The Tyrosine Kinase Syk Regulates TPL2 Activation Signals*

Tpl2/Cot is a serine/threonine kinase that plays a key physiological role in the regulation of immune responses to pro-inflammatory stimuli, including tumor necrosis factor-α (TNF-α). TNF-α stimulates the JNK, ERK, and p38 mitogen-activated protein kinases and the NF-κB pathway by recruiting RIP1 and TRAF2 to the TNF receptor 1. Here we showed that Tpl2 activation by TNF-α signals depends on the integrity of the Tpl2-interacting proteins RIP1 and TRAF2, which are required for the engagement of the ERK mitogen-activated protein kinase pathway. However, neither RIP1 nor TRAF2 overexpression was sufficient to activate Tpl2 and ERK. We also showed that Tpl2 activation by TNF-α depends on a tyrosine kinase activity that is detected in TNF-α-stimulated cells. Based on both genetic and biochemical evidence, we concluded that in a variety of cell types, Syk is the tyrosine kinase that plays an important role in the activation of Tpl2 upstream of ERK. These data therefore dissect the TNF receptor 1 proximal events that regulate Tpl2 and ERK and highlight a role for RIP1, TRAF2, and Syk in this pathway.

Despite the profound biological effects of overexpressed Tpl2, Tpl2 knock-out (Tpl2 Ϫ/Ϫ ) mice are viable and have no obvious phenotypic defects (10). However, a detailed analysis of these mice revealed a critical role for Tpl2 in the regulation of innate and adaptive immunity and in the response to inflammatory signals through the modulation of the ERK MAPK signaling pathway. Lipopolysaccharide (LPS)-stimulated Tpl2 Ϫ/Ϫ macrophages exhibit impaired synthesis of tumor necrosis factor-␣ (TNF-␣) and cyclooxygenase-2, because of a specific defect in ERK activation (10,11). As a result, Tpl2 Ϫ/Ϫ mice are resistant to LPS/ D-galactosamine-induced septic shock (10). In other studies we found that Tpl2 is also required for the transduction of ERK activation signals initiated by CD40 engagement and contributes to IgE synthesis in B lymphocytes stimulated with interleukin 4 and anti-CD40 mAb (12). Similarly, we found that ERK phosphorylation induced by TNF-␣ is also severely impaired in Tpl2 Ϫ/Ϫ macrophages and that the ERK activation signals are induced by triggering the p55 TNF receptor 1 (TNFR1) (12). Fibroblasts isolated from Tpl2 Ϫ/Ϫ mice demonstrate additional defects in NF-B transactivation and JNK signal transduction in response to TNF stimulation (13). Taken together, these findings may provide an explanation to the reported protection of Tpl2 knock-out mice from TNF-␣-induced inflammatory bowel disease (14).
The mechanism by which Tpl2 is regulated by and contributes to the transduction of pro-inflammatory signals is currently the subject of intense investigation. Previous studies demonstrated that Tpl2 forms a stoichiometric, high affinity complex with the cytoplasmic NF-B inhibitory protein p105 NF-B1 (15) and that the Tpl2:NF-B1 binding stabilizes but inactivates Tpl2. LPS and TNF signals promote the dissociation of Tpl2 from p105 (16). Unbound Tpl2 is active but unstable and undergoes rapid degradation via the proteasome (16). As expected, the steady-state levels of Tpl2 are very low in NF-B1 Ϫ/Ϫ macrophages, and the ability of LPS to activate ERK and its downstream target cyclooxygenase-2 is severely impaired in these cells (16). Although not fully defined, this regulatory event occurs downstream of the IB kinase (IKK) signaling complex that is responsible for the stimulus-dependent serine phosphorylation of NF-B1 and its subsequent degradation by the proteasome (17). Indeed, the inactivation of the IKK␤ catalytic subunit of the IKK signaling complex results not only in impaired NF-B signaling but also in diminished activation of Tpl2 and ERK in response to TNF-␣ or LPS (18,19). IKK␤ is also required for the direct phosphorylation of Tpl2 at Thr 290 , which regulates Tpl2 binding to NF-B1 (20,21). However, although IKK␤ and NF-B1 are required for the activation of Tpl2, additional molecules that regulate Tpl2 are likely to exist. Thus, a role for protein-tyrosine kinases in Tpl2 signaling has been recently highlighted by the inhibitory effects of the broad spectrum tyrosine kinase inhibitor herbimycin-A on LPS-induced Tpl2 activation (22).
The binding of soluble TNF-␣ to TNFR1 induces the recruitment of receptor-interacting protein-1 (RIP1) and TNF receptor-associated factor-2 (TRAF2) to the cytoplasmic C terminus of TNFR1 (23), which results in the transduction of signals that regulate the MAPK pathways and NF-B (24 -26). In the present study, we focused on the mechanisms of Tpl2 activation by TNF-␣, upstream of ERK. Our data demonstrate that Tpl2 plays an important role in the TNF-␣-mediated activation of MEK and ERK across different cell lineages. In addition, our data show that RIP1 and TRAF2 associate with Tpl2 and that their expres-sion is necessary but not sufficient for the transduction of Tpl2 activation signals engaged by TNF-␣. Finally, we provide biochemical and genetic data that demonstrate that the activation of the Tpl2/MEK/ERK cascade by TNF-␣ depends on the tyrosine kinase Syk.

MATERIALS AND METHODS
Cell Culture and Transfections-Cell culture, transfections, and reporter assays were performed as described previously (10,27,28). H33HJ-JA1 and Jurkat E6.1 cells were purchased from European Collection of Animal Cell Culture and maintained in RPMI supplemented with 10% fetal calf serum (Invitrogen). HeLa and MCF7 cells were obtained from the Cancer Research UK Cell Culture facility, London, UK, and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Macrophage and fibroblast cultures were performed as described previously (12). Chemical inhibitors (LY, BisI, PP2, PP3, Pic, and HerbA) were purchased from Calbiochem and dissolved in Me 2 SO prior to use. A Myc-tagged Syk expression vector was kindly provided by Professor Steve Watson, University of Birmingham, UK.
Immunocomplex Kinase Assays-To assay Tpl2 kinase activity, 400 -600 g of whole cell lysate was immunoprecipitated for 4 -5 h with anti-Tpl2 antibody M20 coupled to protein-A-Sepharose (Santa Cruz Biotechnology) or unconjugated protein-A beads as a control. Bound proteins were washed twice with kinase lysis buffer (20 mM Tris, pH 7.6, 0.5% Triton X-100, 250 mM NaCl, 3 mM EGTA, 3 mM EDTA, 1 mM Na 3 VO 4 , 0.5 mM sodium pyrophosphate, 25 mM ␤-glycerophosphate, 10 g/ml leupeptin, 5 g/ml aprotinin, and 1 mM DTT) and twice with kinase reaction buffer (KRB: 20 mM MOPS, pH 7.2, 5 mM EGTA, 1 mM Na 3 VO 4 , 25 mM ␤-glycerophosphate, and 1 mM DTT). The beads were resuspended in 20 l of KRB containing 200 M ATP, 22 mM MgCl 2 , 0.4 g of inactive GST-MEK1 (Upstate Biotechnology, Inc.), and 1 g of inactive GST-ERK2 (Upstate Biotechnology, Inc.). Following incubation for 30 min at 30°C in a shaking incubator, the beads were pellet by centrifugation, and 6 l of the supernatant was transferred to a fresh tube together with 10 l of KRB, 10 l of MBP (Invitrogen) substrate (2 mg/ml stock in KRB), and 10 l of diluted [␥-32 P]ATP (10 Ci of [␥-32 P]ATP in KRB containing 500 M ATP and 75 mM MgCl 2 ). The reaction was incubated for 10 min at 30°C in a shaking incubator, and 10 l were spotted on a 2 ϫ 2-cm P81 paper square (Whatman) in duplicate or triplicate. The squares were washed three times with 0.75% phosphoric acid for 5 min per wash, once with acetone for 5 min, and then left to dry. The [␥-32 P]ATP that was incorporated to MBP was measured in a scintillation counter, and specific activity was calculated. Immunocomplex JNK and ERK in vitro kinase assays using GST-c-Jun (Cell Signaling Technology) and GST-Elk1 (Cell Signaling Technology) as substrates, respectively, were performed as described previously (12,29). For IKK kinase assays, cells were lysed in kinase lysis buffer and cleared by centrifugation (13,000 rpm, 10 min). 500 g of the lysates were then immunoprecipitated with 1.2 g of the anti-IKK␤ mAb H4 (Santa Cruz Biotechnology) and protein-G beads for 3 h. Bound proteins were washed twice with kinase lysis buffer and twice with IKK kinase reaction buffer (20 mM Hepes, pH 7.6, 20 mM MgCl 2 , 1 mM Na 3 VO 4 , 20 mM ␤-glycerophosphate), and the kinase reaction was performed in 20 l of reaction buffer containing 10 M ATP, 5 Ci of [␥-32 P]ATP, and 1 g of GST-IB␣ (Upstate Biotechnology, Inc.) for 30 min at 30°C in a shaking incubator. 20 l of SDS-gel loading buffer was then added to terminate the reaction, and the products were resolved on a 10% gel, transferred to nitrocellulose, and exposed to autoradiography. The membrane was then incubated with anti-IKK␤ polyclonal antibody H470 to confirm similar levels of immunoprecipitated IKK. Activation of Syk was measured by autophosphorylation assays, as described previously (30). Briefly, anti-Syk immunoprecipitates were washed once in kinase lysis buffer and three times with Syk kinase reaction buffer (25 mM Hepes, pH 7.5, 5 mM MgCl 2 , 5 mM MnCl 2 ). The kinase reaction was performed in 30 l of reaction buffer containing 20 Ci of [␥-32 P]ATP for 10 min at 30°C in a shaking incubator. 20 l of SDS gel loading buffer was then added to terminate the reaction, and the products were resolved on a 9% gel, which was dried and exposed to autoradiography.
Antibodies, Immunoblotting and Immunoprecipitations-Antibodies to phosphorylated IBa, MEK, ERK, Syk, and p38 and "total" ERK and p38 were purchased from Cell Signaling Technologies and used in immunoblotting at 1:1000 dilution. Msk1, p50/NF-B1, Akt, Syk 4D10 mAb, and phosphotyrosine 4G10 antibodies were purchased from Upstate Biotechnology, Inc. Polyclonal antibodies against IB␣, p65, TRAF2, and Syk were purchased from Santa Cruz Biotechnology. Agarose and HRP-conjugated anti-Tpl2 polyclonal antibody M20 was also obtained from Santa Cruz Biotechnology. For immunoprecipitations in 293 cells, 500 g of cell extracts were incubated overnight with 2 g of anti-RIP1 antibody (Pharmingen), and the precipitated proteins were analyzed by immunoblot. Syk was immunoprecipitated from 400 g of total cell lysates using 1 g of anti-Syk C-20 (Santa Cruz Biotechnology) or 4D10 antibody and immunoblotted with 4G10 mAb. For detection of NF-B1/Tpl2 interactions, a goat polyclonal NF-B1 antibody (Santa Cruz Biotechnology) was first coupled to protein A-Sepharose beads using dimethyl pimelimidate. The coupled antibody was then incubated with 2 mg of protein extracts overnight, and immunoprecipitated NF-B1 was immunoblotted with either HRP-conjugated Tpl2 or a rabbit polyclonal anti-NF-B1 (Upstate). A similar approach was used to detect endogenous TNFR1/Syk interactions in TNF-␣-stimulated macrophages, using an anti-TNFR1 antibody from R & D Systems.
RNA Interference-For delivery of small interfering RNAs (siRNA), 1 ϫ 10 5 MCF7 cells were plated in each well of a 24-well plate (Costar), and the next day siRNA duplexes were transfected using the siIM-PORTER transfection reagent (Upstate Biotechnology), according to the instructions of the manufacturer. Syk was targeted with the Syk siRNA duplexes GUCGAGCAUUAUUCUUAUAdTdT (and its antisense) and CCUCAUCAGGGAAUAUGUG (and its antisense). As a control, the firefly luciferase siRNA duplex CGUACGCGGAAUACU-UCGAdTdT (and antisense) was used. All siRNAs were synthesized by MWG Biotec, Ebersberg, Germany. Following a 6-h incubation with Syk or luciferase siRNA duplexes, an equal volume of media was added. Cells were then left to recover for 10 h and were re-plated in 24-well dishes. The next day, the cells were subjected to a second round of transfection to increase the siRNA-mediated gene suppression. Twenty four hours later, cultures were serum-starved overnight and then stimulated with TNF-␣, before lysis.

RESULTS
Tpl2 Is Largely Responsible for the Engagement of the MEK/ERK Pathway by TNF-␣-The role of Tpl2 in TNF-induced ERK signaling was first examined in bone marrow-derived macrophages (BMDM) isolated from the Tpl2 ϩ/ϩ and Tpl2 Ϫ/Ϫ mice. Cell cultures were stimulated with human TNF-␣, which selectively engages the mouse TNFR1. Western blots of cell lysates harvested at sequential time points were probed with antibodies against the phosphorylated and active forms of ERK1 and ERK2, the total (phosphorylated and nonphosphorylated) ERK, and the phosphorylated form of the ERK kinase MEK. The results confirmed our previous finding that the TNF-␣-mediated ERK phosphorylation is severely impaired, although not abolished, in Tpl2 Ϫ/Ϫ BMDM (12) (Fig.  1A). Moreover, the results of the present analysis extend this finding by the demonstration that MEK phosphorylation is also significantly reduced in TNF-␣ stimulated Tpl2 Ϫ/Ϫ BMDM (Fig. 1A). Therefore, Tpl2 functions upstream of MEK in the molecular pathway that is initiated by TNFR1 and results in the phosphorylation of ERK.
To confirm that the effect of Tpl2 on TNF-induced ERK signaling is not a macrophage-restricted phenomenon, we performed similar analysis in immortalized Tpl2 ϩ/ϩ and Tpl2 Ϫ/Ϫ fibroblasts. The results showed diminished ERK phosphorylation in TNF-stimulated Tpl2 Ϫ/Ϫ fibroblasts (Fig. 1B). Similarly, the TNF-mediated activation of ERK, as measured by in vitro kinase assays using GST-Elk1 as a substrate, was significantly reduced in MCF7 breast carcinoma cells transfected with low amounts of Tpl2KM, a catalytically inactive Tpl2 (Fig. 1C). Therefore, Tpl2 Tp12KM, a plays an important role in TNF-induced ERK signaling in a range of different cell types. However, additional Tpl2independent mechanisms of ERK signaling are also engaged by TNF stimulation.
RIP1 and TRAF2 Are Necessary for the Activation of the Tpl2/ERK Cascade in Response to TNF-␣ Stimulation-However, RIP1 or TRAF2 overexpression is not sufficient for ERK activation. In an effort to characterize the pathway through which Tpl2 regulates TNFR1-transduced ERK activation signals, we first examined whether Tpl2 is recruited to a RIP1-containing complex in TNF-stimulated cells. To this end, endogenous RIP1 was immunoprecipitated from Tpl2 ϩ/ϩ macrophages before, and 2.5 and 5 min after stimulation with TNF-␣. Immunoprecipitates were then immunoblotted with either an HRP-conjugated Tpl2 antibody or with anti-RIP1. The results showed weak association of RIP1 and Tpl2 in unstimulated cultures, which was significantly increased following treatment with TNF-␣ ( Fig. 2A). The ability of Tpl2 to complex with RIP1 was confirmed in human embryonic kidney (HEK) 293 cells transiently transfected with Myc-tagged Tpl2 or, as a negative control, FLAG-tagged Msk1. RIP1 immunoprecipitates from these cultures were immunoblotted with anti-Tpl2, Msk1, or RIP1 antibodies. It was found that overexpressed Tpl2, but not Msk1, associates with endogenous RIP1 (Fig. 2B). Similar experiments were performed in cells transfected with Myc-tagged NF-B1, a Tpl2-interacting protein.
RIP1 also interacts with TRAF2 following stimulation with TNF-␣ (23, 25). TRAF2 co-immunoprecipitates with overexpressed Tpl2 (28), but the physiological role of TRAF2 in TNF-induced Tpl2/MEK/ERK signaling is unknown. This issue was addressed in fibroblasts isolated from TRAF2 ϩ/ϩ and TRAF2 Ϫ/Ϫ mice (Fig. 2D). Examination of the effects of TNF on ERK phosphorylation in TRAF2 ϩ/ϩ fibroblasts demonstrated significant activation of ERK at 10 and 20 min, which returned to basal levels by 30 min of treatment (Fig. 2E). Most interestingly, the activation of ERK in TRAF2 Ϫ/Ϫ cells was reduced at 10 min and was abolished at 20 min of stimulation with TNF-␣ (Fig. 2E). Therefore, TRAF2 plays an important role in TNFR1-transduced ERK activation signals. This observation raised the question of whether TRAF2 is required for Tpl2 activation in response to TNF-␣. To this end, TRAF2 ϩ/ϩ and TRAF2 Ϫ/Ϫ fibroblasts were treated with TNF-␣ for 10 or 20 min or left untreated, and lysates from these cultures were immunoprecipitated with anti-Tpl2 antibody and subjected to an in vitro coupled kinase assay using inactive MEK, ERK2, and MBP as sequential substrates. The results showed significant increase in the endogenous Tpl2 kinase activity in TNF-stimulated TRAF2 ϩ/ϩ fibroblasts, which was absent in TRAF2 Ϫ/Ϫ cells (Fig. 2F). We conclude that TRAF2 is required for TNFR1 signaling on the Tpl2/MEK/ERK axis.
Overexpression of RIP1 or TRAF2 activates the JNK, p38, and NF-B pathways (31)(32)(33). To determine whether overexpression of TRAF2 and RIP1 also results in the activation of the Tpl2/MEK/ERK pathway, HEK293 cells were first transfected with RIP1, and its effects on the endogenous ERK2 activation were examined by in vitro kinase assays using GST-Elk1 as a substrate. As a control, the effects of transfected RIP1 on JNK1 activation were also assessed by in vitro kinase assays using GST-c-Jun as a substrate. RIP1 overexpression was found to increase the kinase activity of endogenous JNK1 but not ERK2 (Fig. 3, A  and B). Unlike RIP1, Tpl2 overexpression potently induced the activation of both ERK2 and JNK1 (Fig. 3, A and B). Similar results were obtained when the phosphorylation status of transfected HA-ERK1 or HA-JNK1 was examined in RIP1 overexpressing HEK293 cells (data not shown).
We next examined the effects of TRAF2 on the activation of the Tpl2/MEK/ERK pathway. To this end, HEK293 cells were transfected with empty vector (Fig. 3C, 1st 4 lanes), TRAF2 (5th lane) or, as a control, Tpl2 (6th lane), and they were either left untreated (1st, 5th, and 6th lanes) or stimulated with TNF-␣ for 10, 20, or 30 min (Fig. 3C, 2nd to 4th lanes). Endogenous ERK2 or, as a control, JNK1 was immunoprecipitated and examined for catalytic activity using GST-Elk1 or GST-c-Jun as a substrate, respectively. The results showed that expression of TRAF2 induced an ϳ2.2-fold increase in activation of JNK but failed to activate the endogenous ERK in these cells. However, transfection of Tpl2 or treatment with TNF-␣ engaged both signaling pathways (Fig.  3C). Similar results were obtained upon co-transfection of TRAF2 and HA-ERK1 (Fig. 3D), suggesting that the inability of TRAF2 to activate ERK is not merely a consequence of insufficient expression levels of TRAF2 versus ERK. The effects of TRAF2 on the kinase activity of Tpl2 . Serum-starved BMDM or fibroblasts from Tpl2 ϩ/ϩ and Tpl2 Ϫ/Ϫ mice were left untreated or stimulated with 25 ng/ml human TNF-␣ for various time intervals as indicated. Cell lysates were analyzed for phosphorylated MEK and ERK or for total ERK protein levels by immunoblot. C, expression of low amounts of catalytically inactive Tpl2 (Tpl2KM) suppresses TNF-induced ERK activation in MCF7 breast carcinoma cells. MCF7 cells (6 ϫ 10 5 cells per 30-mm dish) were transfected with 1 g of HA-ERK1 and 80 ng of a Tpl2KM construct (9) or control vector and 24 h later were serum-starved overnight. Next day, the transfected cells were stimulated with 40 ng/ml TNF-␣ or left untreated and ERK activation was determined by in vitro kinase assays (IVK) using GST-Elk1 as a substrate.
was also determined. Although TNF stimulation significantly increased the catalytic activity of Tpl2 in a time-dependent manner, transfection of TRAF2 had no effect (Fig. 3D). Taken together, these data suggest that the expression of RIP1 and TRAF2 is necessary but not sufficient for the activation of the Tpl2/ERK cascade in response to TNF-␣ stimulation.
The Syk Tyrosine Kinase Inhibitor Piceatannol Suppresses the TNFmediated Activation of the Tpl2/MEK/ERK Pathway-On the basis of the preceding data, we hypothesized that activation of Tpl2, following its recruitment to the TNFR1-RIP1-TRAF2 complex, requires additional TNF-induced signals. Previous studies had shown that tyrosine kinase signaling may play an important role in Tpl2/MAPK activation by LPS in RAW264.7 macrophages (22). To determine whether tyrosine kinases are also required for the activation of Tpl2 by TNF-␣ in BMDM, we treated these cells with the broad spectrum tyrosine kinase inhibitor herbimycin-A (HerbA), prior to stimulation with TNF-␣, and we analyzed their lysates for the expression and phosphorylation of MEK and ERK. The results showed a dramatic reduction in both MEK and ERK activation in TNF-␣-stimulated macrophages pretreated with HerbA, compared with cells exposed to TNF-␣ alone (Fig. 4A). On the basis of this observation, we proceeded to evaluate the effects of chemical inhibitors that target specific tyrosine kinases. To this end, Tpl2 ϩ/ϩ macrophages were treated with the Syk tyrosine kinase inhibitor piceatannol (Pic) (34), the PI 3-kinase inhibitor LY (35), the Src tyrosine kinase family inhibitor PP2, or its inactive control compound PP3, prior to stimulation with TNF. Moreover, we examined whether BisI (36), which inhibits the protein kinase C isoforms ␣, ␤I, ␤II, ␥, ␦, and ⑀, affects TNF-induced MEK/ERK signaling. The results showed a dramatic reduction in TNF-mediated phosphorylation of MEK and ERK by piceatannol but not by LY, PP2, PP3, or BisI (Fig. 4A). To confirm that LY was otherwise functional, we determined the kinase activity of Akt, an established PI 3-kinase target (37), using in vitro kinase assays with cross-tide as the substrate (38). TNF induced an increase in Akt kinase activity that was abolished in LY-treated cells (data not shown). Moreover, BisI suppressed the 12-O-tetradecanoylphorbol-13-acetate-induced phosphorylation of ERK (data not shown), which is known to occur via a protein kinase C-dependent (39) but Tpl2-independent (10) pathway.
Piceatannol could, theoretically, target the Tpl2-independent arm of TNF-induced ERK phosphorylation (see Fig. 1). To address this issue, immunoblot was re-probed for RIP1 expression. Whole cell lysates (WCL; 50 g) were also analyzed to confirm expression of transfected Msk1 and Tpl2. C, endogenous RIP1 co-precipitates with overexpressed Myc-tagged NF-B1, a Tpl2-interacting protein. D, expression of TRAF2 and Tpl2 in TRAF2 ϩ/ϩ and TRAF2 Ϫ/Ϫ fibroblasts as determined by immunoblot analysis of whole cell lysates. E, TRAF2 contributes to TNF-induced ERK activation. TRAF2 ϩ/ϩ and TRAF2 Ϫ/Ϫ fibroblasts were either left untreated or stimulated with TNF-␣, as indicated. Cell lysates were immunoblotted for phosphorylated or total ERK. Data are representative of four independent experiments. F, TRAF2 is required for TNF-induced Tpl2 activation. TRAF2 ϩ/ϩ and TRAF2 Ϫ/Ϫ fibroblasts were stimulated with TNF-␣ for 0, 10, or 20 min, as indicated, and lysed. Endogenous Tpl2 was immunoprecipitated from these lysates and subjected to in vitro coupled kinase assay using inactive MEK, ERK2, and MBP as sequential substrates. The relative increases in MBP phosphorylation (ϮS.D.) from three independent experiments are shown.
Tpl2 ϩ/ϩ and Tpl2 Ϫ/Ϫ BMDM were pretreated with piceatannol and then stimulated with TNF-␣ for 10 min. Piceatannol inhibited the TNFmediated activation of ERK in Tpl2 ϩ/ϩ but not in Tpl2 Ϫ/Ϫ BMDM (Fig.  4B), suggesting that Syk or a Syk-related kinase functions upstream of Tpl2 in TNFR1 signaling. The inhibition of ERK phosphorylation in response to TNF-␣ was not restricted to mouse macrophages, as piceatannol also suppressed the TNF-induced phosphorylation of ERK in H33HJ-JA1 Jurkat T cells, as well as in HeLa human cervical carcinoma cells (Fig. 4C). Both H33HJ-JA1 and HeLa are known to express Syk (40,41). The effects of piceatannol on TNF-induced Tpl2 activation were FIGURE 3. Overexpression of RIP1 or TRAF2 is insufficient to activate the ERK pathway. A and B, RIP1 overexpression activates JNK but not ERK. HEK293 cells were transfected with RIP1 or, as a control, Myc-tagged Tpl2. Anti-ERK2 immunoprecipitates (I.P.) of total cell lysates were subjected to in vitro kinase assay (IVK) using GST-Elk1 as a substrate (A). vec, vector. To confirm that under these experimental conditions the expression of RIP1 was functional, the same lysates were immunoprecipitated using anti-JNK1 antibody prior to examination of JNK kinase activity by in vitro kinase assays (B). C, TRAF2 overexpression fails to activate ERK. HEK293 cells were transfected with TRAF2 or, as a control, Myc-tagged Tpl2. Anti-ERK2 immunoprecipitates of total cell lysates were subjected to in vitro kinase assay using GST-Elk1 as a substrate. Parallel cultures were stimulated with 40 ng/ml TNF-␣ prior to lysis and evaluation of ERK2 kinase activity. To confirm that under these experimental conditions the expression of TRAF2 was functional, the same lysates were immunoprecipitated using anti-JNK1 antibody prior to examination of JNK kinase activity by in vitro kinase assays using GST-c-Jun as a substrate. D, TRAF2 overexpression does not promote the phosphorylation of co-expressed HA-ERK1 in HEK293 cells. E, TRAF2 overexpression does not activate Tpl2. HEK293 cells were transfected with TRAF2 or stimulated with TNF-␣, as indicated, and Tpl2 activity was assessed by coupled in vitro kinase assays. Data shown are the mean values (ϮS.D.) from three independent experiments. The inset shows the levels of immunoprecipitated Tpl2 in a representative experiment.

FIGURE 4. The Syk tyrosine kinase inhibitor piceatannol severely impairs the TNF-mediated activation of the Tpl2/MEK/ERK pathway.
A, the tyrosine kinase inhibitor HerbA and the Syk tyrosine kinase inhibitor piceatannol (Pic) severely impair TNF-induced phosphorylation of MEK and ERK in BMDM. Serum-starved Tpl2 ϩ/ϩ macrophages were pretreated with HerbA (1 g/ml), piceatannol (30 M), the protein kinase C inhibitor BisI (10 M), the PI 3-kinase inhibitor LY (20 M), and the Src family tyrosine kinase inhibitor PP2 (20 M) or its control compound PP3 (20 M) and then stimulated with 25 ng/ml TNF-␣ for 10 min. Lysates were analyzed for phosphorylated MEK and ERK or for total ERK by immunoblot. DMSO, Me 2 SO. B, piceatannol does not suppress ERK phosphorylation in Tpl2 Ϫ/Ϫ macrophages. Tpl2 ϩ/ϩ and Tpl2 Ϫ/Ϫ macrophages were pretreated with piceatannol for 1.5 h and then stimulated with TNF-␣ for 10 min or left untreated. Cell lysates were examined for ERK phosphorylation by immunoblot. C, piceatannol suppresses ERK phosphorylation in the Jurkat T cell lymphoma clone H33HJ-JA1 and the cervical carcinoma cell line HeLa. Cells were pretreated with piceatannol as in B and then they were either left untreated or stimulated with TNF-␣ at 40 ng/ml (H33HJ-JA1) or 30 ng/ml (HeLa). Lysates were analyzed for ERK phosphorylation by immunoblot. D, piceatannol inhibits TNF-induced Tpl2 kinase activity. Lysates from Tpl2 ϩ/ϩ BMDM treated as in A were subjected to in vitro coupled kinase assay using inactive MEK/ERK and MBP as substrates, respectively. Data shown are the mean values (ϮS.D.) from four independent experiments. then determined by in vitro coupled kinase assays. The results of this analysis showed a 3.7-fold increase in Tpl2 kinase activity following stimulation with TNF, which was reduced by more than 70% in piceatannol-treated cultures (Fig. 4D).
We next examined whether piceatannol treatment impinges on the TNF-mediated dissociation of Tpl2 from NF-B1, a critical step for Tpl2 activation (16,18). To this end, Tpl2 ϩ/ϩ macrophages or H33HJ-JA1 Jurkat cells were either left untreated or treated with piceatannol and then stimulated with TNF-␣. Endogenous NF-B1 was immunoprecipitated from these cultures and immunoblotted with antibodies against Tpl2 or NF-B1. As reported previously (18), TNF treatment promoted the dissociation of the long form of Tpl2 (Tpl2 L ) from NF-B1. Most interestingly, this effect was blocked by pretreatment with piceatannol in both lines (Fig. 5A). As IKK␤ and its regulatory subunit, IKK␥, are critical upstream regulators of the NF-B1-Tpl2 complex (18), the effect of piceatannol on TNF-␣-induced IKK activity was examined. The results showed that this inhibitor suppresses the TNF-mediated phosphorylation of IB␣ (Fig. 5B) and the catalytic activity of endogenous IKK␤ in macrophages (Fig. 5C) and H33HJ-JA1 cells (data not shown). Therefore, piceatannol inhibits the TNF-mediated activation of the Tpl2/MEK/ERK pathway by functioning upstream of the IKK signaling complex.
Role of Syk in TNF-mediated Activation of the Tpl2/MEK/ERK Pathway-The preceding data implicate Syk in Tpl2 signaling inasmuch as the Syk inhibitor piceatannol suppresses the TNF-mediated activation of Tpl2 and its downstream targets MEK and ERK. Piceatannol, however, may also target kinases other than Syk. We therefore sought to obtain additional evidence for the role of Syk in TNFR1 signaling upstream of the Tpl2/MEK/ERK pathway. To this end, we first examined the effects of TNF stimulation on Syk tyrosine phosphorylation, a prerequisite for its activation. Total cell lysates from Tpl2 ϩ/ϩ BMDM were isolated before and 1, 2.5, 5, or 10 min after stimulation with TNF-␣ and immunoblotted with the anti-phosphotyrosine antibody 4G10. The results showed increased tyrosine phosphorylation of an ϳ72-kDa protein within 2.5 min of stimulation (Fig. 6A). Re-probing the same immunoblot with an anti-Syk polyclonal antibody verified the identity of this protein as Syk. To confirm its phosphorylation in response to TNF-␣, Syk was immunoprecipitated from TNF-stimulated and control untreated Tpl2 ϩ/ϩ and Tpl2 Ϫ/Ϫ BMDM and immunoblotted with anti-phosphotyrosine (4G10) or anti-Syk antibodies (Fig.  6B). Furthermore, Syk activation was confirmed by in vitro kinase assays of immunoprecipitated Syk before and after TNF stimulation. In these experiments, the activation of Syk was dramatically suppressed in cells that were treated with piceatannol prior to TNF stimulation (Fig. 6C). Interestingly, Syk was also detected in TNFR1 immunoprecipitates from TNF-stimulated macrophage lysates (Fig. 6D).
Next, we sought to obtain genetic evidence for the role of Syk in the TNF-mediated activation of the Tpl2/MEK/ERK pathway. To this end, we first used a Jurkat T cell derivative (Jurkat E6.1) that lacks expression of Syk as a result of a single nucleotide insertion in the Syk open reading frame, leading to premature translational termination (40). The absence of Syk expression in Jurkat clone E6.1 and the expression of significant levels of Syk in H33HJ-JA1 Jurkat T cells (40) were confirmed by immunoblotting (Fig. 7A). H33HJ-JA1 and Jurkat E6.1 cells were then compared for the activation of the Tpl2/MEK/ERK cascade following stimulation with TNF-␣. The results showed significant activation of ERK in TNF-treated H33HJ-JA1 but diminished response in Jurkat E6.1 cells (Fig. 7B). The kinase activity of Tpl2 mirrored the phosphorylation of ERK, showing diminished response in Jurkat E6.1 lymphocytes (Fig. 7C). Moreover, TNF-stimulated Jurkat E6.1 cells also showed delayed and reduced IB␣ degradation compared with H33HJ-JA1 cultures (Fig.  7B), consistent with the reported critical role of IKK␤/IKK␥ in TNFinduced Tpl2 activation (18,19).
To explore further the role of Syk in TNFR1 signaling upstream of Tpl2, we performed knock-down experiments in MCF7 breast carcinoma cells, which are known to express Syk (42), using Syk-specific small interfering RNAs (siRNAs). Lysates of MCF7 cells transfected with siRNAs targeting Syk or the unrelated luciferase gene, were analyzed for the expression of Syk, as well as Tpl2 and ␤-actin as controls, by immunoblotting. Transfection of the Syk siRNA induced a significant reduction in the expression of Syk but not Tpl2 or ␤-actin (Fig. 7D). Lysates of parallel cultures, stimulated with TNF-␣ for 15 min or left untreated, were analyzed for the expression of phosphorylated and total ERK by immunoblotting. The results demonstrated significant phosphorylation of ERK in luciferase but not Syk siRNA-transfected cells (Fig. 7E). In addition, although stimulation of luciferase siRNA-transfected cells with TNF-␣ resulted in robust IB␣ degradation, this effect was significantly impaired in Syk siRNA-transfected cultures (Fig. 7E). Analysis of cell lysates for Tpl2 kinase activity using in vitro coupled kinase assays confirmed that Syk plays an important role in the TNFmediated activation of Tpl2 (Fig. 7F). immunoprecipitates (IP) from 2 mg of these lysates were immunoblotted (IB) with HRPconjugated Tpl2 or NFB1 antibodies. Tpl2 and NFB1 expression in 50 g of whole cell lysate (WCL) from unstimulated cultures is also shown. B, piceatannol (Pic) suppresses the TNF-mediated phosphorylation of IB␣. Tpl2 ϩ/ϩ BMDM were pretreated with 30 M piceatannol and then stimulated with TNF-␣ for 10 min or left untreated. Lysates were analyzed for IB␣ and ERK phosphorylation by immunoblot. The levels of total ERK were also evaluated as a loading control. Note that the effect of Pic on IB␣ phosphorylation precedes that on ERK. C, treatment of Tpl2 ϩ/ϩ BMDM with piceatannol suppresses the TNF-induced IKK␤ kinase activity. Cells were pretreated with 30 M piceatannol and then stimulated with TNF-␣ for 5 min or left untreated. Lysates were immunoprecipitated with an anti-IKK␤ mAb and analyzed for catalytic activity by in vitro kinase assays (IVK) using GST-IB␣ as a substrate. The immunoprecipitated IKK␤ was immunoblotted and probed with a polyclonal anti-IKK␤ to confirm that similar levels of the enzyme were analyzed.
The effects of Syk overexpression on TNF-induced Tpl2/MEK/ERK activation were then investigated. To this end, HEK293 cells, which express very low levels of endogenous Syk (Fig. 8), were transiently transfected with HA-ERK1 in the presence or absence of a Myc-tagged Syk expression vector. Cells were then stimulated with TNF-␣ or left untreated, and lysates were examined for expression of Syk and phosphorylated and total HA-ERK1 by immunoblot. Lysates were also immunoprecipitated with anti-Tpl2 and exposed to in vitro kinase assays using GST-MEK as a substrate. The results showed that Syk overexpression significantly enhances the TNF-␣-mediated phospho-FIGURE 6. Syk becomes tyrosine-phosphorylated following TNF stimulation. A and B, TNF-␣ promotes the tyrosine phosphorylation of Syk. Tpl2 ϩ/ϩ BMDM were stimulated with TNF-␣ for various time intervals or left untreated, as indicated. Cell lysates were subjected to immunoblot (I.B.) using the phosphotyrosine mAb 4G10. The blot was then stripped and re-probed with an antibody raised against the C terminus of Syk. The relative levels of Syk phosphorylation (ratio of phosphorylated versus total protein) were quantitated by using the Scion Image processing and analysis software and are expressed as fold increase relative to unstimulated control (A). To confirm that Syk is the tyrosine-phosphorylated protein detected following TNF stimulation, Syk was immunoprecipitated (IP) from untreated and TNF-activated Tpl2 ϩ/ϩ BMDM and immunoblotted with the phosphotyrosine mAb 4G10. The blot was then re-probed with an anti-Syk antibody (B). C, Syk activation following TNF stimulation of Tpl2 ϩ/ϩ BMDM. Tpl2 ϩ/ϩ macrophages were pretreated with Pic or left untreated and then stimulated with 25 ng/ml TNF-␣, as indicated. Syk was immunoprecipitated and examined for auto-phosphorylation by in vitro kinase assays, as described under "Materials and Methods." D, Syk is recruited to the TNF receptor signaling complex following TNF stimulation. Lysates (2.5 mg) from TNF-␣ stimulated Tpl2 ϩ/ϩ BMDM and untreated controls were immunoprecipitated with an anti-TNFR1 antibody coupled to protein-A-Sepharose beads. The immunoprecipitates were then immunoblotted for either Syk or TNFR1. Two additional experiments were performed and gave similar results. rylation of ERK1 and the activation of endogenous Tpl2 (Fig. 8). Taken together, the preceding data demonstrate that Syk plays an important role in the activation of the Tpl2/MEK/ERK pathway in response to TNF-␣.

DISCUSSION
This study dissects the TNF receptor-proximal events that lead to the engagement of the ERK MAPK cascade by TNF-␣ in macrophages. It shows that the stimulation of TNFR1 results in ERK activation via two independent signaling pathways. The first involves the activation of Tpl2 and is responsible for the majority of the TNF-␣-induced MEK/ ERK signals (Fig. 1). The second pathway is Tpl2-independent and mediates the relatively low level MEK and ERK phosphorylation observed in Tpl2 Ϫ/Ϫ BMDM and MEFs.
Mutational analysis of the cytoplasmic tail of TNFR1 has identified a region called the "death domain" as being responsible for ERK signaling (43). This region binds the adaptor proteins TRADD (44) and MADD (45). TRADD acts as a platform for the recruitment of RIP1 and TRAF2, which mediate the activation of NF-B and MAPKs, and of FADD, which mediates caspase activation and apoptosis (46,47). MADD overexpression results in the activation of JNK, p38, and ERK MAPKs (45), but its physiological role in TNF signal transduction is currently unknown.
Previous studies using RIP1 Ϫ/Ϫ cells had shown that RIP1 is required for the TNF-␣-induced activation of NF-B, JNK, p38, and ERK (24,26,48), suggesting that bifurcation of signals originating in TNFR1 must occur downstream of RIP1. TRAF2, which interacts with RIP1, is obligatory for JNK activation and contributes to the activation of NF-B by TNF-␣ (49). Its physiological contribution to TNF-induced ERK signaling is, however, unknown. Data presented here demonstrate that TRAF2 plays an important role in the engagement of ERK in response to TNF stimulation (Fig. 2D). Thus, we have shown that TNF-␣-induced Tpl2 activation and ERK phosphorylation are significantly impaired in TRAF2 Ϫ/Ϫ fibroblasts. Given that Tpl2 co-immunoprecipitates with both RIP1 (Fig. 2, A and B) and TRAF2 (28), we propose that, following TNFR1 stimulation, Tpl2 is recruited to and is activated by the RIP1-TRAF2 signaling complex (see proposed model in Fig. 9).
It is of interest that although TRAF2 and RIP1 are essential for the activation of the Tpl2/MEK/ERK axis by TNFR1, their overexpression is not sufficient to activate ERK. Thus, although overexpressed TRAF2 and RIP1 readily engage JNK, they do not activate Tpl2 and ERK (Fig. 3).
These findings suggest that Tpl2 is activated by TRAF2-and RIP1transduced signals operating in concert with additional converging signals, also induced by TNF-␣. Another molecule that behaves similar to TRAF2 and RIP1 is IKK␤ that plays a critical role in the TNF-␣-mediated activation of both the NF-B (50 -52) and the Tpl2/MEK/ERK pathways (18). Tpl2 activation is severely impaired in IKK␤ Ϫ/Ϫ fibroblasts treated with TNF-␣ (18). However, overexpression of IKK␤ in HEK293 cells failed to induce the phosphorylation of co-expressed ERK1 (data not shown). The preceding data, coupled with the results of the published studies, raised the possibility that in addition to RIP1/ TRAF2/IKK␤, other molecules are required for the activation of Tpl2 in response to TNF stimulation. Here we demonstrate that one of these molecules is the tyrosine kinase Syk.
Tyrosine kinases have been implicated in signal transduction by a plethora of cytokines and growth factors. A recent report has shown that the broad spectrum tyrosine kinase inhibitor herbimycin-A severely impairs the ability of LPS to activate Tpl2 and its downstream target MEK in macrophages (22). Several candidate tyrosine kinases were considered unlikely targets of herbimycin-A in the preceding experiments. Specifically, the Src kinases were unlikely to be responsible for the herbimycin effect, because Src is not expressed in BMDM (53). Moreover, macrophages from triple knock-out mice lacking the Src family kinases Hck, Fgr, and Lyn exhibit normal LPS-induced ERK phosphorylation (53), which is Tpl2-dependent (10). In agreement with these predictions, PP2, an inhibitor of the Src kinase family, did not interfere with activation of ERK by TNF-␣ (Fig. 4A). Other tyrosine kinases, such as Tyk2 and FAK, are also unlikely targets of herbimycin-A as Tyk2 Ϫ/Ϫ and FAK Ϫ/Ϫ cells also exhibit normal ERK signaling in response to LPS and TNF-␣, respectively (54,55). Akt, a downstream target of PI 3-kinase, which has been reported to phosphorylate Tpl2/ Cot at its C terminus (56), is not involved in the activation of Tpl2 in response to TNF-␣, as suggested by the finding that the TNF-induced  We propose that following stimulation with TNF-␣, Syk becomes activated and acts in concert with RIP1 and TRAF2 to activate Tpl2 upstream of MEK and ERK. Syk appears to modulate signaling on the Tpl2/MEK/ERK axis by functioning upstream of IKK␤, which controls the dissociation of Tpl2 from NF-B1, a prerequisite for Tpl2 activation. A role for additional tyrosine kinases in TNF-␣-induced Tpl2/ERK signaling cannot be excluded. Arrows indicate the main ERK pathway. Dotted lines depict pathways that have not been fully characterized yet (see text for details).
activation of ERK in BMDM is not affected by the PI 3-kinase inhibitor LY294002 (Fig. 4A).
In addition to the above kinases, we examined the role of the tyrosine kinase Syk in Tpl2/ERK signaling. This investigation was prompted by a number of published observations. First, Syk is expressed in a plethora of cell types, including monocytes, B and T lymphocytes (57), epithelial cells (41,42,58), and fibroblasts (59). Second, Syk is recruited to the TLR4 signaling complex following LPS stimulation (60) and undergoes tyrosine phosphorylation in LPS-treated neutrophils and macrophages (60,61). Third, Syk has been implicated in the activation of JNK and NF-B in neutrophils and T cells exposed to TNF-␣ (30,62), and its knock-down in epithelial cells affects TNF-induced phenotypic effects, such as IL6 synthesis and ICAM1 expression (58).
Data presented in this report, demonstrate that Syk is indeed involved in the activation of Tpl2. Thus, Syk became tyrosine-phosphorylated following TNF stimulation, and the Syk inhibitor piceatannol significantly reduced the activation of Tpl2 as well as the phosphorylation of MEK and ERK in TNF-␣-stimulated Tpl2 ϩ/ϩ but not Tpl2 Ϫ/Ϫ macrophages (Fig. 4). This finding suggests that Syk has an active role upstream of Tpl2 signal transduction and is not merely a scaffold for components of the TNFR1 pathway. Similarly, we have found that piceatannol suppressed the LPS-mediated activation of the Tpl2/MEK/ ERK pathway in macrophages (data not shown), consistent with its reported ability to confer protection against LPS-induced septic shock in mice (63). Moreover, we have shown that cell lines that either lack or express reduced levels of Syk exhibited diminished activation of Tpl2 and ERK in response to TNF-␣ treatment, whereas the overexpression of Syk augmented TNF-induced Tpl2/ERK signaling (Figs. 7 and 8). More importantly, interference with Syk levels and function by either knock-down of Syk expression or piceatannol treatment hindered the TNF-mediated activation of IKK␤ (Figs. 5 and 7), which is required for the inducible dissociation of Tpl2 from NF-B1 and the subsequent activation of Tpl2 (18,19). These findings suggest that Syk functions upstream of the IKK signaling complex to control the activation of Tpl2 via dissociation from NF-B1.
In summary, the present study dissects a signaling pathway that is triggered by TNF-␣ and results in ERK MAPK activation (Fig. 9). The data show that TNF-␣ activates Tpl2 via signals transduced by RIP1 and TRAF2 and that Tpl2 is required for the activation of the MEK/ERK axis. Genetic and biochemical evidence presented in this study also shows that Tpl2 activation depends on the tyrosine kinase Syk, which is activated following TNF-␣ stimulation. Given the key role of Tpl2 in immunity and inflammation, understanding the mechanisms of its activation may have significant implications for the design of novel therapeutic strategies against a variety of inflammatory syndromes.