase gene driven by four tandem copies of the NFB consensus sequence fused to a TATA-like promoter from the herpes simplex virus thymidine kinase gene. 5 g of total protein from each sample was used to perform the luciferase assays.
Immunoprecipitation-Cell samples were washed with ice-cold PBS twice and were lysed with radioimmune precipitation assay buffer (150 mM NaCl, 50 mM Tris, pH 7.2, 1% deoxycholate, 1% Triton X-100, 0.1% SDS) at 4°C for 30 min. Protein concentrations were determined using a Bio-Rad kit and 500 g of cellular protein, and 5 l of primary antibody was mixed with 1 ml of radioimmune precipitation assay buffer at 4°C for 1 h. Then 50 l of Protein-A-agarose beads (Santa Cruz Biotechnology) was added to the mixture, and the mixture was rotated for 4 h. The beads were spun down at 5000 rpm for 5 min at 4°C and washed with ice-cold PBS three times prior to analysis of immunoprecipitates.
In Vitro Kinase Assay-Kinases (NIK, MEKK1, IKK␣, or IKK␤) were immunoprecipitated with their respective antibodies and then mixed with 1 g of the appropriate protein substrate (IKK␣, IKK␤, or IB␣) in 0.3 mM cold ATP, 10 Ci of [␥-32 P]ATP, and 10 l of kinase buffer (40 mM Hepes, 1 mM ␤-glycerophosphate, 1 mM nitrophenol phosphate, 1 mM Na 3 VO 4 , 10 mM MgCl 2 , and 2 mM dithiothreitol). Reaction mixtures were incubated at 30°C for 30 min (or shorter times as indicated), and reactions were then terminated by the addition of SDS-PAGE loading buffer at 98°C for 5 min. Proteins were separated by 10% SDS-PAGE and transferred to nitrocellulose membrane and exposed to x-ray film.
Detection of Cellular ROS Production Using H 2 DCFDA-Stock solutions of H 2 DCFDA (Molecular Probes, Eugene, OR) were generated in Me 2 SO at a concentration of 50 g/ml immediately prior to use. Cells were washed three times with PBS prior to simultaneous treatment with H 2 DCFDA (10 M) and IL-1␤ (1 ng/ml) for 20 min in PBS at 37°C in the dark. For samples infected with adenoviral vectors or transfected with siRNAs, this was done 48 h prior to stimulation with IL-1␤. When DPI (10 M) was used to inhibit NADPH oxidases, it was added at the time of IL-1␤ stimulation. Cells were washed in PBS at 20 min post-stimulation and were then fixed for 10 min in 4% paraformaldehyde. Cells were subsequently mounted in 4Ј,6-diamidino-2-phenylindole containing antifadent and were examined by fluorescent microscopy for DCF signal. Exposure times were constant for all experimental samples.

Cellular H 2 O 2 Influences IL-1␤-mediated NFB Activation-Previ-
ous studies have demonstrated that IL-1␤ stimulation of various cell types leads to cellular ROS production (53-55). Others have also demonstrated that H 2 O 2 influences NFB transcriptional activation follow-ing a number of stimuli, including TNF, UV, and IL-1␤ (3,6). Furthermore, direct treatment of cells with H 2 O 2 has the ability to activate NFB (6,56,57). However, one study has also suggested that NFB activation by IL-1␤ occurs in the absence of induced ROS in epithelial cells (42). In the present study, we sought to better understand potential mechanisms by which H 2 O 2 influences IL-1␤-mediated activation of NFB. The MCF-7 breast cancer cell line was chosen for these studies, because it expresses a very low level of endogenous GPx-1 (an antioxidant enzyme responsible for degrading cellular H 2 O 2 to water) and enables efficient modulation of cellular H 2 O 2 levels through the overexpression of recombinant human GPx-1 (6).
We first sought to confirm that IL-1␤ stimulation of MCF-7 cells led to an increase in cellular H 2 O 2 . Indeed, our studies demonstrated that IL-1␤ stimulation of MCF-7 cells enhanced H 2 DCFDA fluorescence, suggesting that H 2 O 2 is elevated following IL-1␤ treatment (Fig. 1A). Furthermore, infection with recombinant adenovirus expressing GPx-1 (Ad.GPx-1), but not the control Ad.BglII empty vector, significantly attenuated H 2 DCFDA fluorescence following IL-1␤ stimulation (Fig.  1A). These findings confirmed that IL-1␤ stimulates cellular H 2 O 2 production in MCF-7 cells and that ectopic expression of GPx-1 could successfully modulate the cellular redox environment.
We next evaluated the extent to which GPx-1 expression could modulate IL-1␤ induction of NFB using a NFB-dependent luciferase reporter. Indeed, we observed that IL-1␤ induction of NFB at 6 h was significantly attenuated following infection with Ad.GPx-1 virus, as compared with the infection with an empty adenoviral control vector (Ad.BglII) (Fig. 1B). In contrast, ectopic expression of GPx-1 had no effect on baseline transcriptional activity of NFB. As a control for the ability of GPx-1 expression to modulate H 2 O 2 induction of NFB, we performed similar studies in MCF-7 cells transiently treated with 1 mM H 2 O 2 for 1 h and assayed for NFB activation at 6 h. Although GPx-1 expression inhibited direct H 2 O 2 induction of NFB to a greater extent than IL-1␤, the level of inhibition was still incomplete (ϳ50%) (Fig. 1C). These results suggested that cellular H 2 O 2 partially contributes to NFB activation by IL-1␤ and that GPx-1 expression more than likely can only partially attenuate H 2 O 2 levels in the cell. The ability of GPx-1 expression to only partially block cellular H 2 O 2 levels is also supported by H 2 DCFDA experiments (Fig. 1A).
H 2 O 2 Influences NFB Activation Upstream of IB␣ Phosphorylation-We next sought to investigate the molecular mechanism through which H 2 O 2 influences IL-1␤-mediated NFB transcriptional activation. IB␣ plays a pivotal role in NFB activation; phosphorylation of IB␣ on two serines (Ser-32/Ser-36) leads to its disassociation from NFB in the cytoplasm, proteasome-dependent degradation of IB␣, and mobilization of NFB to the nucleus. To evaluate whether H 2 O 2 influences NFB activation downstream of IB␣ (i.e. after NFB dissociates from IB␣ and moves to the nucleus), we used an antisense IB␣ cDNA expressed from recombinant adenovirus (Ad.IB␣(AS)) that represses IB␣ protein levels and induces nuclear translocation of NFB (51). As previously shown in HeLa cells (51), infection with Ad.IB␣(AS) significantly reduced the level of IB␣ protein in MCF-7 cells (data not shown) and led to the induction of NFB transcriptional activity in the absence of cytokine stimulation ( Fig. 2A). In the presence of antisense IB␣ mRNA, overexpression of GPx-1 did not alter NFB transcriptional activity, suggesting that H 2 O 2 levels do not directly influence the transcriptional activity of the NFB complex once it has been mobilized from IB␣.
We next sought to evaluate whether H 2 O 2 might influence IL-1␤-dependent NFB activation by regulating factors that phosphorylate IB␣.
Indeed, the majority of NFB activation following IL-1␤ stimulation is mediated by IB␣ serine phosphorylation, as indicated by a nearly complete block in NFB activation by adenoviral expression of the dominant negative IB␣ mutant (IB␣S32/36A) (Fig. 2B). Furthermore, overexpression of GPx-1 and IB␣S32/36A gave a similar level of inhibition as seen with IB␣S32/36A alone, suggesting that GPx-1 regulates components of the NFB pathway upstream of IB␣. Based on previous reports for other stimuli demonstrating redox-dependent activation of the IKK complex (6, 48, 58 -61), we reasoned that H 2 O 2 might also influence activation of the IKK complex following IL-1␤ stimulation.
To approach this question, we used two dominant negative adenoviral vectors that express mutant forms of either the IKK␣ (Ad.IKK␣KM) or IKK␤ (Ad.IKK␤KA) subunits of the IKK complex. Results from these analyses demonstrated that Ad.IKK␣KM or Ad.IKK␤KA infection of MCF-7 cells inhibited ϳ50% of IL-1␤-dependent NFB activation (Fig.  2C). The level of inhibition observed by overexpression of GPx-1 was similar to that seen with IKK␣KM and slightly less than that seen with IKK␤KA. Co-infection with Ad.IKK␣KM and Ad.IKK␤KA gave nearly complete inhibition of NFB activation, suggesting that these two subunits of the IKK complex primarily control NFB activation by IL-1␤. To address whether GPx-1-sensitive signaling was directed through IKK␣ or IKK␤, we performed co-expressing studies with each of these IKK mutants and GPx-1. We reasoned that if H 2 O 2 influenced IKK␣ activation, inhibition of NFB in the presence of IKK␣KM would be unaffected by GPx-1 co-expression. Furthermore, if this hypothesis was true, we would expect to observe enhanced inhibition of NFB following co-infection with Ad.IKK␤KA plus Ad.GPx-1, as compared with Ad.IKK␤KA or Ad.GPx-1 infection alone. The reverse scenario would be true if H 2 O 2 selectively influenced IKK␤ activation (i.e. Ad.IKK␣KM plus Ad.GPx-1 infection would provide an enhanced level of inhibition, FIGURE 2. GPx-1-mediated clearance of H 2 O 2 inhibits NFB activation upstream to IB␣ stabilization by acting on a pathway that leads to IKK␣ activation. MCF-7 cells were infected with the indicated recombinant adenoviral vectors at an m.o.i. of 500 particles/cell for each virus. After 24 h, cells were re-infected with Ad.NFBLuc for 24 h at 500 particles/cell prior to direct analysis (A) or stimulation with IL-1␤ (1 ng/ml) (B and C). A, the ability of antisense IB␣ mRNA expression (Ad.IB␣(AS)) to constitutively induce NFB transcriptional activation was assessed in the presence of Ad.BglII (control vector) or Ad.GPx-1 co-infection. B and C, NFB transcriptional activation was evaluated at 6 h following treatment with IL-1␤ by assessing the relative luciferase activity in 5 g of protein lysate. In panel B, the mean relative light units (RLU/min) (ϮS.E., n ϭ 3) are given for each sample as an index of NFB transcriptional activation. In C, the mean change (ϮS.E., n ϭ 3) in NFB transcriptional activation following IL-1␤ stimulation is given (the baseline level of luciferase activity, as determined from Ad.BglII-infected cells in the absence of IL-1␤, was subtracted from all experimentally induced values). Statistical comparisons of marked groups using the Student's t test are given with p values.
as compared with each vector alone). Our results from these experiments (Fig. 2C) demonstrated significant synergism in the inhibition of IL-1␤-dependent NFB activation in the presence of both IKK␤KA and GPx-1 expression, as compared with each individually. In contrast, inhibition of NFB was similar following IKK␣KM plus GPx-1 overexpression, as compared with expression of IKK␣KM or GPx-1 alone (Fig. 2C). These findings suggest that GPx-1 acts to inhibit IKK␣, but not IKK␤, activation following IL-1␤ stimulation. Because both IKK␣ and IKK␤ contribute to NFB activation by IL-1␤, we conclude that ROS control only half of the NFB activation pathways in response to IL-1␤.
H 2 O 2 Regulates IKK Kinase Activity-Given that GPx-1 expression appeared to modulate IKK␣ activation following IL-1␤ stimulation, we next sought to directly confirm that H 2 O 2 could activate the IKK complex in vivo. To this end, we used in vitro kinase assays to monitor IKK activity following IL-1␤ stimulation and compared these results to that observed following direct H 2 O 2 stimulation of MCF-7 cells. This assay utilized immunoprecipitated IKK␣ or IKK␤, followed by in vitro phosphorylation of GST-IB␣ in the presence of [␥-32 p]ATP. Following IL-1␤ stimulation, both IKK␣ and IKK␤ kinase activities were substantially increased, peaking at ϳ15-30 min following stimulation (Fig. 3A). The total cellular protein levels of both kinases were also examined, and no changes occurred following IL-1␤ stimulation (Fig. 3C). However, because IKK␣ and IKK␤ form a complex in vivo (as evident by immunoprecipitation of IKK␣ with an IKK␤-specific antibody, Fig. 3C), it was impossible to separate the extent to which IKK␣ and/or IKK␤ was activated in the IKK complex. Similar changes in IKK␣-and IKK␤-mediated GST-IB␣ phosphorylation were seen following treatment of cells with 1 mM H 2 O 2 (Fig. 3B). Collectively, these studies demonstrated that H 2 O 2 can directly activate the IKK complex in vivo and suggested a plausible mechanism for IL-1␤-induced regulation of NFB.
H 2 O 2 -dependent Activation of NIK Contributes to NFB Activation by IL-1␤-We next sought to investigate the molecular mechanism by which H 2 O 2 preferentially modulated IKK␣ following IL-1␤ stimulation. Previous reports have implicated the redox-dependent activation of NIK following H 2 O 2 treatment of cells (62). Furthermore, NIK has previously been demonstrated to preferentially phosphorylate IKK␣ over IKK␤ (36). Hence, NIK was an obvious potential candidate for the redox regulation of NFB by IL-1␤. To investigate the involvement of NIK in NFB activation following IL-1␤ stimulation, we utilized a dominant negative NIK mutant (49,50). Adenovirus-mediated expression of this NIK mutant in MCF-7 cells (Fig. 4A) significantly inhibited both H 2 O 2 -and IL-1␤-mediated activation of NFB (Fig. 4B). To evaluate whether the GPx-1-sensitive component of NFB activation by IL-1␤ was mediated through NIK, we performed co-infection experiments with Ad.NIK(DN) and Ad.GPx-1 virus. Results from these experiments demonstrated that expression of dominant negative NIK or GPx-1 significantly inhibited NFB activation to similar extents following H 2 O 2 (Fig. 4C) or IL-1␤ stimulation (Fig. 4D). Importantly, the level of NFB inhibition seen following combined infection of cells with Ad.NIK(DN) and Ad.GPx-1 virus was similar to that seen following infection with each vector alone for both H 2 O 2 (Fig. 4C) and IL-1␤ (Fig. 4D) stimulations. These results support the hypothesis that H 2 O 2 activation of NIK plays an important role in NFB activation by IL-1␤.
Redox Activation of NIK Preferentially Regulates IKK␣ following IL-1␤ Stimulation-Our data thus far have implicated NIK in the redoxdependent activation of NFB by IL-1␤. Furthermore, our results suggest that activation of IKK␣ was involved in this pathway of redox activation. To this end, we hypothesized that H 2 O 2 activation of NIK enhanced its ability to phosphorylate IKK␣. Given the apparent specificity of the redox component of IL-1␤ signaling for the IKK␣ subunit of the IKK complex (Fig. 2C), we first sought to evaluate whether NIK specifically controlled IKK␣ activation in response to IL-1␤ or H 2 O 2 stimulation. To this end, we generated bacterial GST fusion proteins for both IKK␣ and IKK␤ and evaluated the ability of NIK to phosphorylate these two proteins following IL-1␤ or H 2 O 2 stimulation. Generation of full-length IKK␣ and IKK␤ GST fusion protein was not successful due to bacterial toxicity. As an alternative strategy, we generated truncated IKK␣ and IKK␤ GST fusion proteins that comprised their respective activation domains phosphorylated by IKK kinases. For IKK␣, this region included the sequence between Gly-131 to Trp-205, a 75-amino acid peptide containing Ser-176 and Ser-180 sites known to be phosphorylated and critical for IKK␣ activity. Similarly, an IKK␤ fusion was generated from Ala-132 to Trp-206. This 75-amino acid peptide contained Ser-177 and Ser-181 sites known to be phosphorylated and critical for IKK␤ activity (Fig. 5A). These two fusion proteins were purified (Fig. 5B) and used for in vitro kinase assays following H 2 O 2 or IL-1␤ treatment. Results from these studies demonstrated that H 2 O 2 and IL-1␤ treatments stimulated immunoprecipitated NIK to phosphorylate GST-IKK␣, but not GST-IKK␤ (Fig. 5C). Furthermore, the extent of NIK activation by these two stimuli (Fig. 5C) was reflected in their respective abilities to activate NFB in transcriptional assays (Fig. 4B) (i.e. IL-1␤ Ͼ H 2 O 2 ). To confirm that the inability of NIK to phosphorylate GST-IKK␤ was not the result of poor fusion protein quality, we evaluated the ability of MEKK1 to phosphorylate GST-IKK␤ following TNF␣ treatment. MEKK1 has been shown to preferentially phosphorylate IKK␤ (36). Indeed, immunoprecipitated MEKK1, from TNF␣treated cells, had a greater ability to phosphorylate GST-IKK␤ as compared with GST-IKK␣ (Fig. 5C). These results confirmed that both GST-IKK␤ and GST-IKK␣ were receptive kinase substrates and demonstrated that IL-1␤ and H 2 O 2 treatments stimulate NIK to preferentially phosphorylate IKK␣.
To provide further evidence that NIK predominantly signals through IKK␣ in the context of IL-1␤-stimulated NFB activation in vivo, we investigated whether expression of dominant negative NIK enhanced inhibition of NFB in the presence of IKK␣KM or IKK␤KA dominant mutants. Results from these experiments (Fig. 5D) demonstrated that Ad.NIK(DN) infection was only able to augment inhibition of IL-1␤mediated NFB activation in the presence of Ad.IKK␤KA but not Ad.IKK␣KM. These results mirrored those seen following co-infection with the IKK mutants and GPx-1 (Fig. 2C), suggesting that NIK acts predominantly through IKK␣, but not IKK␤, in vivo following IL-1␤ stimulation to induce NFB.
To directly evaluate the redox dependence of NIK activation following IL-1␤ stimulation, we next asked whether GPx-1 overexpression inhibited the ability of NIK to phosphorylate GST-IKK␣ in an in vitro kinase assay. As a control, we first tested whether activation of NIK, following treatment of MCF-7 cells with H 2 O 2 , would be inhibited by GPx-1 overexpression. Indeed, as shown in Fig. 5E, immunoprecipitated NIK from H 2 O 2 -treated cells had significantly attenuated GST-IKK␣ kinase activity following Ad.GPx-1 infection. Similarly, IL-1␤stimulated NIK kinase activity was significantly reduced in the presence of GPx-1 expression (Fig. 5E) 5). However, the extent of NIK activation following in vitro treatment with 1-10 M H 2 O 2 was still significantly lower than the level of NIK activation seen following direct treatment of MCF-7 cells with 1 mM H 2 O 2 for 30 min (Fig. 6A, lane 1). These findings support the notion that other cellular factors may be required to facilitate the redox-dependent activation of NIK and that these factors are predominantly lost during the immunoprecipitation of inactive NIK.
To test whether other cellular factors were required for the H 2 O 2 -dependent activation of NIK, we performed in vitro NIK activation assays in crude cell lysates harvested from both untreated and H 2 O 2 treated MCF-7 cells. Crude lysates were first treated with 1 M to 1 mM H 2 O 2 for 30 min, and then NIK was immunoprecipitated and assayed for its ability to phosphorylate GST-IKK␣. Crude lysates generated from cells treated with 1 mM H 2 O 2 served as an internal control for the maximal achievable NIK activation at each in vitro concentration of H 2 O 2 used to stimulate NIK in crude lysates. Several interesting findings emerged from these studies (Fig. 6B). First, as seen following in vitro exposure of immunoprecipitated NIK (Fig. 6A, lanes 2 and 3), only 1 and 10 M concentrations of H 2 O 2 were able to activate NIK in crude lysates from   1, 7, and 9 to lanes 3 and 5). This inhibition was also evident in lysates derived from untreated cells treated with increasing concentrations of H 2 O 2 (compare lanes 4, 6, 8, and 10). In summary, these results substantiate the findings that only very narrow ranges of H 2 O 2 can activate NIK and that unknown cytoplasmic factors enhance the redox-dependent activation of NIK.
The above studies suggested that H 2 O 2 was able to activate NIK to phosphorylate IKK␣. However, an alternative possibility was that some other unknown redox-regulated IKK␣ kinases (i.e. TAK1, etc.) might associate with NIK and thereby co-immunoprecipitate in our kinase assays and complicate the assignment of GST-IKK␣ phosphorylation to NIK. To formally address this possibility, we expressed FLAG-tagged wild type NIK (WT-NIK) or the dominant negative NIK mutant (DN-NIK) and assessed IKK␣ kinase activity following immunoprecipitation of the FLAG tag from H 2 O 2 -treated MCF-7 cell lysates. We reasoned that if H 2 O 2 activated an alternative IKK␣ kinase that associated with NIK, this would be revealed as residual H 2 O 2 -induced IKK␣ kinase activity in precipitates of the kinase-dead DN-NIK mutant. Results from these experiments are shown in Fig. 6C. Immunoprecipitated FLAGtagged WT-NIK demonstrated a significant increase in its ability to phosphorylate GST-IKK␣ following H 2 O 2 treatment (Fig. 6C, compare  lanes 7 to 8). This level of IKK␣ kinase activity was similar to what had been seen following immunoprecipitation of endogenous NIK (com- pare lanes 1 to 2). In contrast, FLAG immunoprecipitates from LacZ (negative control without a FLAG tag) or FLAG-tagged DN-NIK transfected cells failed to demonstrate H 2 O 2 -induced IKK␣ kinase activity (lanes 3-6). These studies suggest that other H 2 O 2 -activated IKK␣ kinases likely do not associate with NIK. However, given that the DN-NIK construct is a truncation mutant (49, 50), we cannot currently rule out that an alternative IKK␣ kinase might associate with the deleted region of NIK. However, we failed to see TAK1 (a known alternative IKK␣ kinase) association with NIK in MCF-7 cells prior to or following IL-1 or H 2 O 2 treatment (data not shown), suggesting that if this occurs it does not involve TAK1.
The third potential mechanism by which H 2 O 2 might enhance NIK activation during IL-1␤ signaling includes H 2 O 2 -mediated inhibition of phosphatases that inactivate NIK. NIK is known for its ability to autophosphorylate, so redox regulation by phosphatases seems reasonable (62). Although it is well recognized that H 2 O 2 -dependent inactivation of phosphatases plays important roles in signaling (45,63), information on phosphatase regulation of NIK is lacking. To approach this question we assessed the effects of okadaic acid (a general Ser/Thr phosphatase inhibitor) on both NFB activation and the activation of NIK to phosphorylate GST-IKK␣. Indeed, treatment of cells with increasing concentrations of okadaic acid significantly increased the transcriptional activation of NFB in MCF-7 cells (Fig. 7A). Similarly, okadaic acid treatment of MCF-7 cells significantly enhanced the IKK␣ kinase activity of immunoprecipitated NIK (Fig. 7B) and also enhanced the association of NIK with TRAF6 (Fig. 7C, compare lanes 1 and 7). These findings suggest that certain Ser/Thr phosphatases may indeed play a role in NIK activation and its association with TRAF6; as such, Ser/Thr phosphatases are potential targets of H 2 O 2 -mediated inhibition following IL-1␤ stimulation.
H 2 O 2 Modulates NIK Association with TRAF6-Results thus far have demonstrated that IL-1␤-mediated activation of NIK leads to enhanced IKK␣ phosphorylation and is partially responsible for activation of NFB. In vitro, the process of NIK activation by H 2 O 2 appears to require unknown cellular factors and may also involve inhibition of phosphatases in vivo. TRAF6, which recruits NIK to the IL-1 receptor complex, is an integral part of generating an active IKK␣ kinase complex following IL-1␤ stimulation (64). To this end, we sought to investigate whether H 2 O 2 modulated the association of NIK with TRAF6. Such a mechanism could explain why cell lysate was required for maximal H 2 O 2mediated activation of immunoprecipitated NIK (Fig. 6, A and B). We evaluated the extent to which IL-1␤ or H 2 O 2 treatment enhanced the association of NIK with immunoprecipitated TRAF6. Results from these experiments demonstrated that IL-1␤ or H 2 O 2 stimulation of MCF-7 cells increased the association of NIK with TRAF6 (Fig. 7C). Furthermore, degradation of H 2 O 2 by GPx-1 expression inhibited this association (Fig. 7C, lanes 3 versus 4 and lanes 5 versus 6). As expected, the increased association between TRAF6 and NIK promoted by phosphatase inhibition (i.e. OA treatment) was unaffected by GPx-1-mediated clearance of H 2 O 2 (Fig. 7C, lanes 7 versus 8). Together with earlier studies, these experiments provide strong support that H 2 O 2 regulates NIK activity by modulating the association between NIK and TRAF6.
Rac1 and NADPH Oxidase Control the Redox-dependent Association of NIK with TRAF6-The source of ROS generation following IL-1␤ stimulation remains complex and controversial. Several studies have indirectly implicated NADPH oxidases as a ROS source, based on the ability of diphenyleneiodonium (a NADPH oxidase inhibitor, DPI) to prevent ROS-dependent activation of IL-1␤ induced genes such as E-selectin, inducible nitric-oxide synthase, c-fos, and collagenase (46,65,66). However, others have suggested that 5-lipoxygenase may be involved in IL-1␤ induction of ROS in lymphoid cells, while NADPH oxidase plays a selective role in monocytic cell-induced ROS following IL-1␤ stimulation (42). Rac1, a small GTPase, plays a central role in cellular ROS generation through certain NADPH oxidases (67). Rac1 has also been linked to IL-1␤ induction of p65NFB in a murine thymoma cell line (47). However, it has been suggested that in epithelial cells, Rac1 and NADPH oxidase do not play a role in NFB activation by IL-1␤ (42). Given the controversy surrounding potential sources of ROS following IL-1␤ stimulation, we sought to investigate the potential role of Rac1/ FIGURE 6. H 2 O 2 can directly activate NIK to phosphorylate IKK␣ in the presence of cellular lysate. The ability of H 2 O 2 to activate IKK␣ kinase activity of NIK was evaluated using in vitro reconstitution experiments with immunoprecipitated NIK. A, NIK was immunoprecipitated from untreated MCF-7 cell lysates and exposed to increasing concentrations of H 2 O 2 in vitro for 30 min prior to performing GST-IKK␣ kinase assays (lanes 2-6). As a positive control, NIK was also immunoprecipitated from MCF-7 cell lysates following an in vivo 30 min treatment with 1 mM H 2 O 2 (lane 1). The reactions were analyzed by SDS-PAGE and transferred to nitrocellulose membrane prior to autoradiography. B, whole cell lysates were generated from MCF-7 cells exposed to an in vivo 30-min treatment with 1 mM H 2 O 2 (odd-numbered lanes) or following no treatment (even-numbered lanes) as marked. These crude lysates were then exposed to increasing concentrations of H 2 O 2 in vitro for 30 min as indicated. NIK was then immunoprecipitated and in vitro kinase assays were performed in the presence of [␥-32 P]ATP and GST-IKK␣ for 30 min at 30°C. The reactions were analyzed by SDS-PAGE and transferred to nitrocellulose membrane prior to autoradiography. C, MCF-7 cells were transfected with LacZ (negative control), FLAG-tagged wild-type NIK, or FLAGtagged dominant negative NIK plasmid expression constructs 48 h prior to treatment with H 2 O 2 (1 mM) for 30 min. Cell lysates were prepared and endogenous NIK or recombinant NIK was immunoprecipitated with anti-NIK or anti-FLAG antibodies, respectively. Immunoprecipitated NIK was then used for in vitro kinase assays with [␥-32 P]ATP and GST-IKK␣. The reactions were analyzed by SDS-PAGE and transferred to nitrocellulose membrane prior to autoradiography and Western blotting to detect FLAG and NIK.
NADPH oxidase in the IL-1␤-induced ROS found in MCF-7 mammary epithelial cells.
Using Rac1 siRNA to inhibit Rac1 and DPI to inhibit NADPH oxidase, we investigated the role of Rac1/NADPH oxidase in ROS production following IL-1␤ stimulation. Results from H 2 DCFDA staining demonstrated that both DPI and Rac1 siRNA effectively reduced ROS production in MCF-7 cells following IL-1␤ stimulation (Fig. 8A). No inhibition in ROS was seen following transfection with a scrambled siRNA control. Rac1 siRNA also effectively inhibited total Rac1 protein levels in cell lysates (Fig. 8B). These findings provide strong evidence that a Rac1-regulated NADPH oxidase controls ROS production following IL-1␤ stimulation.
Given that Rac1 was in part required for the stimulation of ROS following IL-1␤ treatment of MCF-7 cells, we next sought to better understand if Rac1 was also required for NIK recruitment to TRAF6 following IL-1␤ stimulation. Rac1 siRNA indeed reduced the ability of NIK to associate with TRAF6 following IL-1␤ stimulation (Fig. 8C). However, this inhibition was not seen following transfection with a scrambled siRNA control. Combined with earlier results, these data suggest that Rac1-mediated H 2 O 2 -dependent activation of NIK, through the inhibition of phosphatases, promotes association of TRAF6 with NIK.
Data demonstrating ligand-independent association of activated NIK with TRAF6 following okadaic acid or H 2 O 2 treatment (Fig. 7) suggested that NIK can associate with TRAF6 prior to its recruitment to the receptor. However, it remained unclear if the redox-dependent activation of NIK was required for TRAF6 recruitment to IL-1R1. To address this question, we used NIK and Rac1 siRNAs to modulate the formation of NIK⅐TRAF6 complex formation following IL-1␤ stimulation. Results from these experiments demonstrated that NIK siRNA effectively inhibited total NIK in cell lysates (Fig. 8B), and as expected, also prevented NIK recruitment to TRAF6 following IL-1␤ stimulation (Fig.  8C). Importantly, inhibition of NIK protein levels had no effect on steady-state levels of TRAF6. To address whether NIK was required for TRAF6 recruitment to IL-1R1, we performed IL-1R1 pull-down assays in the presence of NIK, Rac1, or scrambled siRNAs (Fig. 8D). If the redox-sensitive complex formation between TRAF6 and NIK was absolutely required for binding of TRAF6 to IL-1R1, we would anticipate that both NIK and Rac1 siRNAs would prevent TRAF6 recruitment to IL-1R1. As shown in Fig. 8D, this was not the case. Rac1 siRNA effectively inhibited both the recruitment of NIK and TRAF6 to IL-1R1. However, NIK inhibition did not alter TRAF6 recruitment to IL-1R1 following IL-1␤ stimulation. This finding provides strong evidence that NIK binding to TRAF6 is not required for the recruitment of TRAF6 to ligand activated IL-1R1. Given that the association of TRAF6 with NIK can be directly activated by H 2 O 2 or okadaic acid in the absence of a ligand signal, our data suggest that redox activation of NIK likely promotes TRAF6⅐NIK complex formation both prior to and following TRAF6 recruitment to the IL-1 receptor.

DISCUSSION
The mechanism by which ROS stimulates the NFB pathway remains quite complex and multifaceted. IL-1␤ induction of the NFB pathway is one example for which molecular mechanisms of ROS action FIGURE 7. The association of NIK with TRAF6 following IL-1␤ stimulation is redox-regulated and is also enhanced by inhibition of Ser/Thr phosphatases. A and B, MCF-7 cells were infected with Ad.NFBLuc (500 particles/cell) for 24 h and were then treated with okadaic Acid (OA) at the indicated concentrations. Cells were harvested at 6 h post-OA treatment for luciferase activity assays (A) or at 30 min post-OA treatment for in vitro kinase assays using immunoprecipitated NIK and GST-IKK␣ as a substrate (B). Autoradiography of kinase products on SDS-PAGE were transferred to nitrocellulose membrane, and Western blots of the same membrane to detect GST are shown in B. C, MCF-7 cells were infected with Ad.GPx-1 or Ad.BglII at 500 particles/cell for 48 h followed by treatment with H 2 O 2 (1 mM), IL-1␤ (1 ng/ml), or OA (100 mM) for 30 min. Cell samples were then harvested and used for immunoprecipitation with anti-TRAF6 antibody. Immunoprecipitates were then analyzed by SDS-PAGE and Western blotting (top panel) using anti-TRAF6 and anti-NIK antibodies and infrared dye-conjugated secondary antibodies on an Odyssey infrared imaging system (LI-COR Biotechnology Lincoln, NE). Infrared quantification of NIK⅐TRAF6 band intensity ratios is plotted in the lower panel (mean Ϯ S.E. n ϭ 3). The Student's t test demonstrated a significant difference (p Ͻ 0.005) for marked comparisons below the graph. remain poorly elucidated. In the present study, we have shown that H 2 O 2 in part controls NFB activation by IL-1␤ by facilitating the activation of NIK and subsequent phosphorylation of IKK␣. ROS-mediated events that appear to be important for NIK-mediated activation of NFB following IL-1␤ stimulation include NIK association with TRAF6 and the inhibition of Ser/Thr protein phosphatases. Such findings suggest that H 2 O 2 may act to promote NIK activation and association with TRAF6 by inhibiting Ser/Thr protein phosphatases.
Rac1 appears to be a central player in the redox control of the IL-1␤ signaling pathway. The ability of Rac1 siRNA and DPI to inhibit ROS production in cells stimulated by IL-1 implicates Rac1-dependent NADPH oxidases as the cellular ROS source. Rac1 siRNA also inhibited both the redox-dependent formation of a TRAF6⅐NIK complex and recruitment of both TRAF6 and NIK to IL-1R1. Together with findings that GPx-1 expression inhibited NIK activation and TRAF6⅐NIK complex formation following IL-1␤ stimulation, these findings suggest H 2 O 2 is the central ROS mediator of this pathway. The importance of Rac1 in IL-1 signaling is supported by previous work demonstrating an association of Rac1 with the IL-1R1 complex through interactions with MyD88 and the IL-1 receptor accessory protein (47). This same study also demonstrated that the dominant negative N17Rac1 mutant prevented IL-1-mediated p65 transactivation. Hence, our studies now clarify that there is a central role for Rac1 in IL-1 signaling via facilitation of the redox activation of downstream effectors.
In vitro reconstitution experiments attempting to directly activate immunoprecipitated NIK with H 2 O 2 demonstrated that one or more cellular factors are required for the redox activation of NIK to phosphorylate GST-IKK␣ (Fig. 6). In vivo, H 2 O 2 promoted NIK association with TRAF6 in the absence of a ligand signal, and GPx-1 expression inhibited IL-1␤-induced TRAF6⅐NIK association and NIK activation (Figs. 7C and 5E). Given the close correlation between redox activation of NIK and its association with TRAF6, we anticipate that TRAF6 is a required cellular component necessary for NIK activation as an IKK␣ kinase. The ability of okadaic acid to directly promote NIK kinase activity and association with TRAF6, in the absence of a ligand signal, suggests that H 2 O 2 -mediated inhibition of protein phosphatases may be responsible for the redox activation of NIK. Hence, we favor a model whereby TRAF6 must recruit to NIK prior to phosphorylating IKK␣. Although NIK binding to TRAF6 may be necessary for an active IKK␣ kinase complex, this association was not necessary for TRAF6 to recruit to IL-1R1; TRAF6 effectively recruited to IL-1 stimulated IL-1R1 in the absence of NIK.
Cumulatively, these studies place NIK as a central redox-regulated signaling molecule in IL-1-mediated activation of NFB. We propose a model whereby IL-1 induces Rac1-dependent ROS production through NADPH oxidase, which in turn leads to NIK activation through the inhibition of protein phosphatases and the recruitment of NIK⅐TRAF6 complexes to the IL-1 receptor.