Role of Receptor-interacting Protein in Tumor Necrosis Factor-α-dependent MEKK1 Activation

Receptor-interacting protein (RIP), a death domain serine/threonine kinase, has been shown to play a critical role in tumor necrosis factor-α (TNF-α)-induced activation of the nuclear factor-κB signaling pathway. We demonstrate here that ectopically expressed RIP induces I-κB kinase-β (IKKβ) activation in intact cells and that RIP-induced IKKβ activation can be blocked by a kinase-inactive form of MEKK1, MEKK1(K1253M). Interestingly, RIP physically associated with MEKK1 both in vitro and in vivo. RIP phosphorylated MEKK1 at Ser-957 and Ser-994. Our data also indicate that RIP induced the stimulation of MEKK1 but not MEKK1(S957A/S994A) in transfected cells. Furthermore, overexpressed MEKK1(S957A/S994A) inhibited the RIP-induced activation of both IKKβ and nuclear factor-κB. We also demonstrated that the TNF-α-induced MEKK1 activation was defective in RIP-deficient Jurkat cells. Taken together, our results suggest that RIP phosphorylates and activates MEKK1 and that RIP is involved in TNF-α-induced MEKK1 activation.

Proinflamatory cytokine tumor necrosis factor-␣ (TNF-␣) 1 stimulates various signaling pathways leading to proliferation, differentiation, or apoptosis of cells through its receptors TNFR1 and TNFR2 (1). TNF-␣, by binding to TNFR1, can activate the apoptotic pathway in which TNFR1 associates with death domain proteins such as TNFR-associated death domain (2) and Fas-associated death domain (3). The receptor complexes then form a death-inducing signaling complex with caspase-8, which subsequently activates the executive caspases, causing apoptotic cell death (4). TNF-␣ can also induce the association of TNFR with TNF receptor-associated factors (TRAFs). The TNFR⅐TRAF complexes activate the nu-clear factor-B (NF-B) pathway, which functions as a cell survival signal (5)(6)(7).
TNFR-bound TRAFs have been shown to induce NF-B activation through down-regulation of I-B, an inhibitor of NF-B (8). Down-regulation of I-B results from I-B phosphorylation by I-B kinases (IKKs) and ubiquitin-dependent degradation of the phosphorylated I-B (9). Three isoforms of IKK have been identified: ␣, ␤, and ␥. Studies using knockout mice have shown that IKK␤ is responsible for the TNF-␣-induced NF-B activation (10), whereas IKK␣ is dispensable for the TNF-␣ signaling (11). Nonenzymatic IKK␥/NEMO induces the formation of an IKK␣ and IKK␤ complex (12). In addition to the activation of the IKKs by IKK␥-mediated complex formation, IKK␣ and IKK␤ can be activated by phosphorylations catalyzed by upstream kinases such as the NF-B inducing kinase (NIK) (13), MEKK1 (14) and Akt (15). NIK and MEKK1 have been reported to directly interact with TRAF2. Thus, TRAF2 has been thought to be an important component of the TNFR signaling complexes in the NF-B activation (16,17). However, a recent study using TRAF2-null mice has demonstrated that the TNF-␣-induced NF-B activation is intact while the TNF-␣-induced c-Jun N-terminal kinase activation is defective in the TRAF2deficient cells (18). Therefore, the major physiological function of TRAF2 may be the activation of the c-Jun N-terminal kinase pathway through physical interaction with mitogen-activated protein (MAP) kinase kinase kinases such as MEKK1 (17) and apoptosis-stimulating kinase 1 (19) rather than the activation of the NF-B pathway.
Receptor-interacting protein (RIP) has been shown to interact with cell surface receptors containing death domains such as TNFR1, Fas, and TNF-related apoptosis-inducing ligand (TRAIL), and with cytoplasmic adaptor proteins containing death domains such as Fas-associated death domain and TNFR-associated death domain (20 -24). RIP can also associate with proteins involved in the NF-B signaling such as TRAF2, Epstein-Barr virus-latent membrane protein 1, p62, and IKK␥ (25)(26)(27)(28). It has been reported previously that RIP-deficient thymocyte cells were defective in TNF-␣-induced NF-B activation but not c-Jun N-terminal kinase activation (29).
Although RIP possesses serine/threonine kinase activity, the biological significance of the kinase activity is not yet clearly understood, and the physiological substrates of RIP have not yet been identified. In the present study, we investigated the mechanism underlying the RIP-mediated IKK activation to understand the role of RIP in the NF-B signaling pathway. We demonstrate here that MEKK1 is a substrate of RIP. The RIP-induced phosphorylation of MEKK1 enhances enzymatic activity of MEKK1, and it then contributes to the TNF-␣induced activation of NF-B.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-Mouse monoclonal anti-hemagglutinin (HA) antibody was purchased from Roche Molecular Biochemicals. Mouse monoclonal anti-FLAG antibody was purchased from Stratagene, and mouse monoclonal anti-Myc antibody was from Santa Cruz, Inc. Human recombinant TNF-␣ was bought from R&D Systems. Mouse monoclonal anti-RIP and anti-MEKK1 antibodies were purchased from Pharmingen. 32  Vector Constructs and Transfection-pRK7-FLAG-IKK␣, pRK5-FLAG-IKK␤, pRK5-FLAG-IKK␤(K44A), and pRK7-FLAG-NIK were kindly provided by Dr. David Goeddel (Tularik Inc.). Fragments of the human RIP gene (20) were isolated by polymerase chain reaction and cloned into the BamHI/XhoI sites of pcDNA3-HA or pGEX-4T-1. Derivatives of MEKK1 were produced by polymerase chain reaction from pcDNA3-HA-MEKK1 (30) and cloned into the BamHI/EcoRI sites of the pCMV2-FLAG vector (Kodak) or the BamHI/SalI sites of pGEX-4T-1 (Amersham Pharmacia Biotech). Point mutations of MEKK1 were produced using the QuikChange site-directed mutagenesis kit (Stratagene). Vector constructs were transfected into human embryonic kidney 293 (HEK293) cells using LipofectAMINE (Life Technologies, Inc.) and RIP(Ϫ) Jurkat cells by electroporation at 0.4 kV and 960 microfarads using an electroporator (Bio-Rad).
Cell Culture and Metabolic Labeling using 32 PO 4 -HEK293 cells were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum. Wild type and RIP(Ϫ) Jurkat cells (31) were grown in RPMI 1640 with 10% fetal bovine serum. Phosphorylation of MEKK1 in cells was examined as described previously (32). Briefly, wild type or RIP(Ϫ) Jurkat cells were incubated in phosphate-free Dulbecco's modified Eagle's medium containing 10% phosphate-free fetal bovine serum for 1 h and labeled with 100 Ci/ml carrier-free 32 PO 4 . Cells were then treated with 20 ng/ml TNF-␣ or 80 J/m 2 UV and incubated further for 10 or 30 min, respectively. Cells were lysed and subjected to immunoprecipitation using anti-MEKK1 anti-FIG. 1. RIP activates IKK␤. Panel A, schematic diagram of RIP and its derivatives (upper section). HEK293 cells were transiently transfected with pcDNA3-HA vector expressing RIP or its mutants, and the expression of HA-tagged proteins was confirmed by immunoblot analysis (lower section). Panel B, effects of RIP and its mutants on IKK activity. HEK293 cells were transfected with pcDNA3-HA vector encoding RIP or its mutants along with pRK7-FLAG-IKK␣ or pRK5-FLAG-IKK␤. pCMV-crmA was also included in the transfection to protect the cells from apoptosis induced by overexpression of RIP. Where indicated, the cells were treated for 20 min with 20 ng/ml TNF-␣ at 48 h after transfection. Cell lysates were examined for IKK␣ or IKK␤ activity by immunocomplex kinase assay using anti-FLAG antibody. The band intensities were quantified using a PhosphorImager BAS2500 (Fuji). IB, immunoblot. body. The immunopellets were analyzed by 8% SDS-PAGE and autoradiography.
Luciferase Reporter Assay-To assess the transcription stimulating activity of NF-B, pcDNA3-3X-NF-B-Luc, and pcDNA3-␤-galactosidase were cotransfected with various vector constructs into the cells, and luciferase activities in the transfected cells were measured using a luciferase assay kit (Promega). Luciferase activity was normalized relative to the coexpressed ␤-galactosidase activity.
Immunocomplex Kinase Assays-IKK activity was measured by the method of Regnier et al. (33). Briefly, cells were transfected with FLAG-IKK␣, FLAG-IKK␤, or other indicated constructs. Where indicated, the cells were treated for 20 min with 20 ng/ml recombinant human TNF-␣ at 48 h after the transfection. The cell lysates were subjected to immunoprecipitation using anti-FLAG antibody. IKK activity in the immunocomplexes was determined by incubating them at 30°C for 30 min in a reaction buffer containing 20 mM HEPES (pH 7.5), 20 mM ␤-glycerophosphate, 10 mM MgCl 2 , 100 mM Na 3 VO 4 , 2 mM dithiothreitol, 20 mM ATP, 100 Ci/ml [␥-32 P]ATP, and 2 g GST-I-B␤. Phosphorylation of GST-I-B␤ was analyzed by 10% SDS-PAGE followed by autoradiography. c-Jun N-terminal kinase 1 and MEKK1 activity was measured by a procedure described previously (34). c-Jun N-terminal kinase 1 and MEKK1 activity was monitored by following the phosphorylation of GST-c-Jun and GST-SEK1(K129R) or GST-IKK␣(1-200), respectively. To assess the phosphorylation of MEKK1 by RIP, HEK293 cells were transfected with HA-RIP or MEKK1(K1253M)-FLAG construct. After 48 h of transfection, the cells were subjected to immunoprecipitation using anti-HA or anti-FLAG antibody. HA-RIP immunoprecipitates were then used for the RIP kinase assay by incubating with MEKK1(K1253M)-FLAG immunoprecipitates at 30°C for 30 min in a reaction buffer containing 20 mM HEPES (pH 7.3), 10 mM MnCl 2 , 10 mM MgCl 2 , and 100 Ci/ml [␥-32 P]ATP as described elsewhere (31).
GST Pulldown in Vitro Binding Assay and Coimmunoprecipitation-Either RIP or MEKK1 was labeled with [ 35 S]methionine by in vitro transcription/translation procedures using the Quick-coupled TNT kit (Promega). The 35 S-labeled proteins were incubated with 1 g of GST fusion proteins in binding buffer consisting of 20 mM Tris (pH 7.4), 100 mM NaCl, 2 mM EDTA, 0.1% Nonidet P-40, 2 mM dithiothreitol, 0.05% bovine serum albumin, and 5% glycerol (35) at 4°C for 1 h. The GSTfusion proteins were recovered using glutathione-Sepharose 4B resins (Amersham Pharmacia), and washed three times with PBS (pH 7.4). The 35 S-labeled proteins bound to the GST fusion proteins were analyzed by 10% SDS-PAGE and autoradiography. Coimmunoprecipitation experiments were performed following a procedure described previously (20) with some modification. Transfected cells were lysed in a buffer containing 50 mM HEPES (pH 7.4), 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 20 mM NaF, and 1 mM phenylmethylsulfonyl fluoride. The cell lysates were subjected to immunoprecipitation using appropriate antibodies. The immunoprecipitates were resolved by SDS-PAGE and analyzed by immunoblot probed with the indicated antibodies.

RESULTS
RIP Induces IKK␤ Activation-The death domain serine/ threonine kinase RIP plays a key role in the TNF-␣-induced activation of NF-B (29). TNF-␣ induces NF-B activation through IKK activation and the subsequent IKK-mediated phosphorylation of I-B. Phosphorylated I-B undergoes ubiquitin-mediated degradation (36). To understand better the role of RIP in the regulation of the NF-B pathway, we investigated the mechanism by which RIP mediates the TNF-␣-induced IKK activation. We transfected HEK293 cells with expression vectors producing HA-tagged RIP variants and FLAG-tagged IKK␣ or IKK␤. We then examined IKK␣ or IKK␤ activity in the transfected cells (Fig. 1). Because overexpression of RIP alone can induce apoptotic cell death (22), a caspase inhibitor crmA was coexpressed to prevent RIP-induced apoptosis. Ectopic expression of RIP or its mutants did not affect IKK␣ activity (Fig.  1B). In contrast, the ectopic RIP induced an increase in the IKK␤ activity in the transfected cells. mitogen-activated protein kinase kinase kinases such as NIK (13) or MEKK1 (14,37). We therefore examined the action of RIP on the upstream kinases of IKK␤, NIK, and MEKK1 ( Fig.  2A). Ectopic RIP induced the stimulation of NIK or MEKK1 activity for phosphorylating IKK␣. A kinase-inactive RIP mutant, RIP(K45R), was also able to activate NIK, but not MEKK1. This suggests that the kinase activity of RIP may be critical for MEKK1 activation, but not for NIK activation. Furthermore, the RIP-induced activation of IKK␤ was inhibited by a kinase-inactive MEKK1 mutant, MEKK1(K1253M), but not by a kinase-inactive NIK mutant, NIK(K293A/K294A) (Fig.  2B). These data suggest that MEKK1 may mediate the RIPinduced activation of IKK␤. The involvement of MEKK1 in the RIP-induced activation of IKK␤ was supported further by data showing that the RIP-induced stimulation of NF-B activity was repressed by coexpression of MEKK1(K1253 M) (Fig. 2C).
MEKK1 Associates Directly with RIP-To test whether RIP could physically associate with MEKK1, we carried out in vitro binding experiments using in vitro translated [ 35 S]methioninelabeled RIP and recombinant GST-MEKK1 variants that were immobilized on glutathione-agarose beads (Fig. 3A). 35 S-Labeled RIP physically associated with GST-MEKK1-M (amino acids 821-1172), which contains the conserved Cdc42/Rac interactive binding motif (38). We also carried out in vitro binding analyses using 35 S-labeled MEKK1 and recombinant GST-RIP fragments (Fig. 3B). The in vitro binding data indicate that the binding of RIP to MEKK1 occurred in the intermediate domain (ID) of RIP (Fig. 3B). Next, we examined whether RIP and MEKK1 would bind to each other in intact cells. HEK293 cells were cotransfected with full-length MEKK1-FLAG and HA-tagged RIP variants (Fig. 3C) or full-length HA-RIP and FLAG-tagged MEKK1 variants (Fig. 3D). The coimmunoprecipitation results indicated that MEKK1 physically interacted with RIP or RIP-ID but not with RIP lacking ID (RIP-⌬ID) (Fig.  3C). RIP bound to MEKK1 or MEKK1-M in intact cells (Fig.  3D). Furthermore, TNF-␣ treatment increased the interaction between RIP and MEKK1 in the cotransfected cells (Fig. 3E). The TNF-␣-dependent increase in the interaction between RIP and MEKK1 was reduced by overexpressed RIP-⌬ID. RIP-⌬ID contains a C-terminal death domain, which is involved in the interaction with TNFR-associated death domain during TNF-␣ signaling (20,21,27). We also examined the physical interaction between the two endogenous RIP and MEKK1 proteins in HEK293 cells. Immunoblot analysis of protein complexes immunoprecipitated with anti-MEKK1 antibody showed that en-dogenous RIP protein physically interacted with endogenous MEKK1 in intact cells and that this interaction was enhanced when the cells had been exposed to TNF-␣ (Fig. 3F).
RIP Phosphorylates MEKK1 at Ser-957 and Ser-994 -Our data in Figs. 2 and 3 suggested that MEKK1 functioned as a downstream signal of RIP and that MEKK1 was physically associated with RIP in intact cells. We therefore examined whether RIP could phosphorylate MEKK1. We used MEKK1(K1253M) to avoid the basal autophosphorylation of MEKK1. RIP and MEKK1(K1253M) were immunoprecipitated from TNF-␣-treated HEK293 cells, and in vitro phosphorylation of the immunopellets was tested (Fig. 4A). Interestingly, RIP enhanced the phosphorylation of MEKK1(K1253M), suggesting that RIP could phosphorylate MEKK1. The phosphorylation of MEKK1 by RIP was examined further using GST fusion proteins as substrates. RIP phosphorylated MEKK1(821-1172) but not NIK, IKK␣, or TRAF2 (Fig. 4B). Next, we constructed various MEKK1 fragments to map the in vitro phosphorylation sites of MEKK1(821-1172) (Fig. 4C). The in vitro phosphorylation study showed that MEKK1 phosphorylation by RIP occurred at a site(s) between amino acids 922 and 1021. MEKK1(922-1021) was then divided further into three fragments to identify the phosphorylation site(s). RIP-dependent in vitro phosphorylation was observed in two fragments: amino acids 953-981 and 982-1021 (Fig. 4C). Next, we replaced the serine or threonine residues in the two fragments with alanine and tested whether the mutations would prevent any phosphorylation by RIP (Fig. 4D). Our data demonstrate that RIP-dependent phosphorylation was abolished by the mutations of MEKK1 at Ser-957 and Ser-994. These findings strongly suggest that RIP phosphorylates MEKK1 at Ser-957 and Ser-994.
Phosphorylation of MEKK1 at Ser-957 and Ser-994 Is Critical for RIP-induced MEKK1 Activation-We examined further the effect of the RIP-induced MEKK1 phosphorylation on the MEKK1 activity (Fig. 6). Ectopically expressed RIP induced a stimulation of the MEKK1 activity phosphorylating its substrates GST-IKK␣(1-200) and GST-SEK1(K129R), whereas RIP(K45R) did not affect the MEKK1 activity (Fig. 6A). MEKK1(S957A/S994A) failed to phosphorylate the substrates regardless of the coexpression of RIP or RIP(K45R). The in-volvement of MEKK1 phosphorylation in RIP-induced MEKK1 activation is supported further by our results showing that TNF-␣ induced the stimulation of MEKK1, but not that of MEKK1(S957A/S994A) (Fig. 6B). These results suggest that the RIP-catalyzed MEKK1 phosphorylation at Ser-957 and Ser-994 is important for TNF-␣-induced MEKK1 activation.
TNF-␣-induced MEKK1 Activation Is Defective in RIP-deficient Jurkat Cells-To understand better the mechanism of the RIP-induced MEKK1 activation, we examined the phosphorylation of MEKK1 and the TNF-␣-stimulated MEKK1 activity in RIP(Ϫ) Jurkat cells (31) (Fig. 7). Our metabolic labeling study using 32 PO 4 demonstrated that TNF-␣ induced an increase in MEKK1 phosphorylation in wild type Jurkat cells but not in RIP(Ϫ) Jurkat cells (Fig. 7A). In comparison, UV irradiation resulted in an increase in MEKK1 phosphorylation both in wild type and RIP(Ϫ) Jurkat cells. These results suggest that RIP mediates the TNF-␣-induced, but not UV-induced, phosphorylation of endogenous MEKK1 in Jurkat cells. Furthermore, the phosphorylation of MEKK1 was recovered in RIP(Ϫ) Jurkat cells when the cells were transfected with a RIP construct (Fig.  7B). However, RIP did not increase the phosphorylation of MEKK1(S957A/S994A) in RIP(Ϫ) Jurkat cells. We also examined the effect of TNF-␣ on the endogenous MEKK activity in phosphorylating IKK␣ or SEK1 in wild type and RIP(Ϫ) Jurkat cells (Fig. 7C). The TNF-␣ treatment resulted in an increase in MEKK1 activity in wild type Jurkat cells but not in RIP(Ϫ) cells. Collectively, these data suggest that RIP is a major mediator of the TNF-␣-induced activation of MEKK1.

DISCUSSION
The death domain Ser/Thr kinase RIP has been shown to be a factor required for TNF-␣-induced NF-B activation (20,22). A study using thymocytes from RIP(Ϫ)-deficient mice has shown complete absence of TNF-␣-induced NF-B activation (29). RIP can activate IKKs through interaction with IKK␥/ NEMO, a regulatory subunit of the IKK complex (27,28), suggesting that RIP activates the NF-B signal pathway through the direct interaction with this IKK complex. However, the role of kinase activity of RIP in the TNF-␣-induced signaling cascades has not yet been clarified. In the present study, we show that RIP can phosphorylate MEKK1 and that the RIPmediated MEKK1 phosphorylation results in both ⍜KK␤ and NF-B activation.
We show in this study that RIP induces the activation of IKK␤ but not IKK␣ in transfected cells. The intermediate domain of RIP alone can also induce the IKK␤ activation. The IKK␤ activity in cells overexpressing wild type RIP appears to be higher than that of cells overexpressing the intermediate  (20,21,39). The oligomerization of RIP might induce a partial activation of the downstream signaling components, whereas the kinase activity might be required for maximal activation. This may reconcile our re-sults with previous reports stating that the kinase activity of RIP is not essential for RIP-induced NF-B activation (20,31).
We show here that MEKK1 mediates RIP-induced IKK␤ activation. IKK␤ can be phosphorylated and activated by NIK (16), MEKK1 (14), and Akt (40). MEKK1 preferentially activates IKK␤, whereas NIK activates IKK␣ more effectively (37). We demonstrated here that the kinase activity of RIP is required for the stimulation of MEKK1 but not for the stimulation of NIK. Involvement of MEKK1 in RIP-induced IKK␤ activation is supported further by the data showing that the kinase-inactive mutant, MEKK1(K1253M) suppressed the RIP-induced activation of IKK␤.
Our data suggest that RIP induces stimulation of the MEKK1-mediated NF-B signaling pathway via a direct interaction of RIP with MEKK1. It has been reported previously that both RIP and MEKK1 can interact with TRAF2 (17) RIP can phosphorylate MEKK1 at Ser-957 and Ser-994, resulting in MEKK1 stimulation. Interestingly, a replacement of one serine residue with alanine diminishes the phosphorylation of the other serine residue in MEKK1 by RIP. These findings imply that there might be a cross-regulatory interaction between the Ser-957 and Ser-994 residues in MEKK1, although the detailed mechanism underlying the MEKK1 phosphorylation by RIP needs to be investigated further. Both of the phosphorylation sites are followed by a proline residue, suggesting S/T-P as a putative RIP phosphorylation site. This motif has been also observed in sites phosphorylated by cyclindependent kinase (41) and MAP kinase (42). It suggests that the properties of the kinase activity of RIP might resemble those of cyclin-dependent kinase or MAP kinase. Interestingly, it has been reported recently that RIP2, one of the RIP kinase family members, possesses MAP kinase kinase activity (43), although RIP2 failed to phosphorylate MEKK1 in our experiments (data not shown).
RIP has been shown to mediate TNF-␣-induced signaling for apoptosis (22,44) as well as for cell survival via NF-B activation (29). RIP can function as a proapoptotic factor when it is cleaved by caspase-8 during TNF-␣-induced apoptosis (45,46). The proapoptotic C-terminal cleavage product of RIP in turn stimulates caspase-8 and caspase-3, although it attenuates the TNF-␣-induced IKK␤ activation (46). In addition, the caspase-8-cleaved RIP does not stimulate MEKK1 (data not shown). Recent studies suggest that MEKK1 is a negative regulator of apoptosis (47,48), whereas the caspase-3-cleaved MEKK1 fragment becomes proapoptotic (49,50). The cleavage of MEKK1 by caspase-3 separates the RIP-catalyzed phosphorylation sites (Ser-957 and Ser-994) from the C-terminal catalytic domain of MEKK1, which is a constitutively active form. Thus, the caspase-3-cleaved active form of MEKK1 cannot be phosphorylated by RIP. Therefore, the caspase-3-induced MEKK1 cleavage and the caspase-8-induced RIP cleavage may prevent RIP from stimulating MEKK1 activity for activation of the IKK␤-NF-B pathway. Furthermore, overexpression of MEKK1(S957A/S994A) enhances TNF-induced apoptotic cell death in HEK293 cells (data not shown). Collectively, our findings and those of others suggest that the RIP-mediated MEKK1 phosphorylation may be important for the cell survival mechanism of the IKK␤-NF-B pathway, whereas the caspase signaling cascades may terminate the IKK␤-NF-B signaling by cleaving both RIP and MEKK1.