Hydrogen Peroxide Signaling through Tumor Necrosis Factor Receptor 1 Leads to Selective Activation of c-Jun N-terminal Kinase*

Binding of tumor necrosis factor-α (TNFα) to its receptor, TNF-R1, results in the activation of inhibitor of κB kinase (IKK) and c-Jun N-terminal kinase (JNK) pathways that are coordinately regulated and important in survival and death. We demonstrated previously that in response to hydrogen peroxide (H2O2), the ability of TNFα to activate IKK in mouse lung epithelial cells (C10) was inhibited and that H2O2 alone was sufficient to activate JNK and induce cell death. In the current study, we investigated the involvement of TNF-R1 in H2O2-induced JNK activation. In lung fibroblasts from TNF-R1-deficient mice the ability of H2O2 to activate JNK was inhibited compared with fibroblasts from control mice. Additionally, in C10 cells expressing a mutant form of TNF-R1, H2O2-induced JNK activation was also inhibited. Immunoprecipitation of TNF-R1 revealed that in response to H2O2, the adapter proteins, TRADD and TRAF2, and JNK were recruited to the receptor. However, expression of the adaptor protein RIP, which is essential for IKK activation by TNFα, was decreased in cells exposed to H2O2, and its chaperone Hsp90 was cleaved. Furthermore, data demonstrating that expression of TRAF2 was not affected by H2O2 and that overexpression of TRAF2 was sufficient to activate JNK provide an explanation for the inability of H2O2 to activate IKK and for the selective activation of JNK by H2O2. Our data demonstrate that oxidative stress interferes with IKK activation while promoting JNK signaling, creating a signaling imbalance that may favor apoptosis.

The lung is an important target for oxidant injury as a consequence of direct inhalation of oxidants or as a result of the production of oxidants during inflammation (1,2). Although oxidants contribute to tissue damage, these species are also formed in virtually every cell type and are an integral part of normal cell function (3). Oxidants are required for proliferation (4 -6), changes in cellular shape (7) and are involved in transcriptional regulation (2). Despite the emerging role of redox signaling in cellular physiology, the exact mechanisms by which oxidants act as signaling molecules are under intense investigation, and many critical targets remain enigmatic.
Oxidants have been demonstrated to regulate the activation of c-Jun N-terminal kinase (JNK), 1 as well as the transcription factor NF-B (2). JNK is a member of the family of mitogenactivated protein kinases, which is well known to be activated by oxidants and a variety of other stresses in many cell types, including lung epithelial cells (1,8,9). The contribution of JNK to many phenotypic outcomes, including survival (10 -13) and apoptosis (14,15), appears to depend upon the cell type, stimulus, the duration of JNK activation as well as the engagement of other signaling modules (13). In this regard, oxidant-induced JNK activation has been linked to apoptosis (16,17). Although oxidants have also been implicated in the activation of the transcription factor NF-B, currently a number of significant controversies exist around the role of redox events in NF-B activation (18,19). Under basal conditions NF-B is sequestered in the cytoplasm through binding to the inhibitor of B (IB). Upon phosphorylation of IB by IB kinase (IKK), IB is rapidly degraded via the 26 S proteasome, allowing NF-B to translocate to the nucleus and activate the transcription of over 100 genes, including genes critical to cell survival (20 -22).
The signaling events that are required for the activation of JNK and NF-B have been investigated in great detail using the ligand tumor necrosis factor-␣ (TNF␣). TNF␣ binds as a trimer to three TNF-R1 monomers causing aggregation of the intracellular death domains. The death domain (DD) containing-protein, TRADD, binds directly to the DD of TNF-R1 and then recruits the adaptor molecule (23), TRAF2. TRAF2 is essential for JNK activation as well as the recruitment of the IKK complex to the receptor (24). Receptor-interacting protein (RIP) is recruited to TNF-R1 through interaction with TRADD and is essential for the activation of IKK (24). Additionally, TRADD can directly interact with FADD, which is involved in executing caspase-dependent cell death. Recently it has been demonstrated that activation of IKK and JNK following stimulation of TNF-R1 is coordinately regulated and that the activation of NF-B regulates the extent and duration of JNK activation. Transcription of NF-B-driven anti-apoptotic genes such as X-IAP is critical in preventing sustained activation of JNK and also in promoting survival (25,26). Additionally, transient activation of JNK can also mediate survival signaling via JunD, which collaborates with NF-B to increase the expression of the survival gene cIAP-2 (13). Conversely, when NF-B-dependent gene transcription is prevented, JNK activa-tion is prolonged allowing the execution of TNF-induced apoptosis (25,26).
Previously we demonstrated that the oxidant hydrogen peroxide (H 2 O 2 ) inhibits IKK, activates JNK, and causes apoptosis (1,8). In the present study we have investigated the role of TNF-R1 in the activation of JNK by H 2 O 2 . We demonstrate here that H 2 O 2 signals to JNK via TNF-R1 and TRAF2 and that IKK and NF-B activation are prevented as a result of degradation of the adaptor protein, RIP, and cleavage of its chaperone, Hsp90.

EXPERIMENTAL PROCEDURES
Cell Culture and Reagents-A line of spontaneously transformed mouse alveolar type II epithelial cells (C10) was propagated in CRML-1066 medium containing 50 units/ml penicillin, 50 g/ml streptomycin, 2 mM L-glutamine, and 10% fetal bovine serum, all from Invitrogen. Murine recombinant TNF␣ was purchased from Calbiochem. The JNK, IKK, TRAF2, TRADD and Hsp90 antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA), and TNF-R1 antibody was obtained from R&D Systems. Primary lung fibroblasts were isolated from C57BL/6 or TNF-R1Ϫ/Ϫ mice (Jackson Laboratories, Bar Harbor, ME) by mincing lungs and propagation of explants in Dulbecco's modified Eagle's medium/F12 medium containing 50 units/ml penicillin, 50 g/ml streptomycin, 2 mM L-glutamine, and 10% fetal bovine serum, all from Invitrogen. p60⌬ cytoplasmic domain plasmid was provided by Dr. Michael Lenardo (National Institutes of Health, Bethesda, MD) and Flag-TRAF2 by Dr. Brian Seed (Massachusetts General Hospital, Boston, MA). NF-B-luciferase and IB-SR plasmids were provided by Dr. Patrick Baeuerle (Micromet, Martinsreid, Germany). CA-IKK was a kind gift from Dr. Michael Karin (University of California, San Diego), and glutathione S-transferase (GST)-IB was provided by Dr. Rosa Ten (Mayo Clinic, Rochester, MN).
Kinase Assays-Cells were exposed to the test agents and, at the indicated times, transferred to ice, washed once with cold PBS, and lysed as described elsewhere (8). Lysates were cleared by centrifugation at 14,000 rpm, 4°C for 10 min. Protein concentrations were determined, and IKK or JNK was immunoprecipitated from 200 g of protein with an IKK␥, JNK, or HA antibody (Santa Cruz) at 4°C for 1.5 h using protein G-agarose beads (Invitrogen). For TNF-R1 immunoprecipitation and JNK kinase assay, TNF-R1 was immunoprecipitated from 700 g of protein using an anti-TNF-R1 antibody from R&D Systems. Precipitates were washed twice with lysis buffer and once with kinase buffer (8). The kinase reaction was performed using 1 g of GST-IB␣ or GST-c-Jun as a substrate and 5 Ci of [␥-32 P]adenosine triphosphate at 30°C for 30 min. Reactions were stopped by the addition of 2ϫ Laemmli sample buffer. Samples were boiled and stored at Ϫ20°C. Proteins were separated on a 15% polyacrylamide gel. Gels were dried and examined by autoradiography.
Transfection-Cells were transfected using LipofectAMINE-Plus (Invitrogen) according to the manufacturer's directions.
Immunoprecipitation and Western Blotting-A fraction of lysates for analysis of adaptor protein expression or co-immunoprecipitation was added to 2ϫ Laemmli sample buffer, boiled, and loaded on a 10% polyacrylamide gel. Proteins were transferred to nitrocellulose (Schleicher & Shuell), and membranes were subsequently blocked in 5% milk in tris-buffered saline (TBS). Levels of HA were detected with a monoclonal antibody (12CA5, Roche Applied Science). All other proteins were detected with antibodies from Santa Cruz according to the following protocol. Membranes blocked overnight in TBS/milk were washed two times for 15 min in TBS containing 0.05% Tween 20 and incubated with the primary antibody for 1 h at 4°C. Membranes were washed three times for 20 min in TBS/Tween 20 and incubated with a peroxidase-conjugated secondary antibody (Vector Laboratories, Burlingame, CA) for 1 h at room temperature. After a 30-min wash with TBS/Tween 20, conjugated peroxidase was detected by ECL according to the manufacturer's instructions (Amersham Biosciences). For immunoprecipitations, cells were grown to confluence in 100-mm dishes, washed three times with PBS, and treated with test agents in PBS. Cells were lysed in immunoprecipitation buffer (50 mM Hepes, pH 7.4, 250 mM NaCl, 0.1% Nonidet P-40, 5 mM EDTA. 0.5 mM phenylmethylsulfonyl fluoride, 1% aprotinin), and lysates were incubated with TNF-R1 antibody for 2 h with rocking at 4°C. After incubation with antibody, protein G-agarose beads (Invitrogen) were added for 1 h. Precipitates were washed three times with lysis buffer, and 2ϫ Laemmli was added followed by boiling, loading onto a 10% polyacrylamide gel, and Western analysis.
Assessment of NF-B Transcriptional Activity-C10 cells were transiently transfected with a 6 B-tk-luc plasmid containing 6 NFB DNA elements and either TRAF2 or pCDNA3. Cells were treated with 1 ng/ml TNF for 4 h. Cells were lysed in Luciferase Assay Lysis Buffer (Promega, Madison, WI), and a luciferase assay was performed as previously described (8).

H 2 O 2 Activates JNK in a TNF-R1-dependent Manner-We
demonstrated previously that H 2 O 2 causes activation of JNK in lung epithelial cells while inhibiting TNF␣-induced activation of IKK (8). To directly demonstrate that H 2 O 2 -induced JNK signaling is mediated by TNF-R1, lung fibroblasts were isolated from TNF-R1 Ϫ/Ϫ mice, and JNK activation was assessed. The results shown in Fig. 1A demonstrate that the ability of H 2 O 2 to activate JNK was substantially decreased in TNF-R1 Ϫ/Ϫ cells compared with wild type controls at time points ranging from 15 min to 2 h. As expected, TNF␣-induced activation of JNK was also abrogated in TNF-R1 Ϫ/Ϫ fibroblasts. Additionally, overexpression of a truncated form of TNF-R1 (p60⌬CD), which lacks the intracellular death domain in C10 cells also resulted in inhibition of the TNF␣-or H 2 O 2induced activation of JNK (Fig. 1B). These results demonstrate that H 2 O 2 -induced activation of JNK requires TNF-R1.
Activation of JNK and IKK following TNF-R1 activation is coordinately regulated; the duration of JNK activation depends on the activity of NF-B-dependent gene products such as XIAP, which associates with TNF-R1 to repress sustained JNK activity (25,26). In agreement with these findings, inhibition of NF-B activity via the expression of a dominant acting version of IB, IB-SR, led to an augmented and sustained JNK activ- ity profile after exposure to TNF␣, illustrating the repression of JNK by NF-B activation downstream of TNF-R1 in lung epithelial cells ( Fig. 2A). To further support a role of TNF-R1 in JNK activation by H 2 O 2 , we addressed whether JNK activation by H 2 O 2 could also be regulated by the NF-B pathway. We therefore expressed a constitutively active version of IKK␤ (CA-IKK␤), and assessed the ability of H 2 O 2 or TNF␣ to activate JNK. Compared with vector controls, CA-IKK␤-expressing cells displayed attenuated JNK activation following treatment with H 2 O 2 or TNF␣ (Fig. 2B), illustrating that JNK activation by H 2 O 2 is also subject to negative regulation by NF-B, analogous to JNK activation by TNF␣.
H 2 O 2 Causes Recruitment of Adapter Proteins to TNF-R1-TNF-R1 is catalytically inactive and requires the recruitment of adaptor proteins to its intracellular domain in order to transmit signals (28). Upon ligand binding, the intracellular domain aggregates and the DD-containing protein, TRADD, is recruited to and binds to the DD of the receptor (23). TRADD forms a platform for TRAF2 and RIP (29), which are required for JNK activation and the recruitment and activation of IKK (24). We therefore investigated whether H 2 O 2 also causes recruitment of TRADD and TRAF2 to TNF-R1. Endogenous TNF-R1 was immunoprecipitated, and TRADD and TRAF2 co-immunoprecipitation was evaluated by Western blotting. As shown in Fig. 3A, TRADD co-precipitated with TNF-R1 in unstimulated cells and increased in response to TNF␣ or H 2 O 2 . Furthermore, in samples from TNF␣-or H 2 O 2 -exposed cells post-immunoprecipitation, TRADD levels were decreased (Fig.  3B). TRAF2 also co-immunoprecipitated with TNF-R1 in control cells, increased following exposure to TNF␣ or H 2 O 2 (Fig.  3C), and decreased in the post-immunoprecipitate lysates (Fig. 3D).
Because it has been demonstrated that IKK is recruited to TNF-R1 (24), we next assessed whether JNK could also be recruited to TNF-R1 after incubation with TNF␣ or H 2 O 2 . As shown in Fig. 3E (30) to the TNF-R1 DD. Hsp90 and RIP form a complex; destabilization of Hsp90 leads to RIP degradation and consequently prevents the activation of IKK and NF-B (31). We therefore investigated whether the lack of IKK activation in cells treated with H 2 O 2 was due to destabilization of RIP and/or Hsp90 and assessed the levels of TRAF2, RIP, and Hsp90 by Western blot analysis. As seen in Fig. 4A, TRAF2 levels were not affected by H 2 O 2 or TNF␣ in a time frame of up to 2 h. Evaluation of Hsp90 revealed that a lower molecular weight species of Hsp90 was present after treatment with H 2 O 2 for 2 h (Fig. 4B) and first appeared after 30 min of exposure (data not shown). This Hsp90 cleavage product did not appear in cells treated with TNF␣ (Fig. 4B). In agreement with these findings, RIP levels were markedly decreased in cells treated with H 2 O 2 for 2 h, whereas TNF␣ exposure did not affect RIP expression compared with sham controls (Fig. 4C, and data not shown). Because RIP and Hsp90 are required for IKK activation, their destabilization by H 2 O 2 provides a plausible explanation for the inability of H 2 O 2 to activate IKK.
Overexpression of TRAF2 Activates JNK and Inhibits the Ability of TNF␣ to Activate IKK and NF-B-The results presented above demonstrate that in response to H 2 O 2 , TRADD and TRAF2 are recruited to the TNF-R1, whereas RIP and Hsp90, which are essential for IKK activation, are destabilized. The absence of RIP at TNF-R1 has been demonstrated to enhance recruitment of TRADD and TRAF2, suggesting a competition for binding at TNF-R1 between TRADD/TRAF2 and RIP (24). Therefore, an enhanced presence of TRAF2 at TNF-R1 in response to H 2 O 2 may sustain JNK activation while inhibiting IKK. To confirm that the effects of H 2 O 2 in C10 cells can be mimicked by TRAF2 accumulation at the receptor, we overexpressed wild type TRAF2 and analyzed the activity of IKK, NF-B, and JNK. TRAF2 overexpression was sufficient to inhibit the ability of TNF␣ to activate IKK and NF-B (Fig. 5,  A and B). Moreover, TRAF2 overexpression also led to activation of JNK under base-line conditions (Fig. 5C). Lastly, we determined whether H 2 O 2 -induced degradation of RIP and cleavage of Hsp90 would alter JNK and IKK activation in response to TNF␣. As expected, pretreatment of C10 cells with H 2 O 2 for 2 h, the time point associated with RIP degradation and Hsp90 cleavage (Fig. 4), led to a complete inhibition of IKK activation by TNF␣. Although under these conditions JNK activation by H 2 O 2 was no longer observed, pretreatment with H 2 O 2 enhanced JNK activation in response to TNF␣ (Fig. 6). These data confirm that H 2 O 2 causes a signaling imbalance at TNF-R1, leading to preferential activation of JNK, while inhibiting IKK. dependent manner and, consequently that H 2 O 2 -induced JNK is substantially decreased in cells lacking TNF-R1 or in cells expressing a truncated version of TNF-R1 lacking the intracellular death domain. We also provided evidence for the presence of TRAF2 and TRADD at TNF-R1 under base-line conditions, in agreement with recent reports demonstrating that adaptor proteins can be present in nonstimulated cells (36). We were only able to detect small changes in the recruitment of endogenously expressed adaptor proteins to TNF-R1 after stimulation of the lung epithelial cells examined here, in agreement with findings by others (19) but in contrast to reports that use expression cassettes or immune cells (24). Interestingly, analogous to observations demonstrating the presence of IKK at TNF-R1, we provide evidence here that JNK can also be recruited to and is active at TNF-R1.
The H 2 O 2 -induced destabilization of RIP and cleavage of its chaperone Hsp90, both of which are essential in the activation of IKK, provide a plausible explanation for the selective activation of JNK, but not IKK, by H 2 O 2 . A similar disruption of adaptor protein recruitment to the death domain of TNF-R1 has been demonstrated in response to geldanamycin (GA), a disruptor of Hsp90, which leads to the destabilization of RIP and consequently inhibits the activation of IKK and NF-B by TNF␣ (30,31). In agreement with our observations on H 2 O 2 , GA did not affect the stability of TRAF2 nor did it affect JNK activation. GA-induced RIP degradation involves the proteasome and appears to occur in a caspase-and lysosome-independ-ent manner (31). It remains to be determined whether these events are also involved in RIP degradation by H 2 O 2 . It is of interest to note that GA brings about a disruption of endothelial nitric-oxide synthase, causing it to produce superoxide, the precursor of H 2 O 2 (37). Thus, it is conceivable that H 2 O 2 production may also contribute to RIP degradation and NF-B repression after GA treatment (31), although this possibility remains to be tested formally.
In cells overexpressing TRAF2, the ability of TNF␣ to activate IKK and NF-B was inhibited, whereas JNK was constitutively active. In contrast to our observations, others have shown that TRAF2 overexpression leads to NF-B activation (22, 38 -40). The discrepancy of their findings compared with our study may stem from the cell type and experimental conditions used. For example, it is possible that IKK activation by TRAF2 was transient, as opposed to JNK, and that we missed its elevated activity, or that the levels of TRAF2 expression necessary to activate IKK and NF-B are higher than those required for JNK activation. Alternatively, it is possible that expression levels and/or the localization of TNF-R family members (40), RIP, Hsp90, IKK␥, and MEKK1 (41), can sustain NF-B activation in response to TRAF2 overexpression in some, but not all, cell types. Findings demonstrating that in RIPϪ/Ϫ cells additional TRAF2 recruitment to TNF-R1 occurs (24) support the notion that a critical balance of RIP and TRAF2 at TNF-R1 is required to activate NF-B. In agreement with these findings, the H 2 O 2 -induced destabilization of RIP and Hsp90 shown in the present study was also associated with the enhanced presence of TRAF2 at TNF-R1, as well as with prolonged activation of JNK.
Sustained JNK activation by H 2 O 2 is not observed in response to TNF␣ because of the NF-B-dependent expression of anti-apoptotic genes, which are recruited to TNF-R1 and inhibit prolonged JNK activity and TNF␣-induced apoptosis (21,26). Conversely, overexpression of a constitutively active form of IKK resulted in an attenuation of H 2 O 2 -induced JNK activation, demonstrating that JNK activation by H 2 O 2 is also subject to negative feedback regulation by NF-B. It will be of interest to determine whether diminution of JNK activation as a result of NF-B activation also represses oxidant-induced apoptosis, analogous to observations with TNF␣.
Although we recently demonstrated that H 2 O 2 is capable of directly oxidizing IKK leading to its inactivation (8)  TNF␣-induced NF-B activation upstream by interfering with the formation of the TNF-R1 signaling complex. It is plausible that H 2 O 2 can affect binding of TNF␣ to its receptor or can promote its shedding. Recent observations demonstrate that redox events affect the binding of TNF␣ to TNF-R1 (19), and H 2 O 2 -induced shedding of soluble TNF-R1 has been demonstrated in lung epithelial cells (42). However, it is unlikely that these events have contributed to our present observations, as we have demonstrated that JNK activation by TNF␣ is enhanced in cells pretreated with H 2 O 2 (Fig. 6), illustrating that TNF␣ is still able to induce signaling. The sites of H 2 O 2 -induced oxidation relevant to adaptor protein recruitment and degradation remain elusive. Cysteines present in the extracellular domain of TNF-R1 occur as disulfide bridges (43) and therefore are unlikely targets for oxidation by H 2 O 2 . However, cysteines contained in the intracellular DD may be prone to oxidation by H 2 O 2 and could promote receptor clustering and adaptor protein recruitment. Furthermore, oxidant-induced dissociation of thioredoxin from apoptosis signal-regulating kinase-1 promotes its activation and binding to TRAF2 (44), providing an additional explanation by which H 2 O 2 promotes TNF-R1-dependent signaling to JNK. It is not known at this time whether H 2 O 2 acting through TNF-R1 is ligand-independent; however failure to activate IKK strongly suggests that H 2 O 2 affects TNF-R1 function in a manner distinct from TNF␣.
The H 2 O 2 -induced signaling imbalance at TNF-R1 may have important ramifications for the cellular fate. It is well known that oxidant-induced JNK activation can mediate apoptosis (16,17). It is of interest to note that TNF␣ itself can alter the redox potential of the cell and uses oxidants to signal (for review, see Ref. 27). It is conceivable that the extent of oxidant production, including H 2 O 2, dictates whether TNF-R1 activation by TNF␣ will lead to cell death or survival. Our present findings point to a putative proapoptotic role for oxidants under inflammatory conditions by causing direct signaling through TNF-R1, in addition to promoting TNF␣-induced apoptosis via c-Jun N-terminal kinase.