Reactive Oxygen Species Mediate Cytokine Activation of c-Jun NH 2 -terminal Kinases*

1 and (TNF (cid:97) ) are known to induce production of reactive oxygen species (ROS), which have been suggested to act as second messengers. Here we demonstrate that ROS production by bovine chondrocytes upon cytokine stimula- tion induces c- jun expression. Since c- jun expression is regulated by its own gene product via phosphorylation by c-Jun NH 2 -terminal kinases (JNKs), we investigated if cytokines and ROS could modulate JNK activity in chondrocyte monolayer cultures. Treatment of bovine chondrocytes with both IL-1 and TNF (cid:97) leads to rapid induction of JNK activity, stimulating JNK activity 7- and 20-fold, respectively. Importantly, the observation that antioxidant treatment antagonizes IL-1 and TNF (cid:97) activation of JNK provides strong evidence that ROS can act as mediators of JNK activity. Moreover, potent activation of JNK is also observed by direct addition of the ROS hydrogen peroxide (H 2 O 2 ) to the chondrocyte cultures. Nitric oxide (NO), a multifunctional ROS, also appears to simulate JNK, albeit to a lesser extent. These findings identify JNK as another molecular target for the actions of NO and H 2 O 2 . In addition, the inhibitory effect of diphenyleneiodonium on JNK activation impli- cates then re- suspended in 30 (cid:109) l of kinase buffer (20 m M HEPES, pH 7.6, 20 m M MgCl 2 , 2 m M dithiothreitol, 25 m M (cid:98) -glycerophosphate, and 0.1 m M Na 3 VO 4 ) containing 20 (cid:109) M ATP and 5 (cid:109) Ci of [ (cid:103) - 32 P]ATP. After being incubated at 30 °C for 20 min, the reaction was terminated by boiling in Laemmli buffer. Samples containing phosphorylated proteins were first resolved by SDS-PAGE, followed by drying, autoradiography, and quantitation by phosphorimaging analysis. Solid-phase In-gel Kinase Assay— After c-Jun-binding proteins were isolated using glutathione-Sepharose beads prebound with GST-c-Jun as described above, proteins were eluted in Laemmli buffer. The eluted proteins were then resolved on a 10% SDS-polyacrylamide gel, which was polymerized in the presence of 40 (cid:109) g/ml GST-c-Jun or GST. After electrophoresis, the gel was washed twice with 20% isopropanol and 50 m M Tris-HCl (pH 7.9) for 30 min. The gel was subsequently washed twice for 30 min with 5 m M (cid:98) -mercaptoethanol and 50 m M Tris-HCl, pH 7.9. The gel was then incubated in 6 M guanidine HCl, 5 m M (cid:98) -mercap-toethanol, 2 m M EDTA, and 50 m M Tris-HCl, pH 7.9, for 1 h, followed by incubation at 4 °C overnight with a few changes of 0.04% Tween 20, 5 m M (cid:98) -mercaptoethanol, 2 m M EDTA, and 50 m M Tris-HCl, pH 7.9. For kinase reaction, the gel was incubated in kinase buffer containing 40 (cid:109) M ATP and 5 (cid:109) Ci of [ (cid:103) - 32

Interleukin 1 (IL-1) 1 and tumor necrosis factor ␣ (TNF␣) are multifunctional cytokines involved in inflammation, cell growth, and apoptosis (1)(2)(3)(4). The pleiotropic effects of these two cytokines are mediated through distinct cell surface receptors (2,3,5). Despite the incomplete understanding of the membrane signaling events following the occupancy of the cytokine receptors, ligand interaction is known to stimulate the release of a wide variety of putative second messengers. Some of the known signal transduction pathways common to both IL-1 and TNF␣ include coupling to G-proteins (6,7), activation of phospholipase A 2 (8,9), calcium mobilization (10,11), and ceramide production (12,13).
In addition to the above messengers, a class of highly diffusible and ubiquitous molecules termed reactive oxygen species (ROS) has recently been recognized to act as signaling intermediates for cytokines, including IL-1 and TNF␣ (14 -16). ROS encompass species such as superoxide, hydrogen peroxide (H 2 O 2 ), nitric oxide (NO), and hydroxyl radicals (17). These highly reactive molecules are known to regulate many important cellular events, including gene expression (16,18,19), transcription factor activation (20), DNA synthesis (21), and cellular proliferation (22).
The signal transduction cascades elicited after exposure to IL-1 and TNF␣ culminate in a nuclear response characterized by the activation of several key transcriptional regulators, including nuclear factor B (20, 23) and AP-1 (16,24). Hence, genes with an AP-1 binding site, such as those encoding the metalloproteinases collagenase and stromelysin, are potential targets for the two cytokines (25,26). AP-1 is composed of the protein products of c-fos and c-jun; their levels of gene expression are also stimulated by cytokines (24). Posttranslational modification of AP-1 transcriptional activity also plays an important role in the control of AP-1-regulated gene expression. Such control is executed in the form of phosphorylation of Fos and Jun proteins by members of the mitogen-activated protein kinase family (27)(28)(29)(30).
The transcriptional activity of the c-Jun protein is greatly enhanced by phosphorylation of two serine residues at positions 63 and 73 in its activation domain (31,32). Phosphorylation of the NH 2 terminus of c-Jun is catalyzed by c-Jun NH 2terminal kinases (JNKs), a subgroup of the mitogen-activated protein kinase family (28 -30). The properties of JNKs that distinguish them from other mitogen-activated protein kinases, such as extracellular signal-related kinases, include their activation by agents such as ultraviolet irradiation, heat shock, protein synthesis inhibitors, and inflammatory cytokines (33,34). Since many of the JNK activators can be regarded as cellular stress, JNKs have also been termed stressactivated protein kinases (30). Inflammatory cytokines, such as TNF␣ and IL-1, have been shown to induce JNKs (35) and c-jun expression (2,24,26,36); however, the second messengers responsible for their activation remain unidentified. Our earlier study demonstrated ROS production upon cytokine stimulation in bovine chondrocytes (16); 2 thus, we examined the molecular links between cytokine-induced ROS generation and its downstream signaling events.

MATERIALS AND METHODS
Reagents-Human recombinant IL-1␤ was generously supplied by Ciba-Geigy, Basel, Switzerland. Human recombinant TNF␣ was from Genzyme Corp. H 2 O 2 was from Fisher Scientific. N-Acetylcysteine (NAC) and ascorbic acid (Asc) were from Sigma. S-Nitroso-Nacetylpenicillamine (SNAP) was purchased from Biomol Research Laboratories. Diphenyleneiodonium (DPI) was from Toronto Research Chemicals. Radioactive isotopes and x-ray films were from DuPont NEN. Glutathione-Sepharose was obtained from Pharmacia Biotech. Purified anti-human p46 JNK1 monoclonal antibody was purchased from PharMingen. This antibody recognizes both the M r 46,000 JNK1 and the related M r 55,000 JNK2 proteins.
Cell Culture-Primary cultures of bovine articular chondrocytes were isolated from bovine articular cartilage as described by Cruz et al. (37). Bovine chondrocytes were plated at 2 ϫ 10 6 cells/ml in 12 ml of Ham's F-12 media containing 3% antibiotics and 5% fetal bovine serum. The cells were allowed to recover for 24 h at 37°C in a humidified atmosphere supplemented with 5% CO 2 . Prior to any treatments, chondrocytes were serum-starved overnight.
Northern Blot Analysis-Total RNA was isolated by the acidified guanidine isothiocyanate method (38) and quantitated by spectrophotometry at 260 nm. Denatured RNA samples (10 -15 g) were analyzed by gel electrophoresis in a 1% denaturing agarose gel, transferred to a nylon membrane (Bio-Rad), and cross-linked with an ultraviolet crosslinker (Stratagene UV Stratalinker 1800). The blots were hybridized with 32 P-labeled human c-jun cDNA, subsequently stripped, and reprobed with 32 P-labeled rat glyceraldehyde phosphate dehydrogenase (GAPDH) cDNA as internal control.
Whole Cell Extract (WCE) Kinase Assay-Following treatment of chondrocytes with various agents, cells were washed twice with phosphate-buffered saline and harvested according to published procedures by Hibi et al. (29). Briefly, the cells were suspended in WCE buffer and rotated at 4°C for 30 min. The supernatant was collected as WCE after being centrifuged at 10,000 ϫ g for 10 min. The protein concentration in the cell lysate was determined by the Bio-Rad protein assay. Ten g of WCE were incubated at 30°C for 20 min in the presence of 40 M ATP, 5 Ci of [␥-32 P]ATP, and 20 g of glutathione S-transferase (GST) or GST-c-Jun (5-89) as substrate. The reaction was terminated by the addition of Laemmli buffer and then boiled for 5 min. The proteins were then separated by SDS-PAGE, followed by drying and autoradiography. Phosphorylation signals were quantitated by PhosphorImager (Molecular Dynamics, Inc.).
Solid-phase (S-P) Kinase Assay-WCE was diluted according to the procedures described by Hibi et al. (29). Extract containing 100 -300 g of protein was added to 10 l of glutathione-Sepharose suspension prebound with 20 g of GST-c-Jun or GST. The mixture was rotated at 4°C for 3 h and centrifuged at 10,000 ϫ g for 20 s. The pelleted glutathione beads were washed with buffer (20 mM HEPES, pH 7.7, 50 mM NaCl, 2.5 mM MgCl 2 , 0.1 mM EDTA, 0.05% Triton X-100, 0.5 g/ml leupeptin, and 100 g/ml phenylmethylsulfonyl fluoride) and then resuspended in 30 l of kinase buffer (20 mM HEPES, pH 7.6, 20 mM MgCl 2 , 2 mM dithiothreitol, 25 mM ␤-glycerophosphate, and 0.1 mM Na 3 VO 4 ) containing 20 M ATP and 5 Ci of [␥-32 P]ATP. After being incubated at 30°C for 20 min, the reaction was terminated by boiling in Laemmli buffer. Samples containing phosphorylated proteins were first resolved by SDS-PAGE, followed by drying, autoradiography, and quantitation by phosphorimaging analysis.
Solid-phase In-gel Kinase Assay-After c-Jun-binding proteins were isolated using glutathione-Sepharose beads prebound with GST-c-Jun as described above, proteins were eluted in Laemmli buffer. The eluted proteins were then resolved on a 10% SDS-polyacrylamide gel, which was polymerized in the presence of 40 g/ml GST-c-Jun or GST. After electrophoresis, the gel was washed twice with 20% isopropanol and 50 mM Tris-HCl (pH 7.9) for 30 min. The gel was subsequently washed twice for 30 min with 5 mM ␤-mercaptoethanol and 50 mM Tris-HCl, pH 7.9. The gel was then incubated in 6 M guanidine HCl, 5 mM ␤-mercaptoethanol, 2 mM EDTA, and 50 mM Tris-HCl, pH 7.9, for 1 h, followed by incubation at 4°C overnight with a few changes of 0.04% Tween 20, 5 mM ␤-mercaptoethanol, 2 mM EDTA, and 50 mM Tris-HCl, pH 7.9. For kinase reaction, the gel was incubated in kinase buffer containing 40 M ATP and 5 Ci of [␥-32 P]ATP at 30°C for 1 h. Then the gel was washed extensively with 5% trichloroacetic acid and 1% sodium pyrophosphate. Finally, the gel was vacuum-dried and subjected to autoradiography.
Western Blot Analysis-Proteins specifically associated with c-Jun were isolated using glutathione-Sepharose beads prebounded with GST-c-Jun as described under "Solid-phase (S-P) Kinase Assay." Eluted proteins were resolved by SDS-PAGE, followed by electroblotting onto Immobilon-P membrane (Millipore Corp.), which was then probed with monoclonal antibodies against p46 JNK1 . The antigen-antibody complexes were visualized by the enhanced chemiluminescence detection system (Amersham Corp.).

RESULTS AND DISCUSSION
In our previous studies, we have implicated the involvement of ROS in TNF␣and IL-1-mediated nuclear response, such as the induction of c-fos gene expression (16). 2 Not only were antioxidants shown to inhibit cytokine induction of c-fos, we and others have also demonstrated enhancement of c-fos expression by exogenous addition of H 2 O 2 in bovine chondrocytes, HeLa cells, and osteoblasts (16,18,19). In view of this, we decided to examine the effects of IL-1, TNF␣, and ROS on another early response gene c-jun, of which the protein product heterodimerizes with Fos to form the AP-1 transcription factor (24). Time course analyses showed that IL-1 and TNF␣ caused a transient increase in the steady-state c-jun mRNA levels with optimal expression at around 30 min (Fig. 1A). To assess the relevance of ROS as putative second messengers in the induction of c-jun by IL-1 and TNF␣, we tested if endogenous ROS were required in cytokine-mediated c-jun regulation. To do this, chondrocytes were first treated with the antioxidants, NAC and Asc, before the addition of cytokines. Both NAC and Asc are effective free radical scavengers; the former is known to increase intracellular glutathione levels, which in turn modulate the concentration of ROS via glutathione peroxidase (17). On the other hand, Asc itself is highly reactive toward radicals and has proven to be a versatile scavenger (39). NAC and Asc, which neutralize the activities of ROS, reduced both IL-1-and TNF␣-induced c-jun mRNA levels (Fig. 1B), implicating ROS in the signaling cascade leading to c-jun induction.
Next we asked whether exogenous addition of ROS could mimic the effects of IL-1 and TNF␣ with respect to induction of c-jun. Hydrogen peroxide, a membrane-permeable reagent physiologically produced in large amounts by cells such as granulocytes (17), has been widely used to assess the effects of ROS. Hydrogen peroxide was found to activate c-jun gene expression with induction first apparent at 30 min (Fig. 1C), whereas the expression of the housekeeping gene GAPDH was not affected. The induction of c-jun mRNA expression by H 2 O 2 was transient and is similar to the c-jun induction profiles obtained with TNF␣ and IL-1. These data provide further evidence that ROS can indeed act as second messengers in signaling c-jun expression in chondrocyte cultures.
The findings that ROS are involved in inducing expression of c-jun prompted us to investigate the underlying mechanisms. Interestingly, the c-jun promoter contains two AP-1-like sequences termed JUN1 and JUN2 (18,40,41). Each of these sites is bound by a heterodimer of c-Jun in association with another transcription factor, ATF2. Hence, the c-jun promoter is autoregulated by its own gene product (40,41). Activation of the c-jun promoter involves phosphorylation of the prebound c-Jun-ATF2 heterodimers by JNKs/stress-activated protein kinases, which are members of the mitogen-activated protein kinase family (28 -30). Phosphorylation of serine-63 and serine-73 in the activation domain of c-Jun increases its activation potential (31,32). These findings, coupled with our analyses of c-jun induction described above, led us to hypothesize that JNKs in chondrocytes might also be regulated by cytokines with ROS as activating intermediates. WCE and S-P assays were used to examine JNK activity. Following the treatment of chondrocytes with various agents, a whole cell lysate was prepared by lysis with detergent. This extract could be assayed directly with a GST-c-Jun fusion protein as substrate in the WCE assay. Alternatively, in the S-P assay, the extract was first incubated with the GST-c-Jun fusion protein to allow binding of endogenous JNKs. Then, unbound proteins were washed off, and kinase activity was analyzed in the presence of radioisotopes. The GST-c-Jun fusion protein used to bind the c-Jun-specific kinases now acted as a substrate in the kinase reaction. We first looked at the effects of cytokines on JNK activity. Treatment with IL-1 and TNF␣ potently activated JNK activity, as measured by phosphorylation of the substrate GST-c-Jun in the WCE assay ( Fig. 2A). Phosphorylation was observed as early as 15 min and gradually decreased over the course of 2 h. Phosphorimaging analysis showed that maximum fold-induction of JNK activities by IL-1 and TNF␣ were 7and 20-fold, respectively. The S-P assay in which c-Jun-specific kinases were first bound to the substrate gave essentially the same results. This indicated that the phosphorylation signals seen in the "whole cell extract" assay were indeed from c-Junspecific kinases. As a control, phosphorylation was not ob-served with the GST moiety alone as substrate (data not shown).
We again exploited the abilities of antioxidants to perturb intracellular ROS levels, to assess the role of endogenous ROS in regulating the c-Jun kinases. Antioxidant treatment antagonized the stimulating effects of IL-1 and TNF␣ on c-Jun kinase activity (Fig. 2B), indicating that in vivo ROS production does play a role in modulating JNK activity. Next we tested the effects of two ROS, H 2 O 2 and NO generated by SNAP, on the activity of JNK. The ROS NO has recently gained substantial recognition as a signaling molecule with the property of both neurotransmitter and hormone (42). NO production by many cell types, including chondrocytes, is also significantly enhanced in the presence of cytokines (16,43). 2 Fig. 2C clearly shows that the direct addition of H 2 O 2 dramatically activated JNK. Interestingly, NO also stimulated the kinase activity, albeit to a much lesser extent. Both NO and H 2 O 2 stimulated JNK after 15 min treatment (Fig. 2C). Signal quantitation indicated that the maximum fold-activation of JNK activity by H 2 O 2 was around 8-fold. These data provided the first direct evidence that ROS can modulate the activity of JNK.
One class of enzymes that are known to give rise to various for different time points, and whole cell lysates were isolated for JNK activity determination. SNAP, which evolves NO upon dissolving, was prepared in ethanol immediately before use and was directly added to culture medium. Extracts were used to determine JNK activity by both the WCE assay and the S-P assay as described under "Materials and Methods." The data shown are representative of three independent experiments. types of ROS is the flavonoid-containing enzymes. Therefore, we examined the effect of DPI, a potent inhibitor of flavonoidcontaining enzymes, such as NADPH oxidase and nitric oxide synthase (44), on IL-1 and TNF␣ induction of JNK activity. As with antioxidants, the addition of DPI also attenuated cytokine stimulation of Jun kinase activity (Fig. 3). In bovine chondrocytes, DPI has previously been shown to inhibit IL-1 and TNF␣ stimulation of intracellular ROS production (16). 2 Hence, the above findings are in keeping with a role for ROS production by flavonoid-containing enzymes in the cytokine stimulation of JNK activity.
Since JNK phosphorylation of c-Jun is critical for its transcriptional activity (28 -32) and c-Jun autoregulates its own promoter (40,41), our observation of enhanced JNK activity upon exposure to cytokines is consistent with the cytokine induction of c-jun mRNA levels in chondrocytes. The exact pathway from ROS release to JNK activation has yet to be identified. Our finding that DPI inhibits cytokine stimulation of JNK implicates the involvement of flavonoid-containing enzymes such as NADPH oxidase. This is particularly noteworthy considering that one of the critical components of NADPH oxidase, Rac1, has been shown recently to be an upstream regulator of JNK (45)(46)(47). Perhaps the enhanced activity of the enzyme NADPH oxidase in the presence of Rac1 gives rise to increased ROS levels, which then initiates a kinase cascade culminating in the activation of JNK. The cellular mechanism of ROS sensing and the identity of the direct sensor molecules await further characterization. JNKs themselves do not appear to be the direct targets of ROS because both H 2 O 2 and SNAP could not activate JNKs immobilized onto GST-c-Jun-glutathione beads (data not shown). Nonetheless, intracellular ROS sensing may occur through the redox regulation of sensor molecules. Such form of regulation has been demonstrated for the transcription factors Fos and Jun in which oxidation-reduction of a key cysteine residue modulates DNA binding activity (48). Furthermore, similar redox modification may also underlie direct modulation of p21 ras activity by ROS such as H 2 O 2 and NO (49,50).
Two JNK isoforms (M r 46,000 JNK1 and M r 55,000 JNK2) have been identified in HeLa cells (29); both of them are capable of phosphorylating the Jun activation domain. To distinguish the different isoforms in bovine chondrocytes, we used a S-P in-gel kinase assay to determine the sizes of JNKs being activated upon cytokine stimulation. After the chondrocytes were treated with IL-1 or TNF␣ in the presence or absence of the antioxidant NAC, whole cell extracts were isolated and first incubated with GST-c-Jun-glutathione beads to selectively bind Jun-specific kinases. Then protein kinase activities were examined following SDS-PAGE on gels that were polymerized with GST or GST-c-Jun. Extracts of cells treated with either IL-1 or TNF␣ displayed very efficient stimulation of two kinases, migrating at positions of M r 46,000 and M r 55,000, that phosphorylated GST-c-Jun but not GST in the gel (Fig. 4). The phosphorylation signal observed at position M r 46,000 was much stronger than that at M r 55,000. The apparent molecular weights of the M r 46,000 and the M r 55,000 c-Jun kinases match well with that of JNK1 and JNK2, respectively. Since this S-P in-gel kinase experiment was designed to assay only for kinases that specifically bind GST-c-Jun, a property unique to JNK, we therefore conclude that the two kinases correspond to JNK1 and JNK2. In addition, Fig. 4 also shows that activation of the two c-Jun kinases were markedly inhibited by the antioxidant NAC. This indicates that both JNK isoforms contribute to the ROS-mediated cytokine stimulation of c-Jun kinase activity. Nevertheless, the higher activity associated with the M r 46,000 JNK1 upon cytokine stimulation might imply a larger contribution of this isoform to the total c-Jun kinase activity.
Western blot analysis was also used to definitively identify the c-Jun-associated kinases in primary chondrocytes. We used a purified anti-human JNK1 monoclonal antibody that also cross-reacts with human JNK2, as seen in Fig. 5, lane 1, with HeLa whole cell extract. The anti-human JNK1 antibody reacted with two proteins having identical sizes as human JNK1 and JNK2 in chondrocyte extract (Fig. 5, lanes 1 and 2). Although the JNK2 band was less prominent than the JNK1 band in the chondrocyte extract, lane 2 indicates only that both JNK1 and JNK2 are present in chondrocytes and does not necessarily reflect the relative amounts of each c-Jun kinase. Interestingly, the JNK1 protein band appears as a doublet in extracts stimulated with either IL-1 or TNF␣ (Fig. 5, lanes  3-5). The upper band of the doublet probably corresponds to phosphorylated, and thus activated, form of JNK1 (51). It is Whole cell lysates were prepared by lysis with detergent. Extracts were used to determine JNK activity by both the WCE assay and S-P assay as described under "Materials and Methods." Three separate experiments were performed with similar results. known that unactivated JNKs can associate with c-Jun; however, activation leads to enhanced binding (51). In this experiment, JNK2 was barely detectable by this antibody, probably due to poor cross-reactivity of the antibody toward the bovine counterpart of human JNK2 or lower amount of bound JNK2, as suggested by the S-P in-gel kinase experiment, or both (Fig.  4). The relatively weak interaction of JNK2 with c-Jun seen in our chondrocyte system is also consistent with similar findings in human cells (29). Taken together, this study demonstrates that ROS is a link between cytokine-receptor interaction and stimulation of JNK activity and c-jun expression. ROS have been implicated as second messengers in modulating diverse cellular functions in various cell types including chondrocytes, neutrophils, and vascular smooth muscle cells (16,20,52,53). Here we provide strong evidence that JNK activation may mediate some of the signaling effects of ROS. Significantly, this report also identifies JNK as another molecular target for the versatile molecule, nitric oxide. This may become more important in pathological conditions in which an abnormally high concentration of cytokines leads to increased NO production (42,54). Overstimulation of JNK activity by cytokine-stimulated ROS production (e.g. NO and H 2 O 2 ) may cause deregulated c-jun expression. The c-jun protein product, as a key component of AP-1, may then induce inappropriate transcriptional responses, leading to overexpression of metalloproteinases and uncontrolled cellular proliferation. Thus, abnormal activation of the JNK cascade may underlie disease conditions such as arthritis and cancer, which often are characterized by overproduction of ROS and metalloproteinases (54 -57). After SDS-PAGE, the resolved proteins were transferred to Immobilon-P membrane, blotted with anti-JNK1 monoclonal antibodies, and detected by enhanced chemiluminescence. The experiment was performed twice with essentially identical results.