15-Deoxy-Δ12,14-prostaglandin J2 Regulates Endogenous Cot MAPK Kinase Kinase 1 Activity Induced by Lipopolysaccharide*

Cot is a MAPK kinase kinase that has been implicated in cellular activation and proliferation. Here, we show that the addition of lipopolysaccharide (LPS) to RAW264 macrophages induces a 10-fold increase of endogenous Cot activity, measured as MAPK kinase kinase 1 activity. Taxol, but not phorbol 12-myristate 13-acetate (PMA), induces a similar activation of Cot. A tyrosine kinase activity is involved in Cot activation by LPS. 15-Deoxy-Δ12,14-prostaglandin J2, but not rosiglitazone, blocks Cot activation by LPS. Furthermore, 15-deoxy-Δ12,14-prostaglandin J2 also inhibited the LPS-induced Cot in vitro. However, 15-deoxy-Δ12,14-prostaglandin J2 does not inhibit MAPK kinase 1 or ERK1/ERK2 activation/phosphorylation induced by PMA and mediated by c-Raf. Considering these data, we propose that the inhibition of LPS-induced Cot activation is one mechanism by which 15-deoxy-Δ12,14-prostaglandin J2 acts as an anti-inflammatory.

Lipopolysaccharide (LPS) 1 is a major cell wall component of Gram-negative bacteria that potently activates macrophages, inducing the production and secretion of immunoregulators such as TNF-␣, interleukin-1␤, and arachidonic acid metabolites (for reviews see Refs. [1][2][3]. Deregulation of the secretion of these pro-inflammatory cytokines is associated with septic shock and other inflammatory conditions such as rheumatoid arthritis. When complexed with a serum protein, LPS binds to the molecule CD14 on the surface of macrophages. The CD14 receptor itself lacks a cytosolic domain, yet it appears to interact with the toll-like receptors that, in turn, can transduce the signal across the plasma membrane (1)(2)(3)(4). Indeed, activation of the toll-like receptors triggers an intracellular signaling cascade that involves the adapter proteins MyD 88 and IRAK, and the TRAF6 proteins (for a review see Ref. 5). The LPS signal transduction pathway involves the rapid tyrosine phosphorylation of several proteins and the activation of different mitogen-activated protein kinase (MAPK) cascades, including the classical MAPK pathway (6 -9). However, the precise molecular mechanism by which LPS activates the extracellular signalregulated kinases 1 and 2 (ERK1 and ERK2) in the classic MAPK pathway is still not fully understood.
The Cot/tpl-2 gene encodes a MAPK kinase kinase (MKKK) that in overexpression studies was originally reported to be capable of triggering several MAPK cascades, namely those leading to activation of the MAPKs ERK1/ERK2, c-Jun Nterminal kinase, p38, and ERK5 (10 -15). However, more recent studies have shown that peritoneal macrophages from Cot/tpl-2 knockout mice do not activate ERK1/ERK2 in response to LPS, although the activation of c-Jun N-terminal kinase and p38 MAPK pathways remain the same as in wild type mice (15). This suggests that under normal conditions, Cot/tpl-2 may act more specifically within the classical MAPK cascade than when overexpressed in cells. Thus, Cot/tpl-2 is one of the three proteins that can activate MAPK kinase 1 (MKK1) in cells, together with Raf and Mos.
The Cot/tpl-2 proto-oncogene was originally identified in a modified C-terminally truncated form. The disruption of the last coding exon of the human Cot gene or the truncation of its murine homologue, tpl-2, increases its specific activity (16), revealing the oncogenic potential of the gene (16 -18). The high levels of Cot expression in malignant human Hodgkin/Reed-Sternberg cells (19) and human breast cancer cells (20) further emphasize this oncogenic potential. The overexpression of Cot in different cells induces the activation of several transcription factors, including activator protein 1 (12,14,21), NFAT (22)(23)(24), NFB (13,23,25,26), and E2F (27). The involvement of Cot in the secretion of TNF-␣ in T-cells and macrophages (15,21) and in the secretion of interleukin-2 in T lymphocytes has also been established (22,23). Furthermore, Cot plays an essential role in inducing transcription of the gene encoding cyclooxygenase-2 (19,24). Thus, Cot is a key component in the activation of T-cells and macrophages. Indeed, signals that activate T-cells increase the levels of Cot mRNA, suggesting that the expression of Cot protein increases during the G 0 transition of T lymphocytes (28,29).
In this paper we identify two extracellular signals that increase the specific activity of endogenous Cot, namely LPS and Taxol. Furthermore, we demonstrate that Cot is the only MKKK that activates MKK1, and hence ERK1 and ERK2, in macrophages in response to these signals. We also provide evidence that 15-deoxy-⌬ 12,14 -prostaglandin J 2 (15d-PGJ 2 ), but not rosiglitazone, blocks the LPS-induced activation of Cot and that it also inhibits Cot activity in vitro. These data indicate that 15d-PGJ 2 inhibits LPS-induced Cot activation through a peroxisome proliferator-activated receptor-␥ (PPAR-␥)-independent mechanism.

EXPERIMENTAL PROCEDURES
Materials-Fetal bovine serum (FBS) was purchased from Invitrogen. Prostaglandin E 2 (PGE 2 ), herbimycin A (Herb A), Gö6850 (bisindolylmaleimide I), and 15d-PGJ 2 were from Calbiochem; rosiglitazone was a generous gift from GlaxoSmithKline; phorbol 12-myristate 13acetate (PMA), LPS, sodium orthovanadate, LY 294002, and myelin basic protein were from Sigma; complete proteinase inhibitor mixture was from Roche; and [ 35 S]methionine and [ 35 S]cysteine labeling mixture was from Amersham Biosciences. Taxol (paclitaxel) was generously provided by Dr. Eva López (Bristol-Myers Squibb Co.). Antiphospho-ERK1/ERK2 and anti-ERK2 antibodies were purchased from Zymed Laboratories Inc. Laboratories and anti-phospho-MKK1 plus anti-MKK1 antibodies from Cell Signaling Technology. Antibodies used to immunoprecipitate Cot and c-Raf were raised in sheep against the Cot immunogenic peptide, CQSLDSALFDRKRLLSRKELE, and bacterially expressed c-Raf protein by the Division of Signal Transduction Therapy at the University of Dundee (performed in P. C.'s laboratory). The rabbit anti-Cot antibody utilized for Western blot analysis has been described previously (22).
Cell Culture, Metabolic Labeling, and Stimulation-RAW264 macrophages were maintained in a 95% air, 5% CO 2 atmosphere in RPMI plus 10% (v/v) heat-inactivated FBS, 100 units/ml penicillin, 100 g/ml streptomycin. One day prior to stimulation, macrophages were plated at a density of 1.2 ϫ 10 6 cells/90-cm plate. Three hours later, the medium was removed and replaced overnight with 8 ml of RPMI with 1% (v/v) heat-inactivated FBS. For metabolic labeling, cells were kept in methionine-, cysteine-and glutamine-free RPMI for 20 min and then Immunoprecipitation and Assay of Cot and c-Raf-After stimulation, the medium was removed, and the cells were solubilized in 0.5 ml of ice cold lysis buffer (50 mM Tris acetate (pH 7.0), 1 mM EDTA, 1 mM EGTA, 1% (w/v) Triton X-100, 1 mM sodium orthovanadate, 10 mM sodium glycerophosphate, 50 mM NaF, 5 mM sodium pyrophosphate, 0.27 M sucrose, 0.1% (v/v) 2-mercaptoethanol, and complete proteinase inhibitor mixture, 1 tablet/10 ml). The samples were then frozen in liquid nitrogen and stored in aliquots at Ϫ70°C until analysis. Cell extracts were thawed at 0°C and centrifuged for 10 min at 24,000 ϫ g, and protein concentration was determined by the Bradford method. The protein c-Raf was immunoprecipitated for 1 h at 4°C by incubating 0.5 mg of cell lysate protein with 1 g of c-Raf antibody that had been coupled to protein G-Sepharose. Cot was immunoprecipitated for 3 h at 4°C by incubating 0.5 mg of cell lysate protein with 1.5 g of Cot antibody covalently bound to protein G-Sepharose. The different immunoprecipitates were washed and subjected to a two-step kinase assay as described previously for c-Raf (30). One unit of Cot or c-Raf activity was that amount of enzyme that incorporated 1 nmol of phosphate into myelin basic protein in 1 min.
Western Blot Analysis-Western blotting was performed by resolving 35 g of cell lysate protein on 10% SDS-polyacrylamide gels. The proteins were transferred to polyvinylidene difluoride membranes and probed with anti-phospho-ERK1/ERK2 or anti-phospho-MKK1 antibodies. The blots were developed using a chemiluminescent method (ECL, Amersham Biosciences). As a control of protein loading on the gels, the polyvinylidene difluoride membranes were re-probed with anti-ERK2 or anti-MKK1 antibodies.

LPS and Taxol Activate
Cot but Not c-Raf-Treatment of RAW264 macrophages with LPS (500 ng/ml) produced a timedependent activation of endogenous Cot activity (Fig. 1). Maximal activation reached about 10-fold the basal levels and was attained 15 min after the addition of LPS. In contrast, LPS stimulation did not augment the activity of c-Raf. Because Taxol is reported to mimic LPS-induced activation of murine macrophages through an interaction with CD14 (31-33), we also examined the effect of stimulating RAW264.7 macrophages with 50 M Taxol for different time periods on Cot and c-Raf activity. Not only did Taxol increase the specific activity of endogenous Cot, although to a lesser extent than LPS, but also Taxol failed to up-regulate c-Raf activity (Fig. 1).
PMA Activates c-Raf but Not Cot-It has been well established that through protein kinase C (PKC), the tumor-promoting phorbol esters can trigger the activation of the classical MAPK cascade in a wide variety of cells, including macrophages. We therefore set out to compare the mechanism by which PMA and LPS activate this signaling pathway in macrophages. Stimulation of RAW264 cells with 0.4 g/ml PMA increased c-Raf activity about 10-fold, with the maximal increase occurring 5 min after stimulation. However, in sharp contrast to LPS or Taxol, PMA did not activate Cot activity (Fig. 1), indicating that PMA and LPS/Taxol activate the MAPK cascade in different ways.
The Mechanism by Which LPS Activates Cot-In Western blots of endogenous immunoprecipitated Cot, a single band that migrated in the position expected for the Cot polypeptide was observed ( Fig. 2A). On this basis, we assessed the possibility that LPS increased Cot activity by modulating its association/dissociation with putative regulatory subunit(s). Unstimulated and LPS-stimulated cells were labeled with [ 35 S]methionine and [ 35 S]cysteine, and endogenous Cot was immunoprecipitated from the cell lysates. The immunoprecipitated proteins were separated by SDS-polyacrylamide gel electrophoresis, and subsequent autoradiography revealed a single radiolabeled band of the expected size for Cot (Fig. 2B). The absence of any other labeled protein implied that Cot activation by LPS does not result from its association with, or dissociation from, any other protein.
Another possible mechanism for Cot activation might involve its phosphorylation at one or more residues. However, treatment of Cot immunoprecipitated from LPS-stimulated cells with high concentrations of the catalytic subunit of protein phosphatases 1 or 2A or protein tyrosine phosphatase 1B did not affect the activity of Cot (data not shown). Moreover, the inhibition of serine/threonine phosphatases PP1 or PP2A, potential regulators of Cot phosphorylation, by incubating RAW cells with okadaic acid, did not increase Cot activity (data not shown). Finally, we also transfected Cot into HEK293 cells, which were then stimulated with the protein tyrosine phosphatase inhibitor orthovanadate, or with orthovanadate plus okadaic acid. However, neither of these treatments increased the activity of Cot (data not shown). Taken together, these data indicate that it is unlikely that serine/threonine phosphorylation of Cot is responsible for the way in which LPS influences its level of activity.

Inhibitors of Protein Tyrosine Kinases Block the Activation of Cot by LPS-Herb
A is an irreversible inhibitor of protein tyrosine kinases, and it has been found to block both LPSstimulated tyrosine phosphorylation and the activation of ERK1/ERK2 in macrophages (6). Therefore, we analyzed the effects that this inhibitor might have on Cot activation. Preincubating RAW264 cells with Herb A prevented the activation of Cot by LPS or Taxol (Fig. 3A), although the same treatment did not affect the activation of c-Raf by PMA. Furthermore, Herb A inhibited the activation of MKK1 and ERK1/ERK2 by LPS or Taxol but not by PMA (Fig. 3B). Similar results were also obtained when Herb A was replaced by genistein or tyrphostin AG825, other protein tyrosine kinase inhibitors not structurally related to Herb A (data not shown).

The Activation of Cot by LPS Is Independent of Phosphatidylinositol 3-Kinase and Protein
Kinase C-It has been reported that protein kinase B, which is activated "downstream" of phosphatidylinositol 3-kinase (PI3K), induces the phosphorylation of the C terminus of Cot and that this event is accompanied by an increase in NF-B activation (26). However, the preincubation of RAW264 cells for 30 min with an inhibitor of PI3K, LY 294002 (50 M), did not block the phosphorylation of ERK1/ERK2 or MKK1 upon stimulation with LPS or PMA (data not shown). Moreover, co-transfection of an active form of PI3K together with Cot in HEK293 cells did not increase the activity of Cot, whether or not the cells were exposed to orthovanadate (data not shown). Protein kinase C⑀ (PKC⑀) has been reported to play a critical role in macrophage activation and in the generation of TNF-␣ (34). Because Cot also participates in TNF-␣ secretion (15,21), we decided to evaluate the role that PKC⑀ plays in regulating activation of the classic MAPK cascade via LPS and PMA. As expected, incubation of RAW cells with Gö6850, an inhibitor of PKC␣, PKC␤1, PKC␦, and PKC⑀, blocked the PMA-induced activation of c-Raf, MKK1, and ERK1/ERK2. However, preincubation with Gö6850 did not affect the LPS-induced activation of Cot or the phosphorylation of MKK1 and ERK1/ERK2 (Fig. 4), indicating that neither PI3K nor PKC⑀ is involved in the activation of Cot by LPS.
To address the specificity of 15d-PGJ 2 , RAW macrophages were also pre-incubated with 5 mM PGE 2 prior to stimulation with LPS. Exposure to PGE 2 did not inhibit the LPS-induced activation of Cot (Fig. 5B). Furthermore, we analyzed the effect of rosiglitazone, a synthetic PPAR-␥ ligand with an affinity that is equal to or exceeds the affinity of 15d-PGJ 2 (40), on Cot activation by LPS stimulation. In contrast to 15d-PGJ 2 , rosiglitazone does not contain the cyclopentenone ring with the ␣,␤unsaturated carbonyl group that has been proposed to produce a biological activity independent of PPAR-␥ (40). Preincubation of RAW264 cells with rosiglitazone (10 M) did not inhibit LPS-induced activation of Cot (Fig. 5B) suggesting that 15d-PGJ 2 acts through a mechanism independent of PPAR-␥.
To examine the effect of 15d-PGJ 2 on Cot activity in vitro, cells were stimulated with LPS, and the activity of Cot immunoprecipitated from these cell lysates was measured in the presence or absence of 15d-PGJ 2 (10 M). At this concentration of 15d-PGJ 2 , Cot activity was inhibited by about 75%. The same concentration of rosiglitazone did not inhibit Cot activity in vitro (data not shown). On the other hand, and in contrast to the inhibition of Cot activity by 15d-PGJ 2 , only a small reduction in PMA-stimulated c-Raf activity was observed in the presence of 15d-PGJ 2 in the in vitro assay (Fig. 5C).

DISCUSSION
In this paper we have demonstrated that the activation of macrophages with LPS triggers a 10-fold stimulation of endogenous MKK1 kinase activity of Cot. This stimulation reaches a maximum after 15 min and is correlated with the time of activation of ERK1/ERK2 observed both here and in previous studies (6,41,42). Our data exclude the possibility that c-Raf connects macrophage stimulation by LPS to the activation of ERK1/ERK2 as has been previously proposed (43)(44)(45)(46)(47)(48)(49)(50)(51). Nevertheless, they are consistent with the report that LPS stimulation does not activate ERK1/ERK2 in peritoneal macrophages derived from Cot/tpl-2 knockout mice (15). Thus, Cot appears to be the LPS-stimulated MKKK that activates the classical MAPK cascade in macrophages.
LPS-induced activation of Cot/tpl-2 does not appear to be triggered by the phosphorylation of the Cot/tpl-2 peptide itself, because treatment with high concentrations of protein phosphatase 2A, protein phosphatase 1, or protein tyrosine phosphatase 1B did not inactivate Cot/tpl-2 immunoprecipitated from LPS-stimulated macrophage extracts. A 47-amino acid deletion of the Cot/tpl-2 C terminus increases its activity (16,18), and it has been shown that Cot mRNA levels are increased by activating stimuli in T-cells (28,29). However, Cot pulldown experiments indicate that neither of these mechanisms can be responsible for the activation of Cot by LPS. However, the stabilization of Cot mRNA levels after 2 h of LPS stimulation in RAW macrophages 2 may involve a slower and secondary mechanism for enhancing Cot/tp1-2 activity (data not shown).
It has been reported recently that in unstimulated cells, Cot/tpl-2 is present in a complex with p105, the precursor of the p50 subunit of the NFB transcription factor (52,53), and that LPS induces the dissociation of Cot from p105 (52). We may have been unable to detect p105 in Cot immunoprecipitates from unstimulated cells because the antibody we used was raised against a synthetic peptide corresponding to a sequence near the C terminus of Cot/tpl-2, which is reported to be near the binding site for p105 (53). Therefore, our antibody may dissociate p105 from Cot or may only recognize unbound Cot. Taking into account these recent findings, LPS-induced activation of Cot may require the dissociation of p105 from Cot, perhaps involving the phosphorylation of p105 or another modification of either p105 or Cot. However, because our antibody does not itself induce the activation of Cot in unstimulated cells, the dissociation of p105 from Cot may be only one of the events required for Cot activation.
The overexpression of Cot with protein kinase B in HEK293 cells has been reported to trigger the phosphorylation of two serine residues at the C terminus of Cot (26). Protein kinase B is activated "downstream" of PI3K; however, PI3K does not appear to be involved in the activation of Cot following LPS stimulation, and hence the induction of the classic MAPK cascade. Indeed, LY294002, an inhibitor of PI3K, does not prevent Cot activation at concentrations that inhibit the activation of protein kinase B. In addition, after LPS stimulation, serine/ threonine-specific (PP1, PP2A) and tyrosine-specific (PTP-1B) phosphatases did not affect the activity of Cot in vitro, nor were inhibitors of serine/threonine-specific (okadaic acid) or tyrosine-specific (orthovanadate) phosphatases capable of activating Cot. On the other hand, we found that incubation of macrophages with compounds reported to inhibit protein tyrosine kinases also inhibited Cot activation by LPS or Taxol but not by PMA. This again indicates that in macrophages, different stimuli activate the classical MAPK cascade, using distinct MKKKs. This hypothesis is reinforced by the fact that 15d-PGJ 2 blocks the activation of Cot/tpl-2 and ERK1/ERK2 by LPS or Taxol but not by PMA, and further, by the finding that Gö6850 blocks the activation of Cot/tp1-2 and ERK1/ERK2 by PMA but not by LPS.
Prostaglandin 15d-PGJ 2 is a natural ligand of PPAR-␥, a family of nuclear receptors that functions in ligand-activated transcription (54). PPAR-␥ ligands have recently been implicated in controlling inflammation and particularly in modulating proinflammatory cytokine production. Indeed, some PPAR-␥ ligands inhibit the production of TNF-␣, interleukin-1␤, interleukin-6, nitric-oxide synthase in monocytes, activated macrophages, or endothelial cells (35)(36)(37)(38)55). In this context, it is noteworthy that the effects of the different PPAR-␥ ligands on inflammation do not always correlate with their ability to activate PPAR-␥. Thus, it has been suggested that some ligands, such as 15d-PGJ 2 , can also act though a mechanism independent of PPAR-␥ (36, 37) and other nuclear receptors. This PPAR-␥-independent activity requires micromolar concentrations of 15d-PGJ 2 (37,38), and it has been proposed that the chemically reactive ␣,␤-unsaturated carbonyl moiety of the cyclopentenone ring alkylates exposed cysteine residues on target proteins, thereby modifying their function (37,40). The fact that 15d-PGJ 2 suppresses both LPS-induced Cot activity in intact cells as well as in vitro indicates that this activity is independent of PPAR-␥. Indeed, rosiglitazone, which does not contain the cyclopentenone ring with the ␣,␤-unsaturated carbonyl moiety, cannot mimic 15d-PGJ 2 -mediated inhibition of LPS-induced Cot activation.
To our knowledge, the concentration of 15d-PGJ 2 in serum has never been measured, and it is therefore unclear whether it may accumulate to the concentrations capable of inhibiting Cot in vivo. Indeed, after the induction of an inflammatory response by injecting carrageenin into rats, the concentration of 15d-PGJ 2 measured in rat inflammatory exudates was reported to be 3 nM (56), far below that needed to inhibit the activation of Cot in our experiments. Therefore, further work is needed to establish whether the effect of 15d-PGJ 2 on Cot is physiologically or pharmacology relevant. Nevertheless, if the anti-inflammatory effects of 15d-PGJ 2 are indeed mediated by its effects on Cot, this would reinforce the potential importance of Cot as a drug target.
In summary, the results presented in this paper together with those obtained from Cot "knockout" mice (15) indicate that Cot plays a key role in the LPS-induced activation of the classical MAPK cascade and in the subsequent production of a number of inflammatory cytokines and inflammatory mediators in macrophages. Thus, Cot is potentially an attractive target for the development of improved anti-inflammatory drugs, because its inhibition would not affect the activation of the classic MAPK cascade by growth factors and other agonists known to activate this pathway in many cells and tissues via the activation of one or more isoforms of Raf.