TNF-induced activation of stress activated protein kinases (SAPKs, Jun NH2-terminal kinases) requires TNF receptor associated factor 2 (TRAF2). TRAF2 is a potent activator of a 95-kDa serine/threonine kinase termed germinal center kinase related (GCKR, also referred to as KHS1), which signals activation of the SAPK pathway. Consistent with a role for GCKR in TNF- induced SAPK activation, a kinase-inactive mutant of GCKR is a dominant negative inhibitor of TRAF2-induced SAPK activation. Here we show that TRAF2 interacts with GCKR. This interaction depended upon the TRAF domain of TRAF2 and the C-terminal 150 aa of GCKR. The full activation of GCKR by TRAF2 required the TRAF2 RING finger domain. TNF treatment of a T cell line, Jurkat, increased both GCRK and SAPK activity and enhanced the coimmunoprecipitation of GCKR with TRAF2. Similar results were found with the B cell line HS-Sultan. These findings are consistent with a model whereby TNF signaling results in the recruitment and activation of GCKR by TRAF2, which leads to SAPK activation.
Tumor necrosis factor is a cytokine secreted predominantly by activated macrophages and monocytes that has a major role in inflammation (reviewed in Ref. 1). TNF triggers the activation of the transcription factors NF-κB and AP-1, which in turn regulate many of the genes involved in the inflammatory response (reviewed in Refs. 2 and 3). AP-1 is a heterodimer consisting of the c-Jun transcription factor and either a member of the Fos or ATF family of transcription factors. Considerable progress has been made in understanding the signaling pathways that couple TNF to AP-1 and NF-κB activation.
Ligand-induced trimerization of the TNF receptor type 1 (TNFR1)3 results in the recruitment of the TNFR-associated death domain (TRADD) protein, which in turn binds two additional proteins, TNFR-associated factor-2 (TRAF2) and receptor interacting protein (RIP) (4, 5, 6, 7, 8). TRAF2 is a member of the TRAF family, a family of proteins that act as signal transducers for the TNFR family members. To date, six TRAF proteins have been isolated and designated as TRAF1 through TRAF6. All TRAF family members have related TRAF-N and TRAF-C domains in their carboxyl terminus. The TRAF-C domain is involved in binding to receptor tails and the formation of hetero- and homodimers between TRAF family members. TRAF2 as well as all the other TRAF members except TRAF1 have two amino-terminal zinc-binding domains, a RING finger and five zinc fingers. (7, 9, 10). In TNFR signaling, both TRAF2 and RIP mediate signals involved in activating NF-κB and AP-1 (11, 12, 13, 14). TRAF2 likely recruits and activates a serine/threonine kinase termed NF-κB-inducing kinase (NIK), which triggers a kinase cascade that leads to the translocation of NF-κB to the nucleus (15, 16, 17). TRAF2 also signals an extracellular signal-regulated kinase cascade that culminates in the activation of stress-activated protein kinases (SAPKs, also referred to as Jun NH2-terminal kinases or JNKs) (11, 12, 13). TRAF5 and TRAF6 also can activate the SAPK pathway. In turn, the SAPKs can phosphorylate c-Jun and ATF2 on residues important for their activation. Also, they phosphorylate Elk1, which interacts with the serum response factor and participates in c-fos activation (reviewed in Refs. 18 and 19). Demonstrating the importance of TRAF2 in TNF-induced SAPK activation, disruption of the TRAF2 gene abrogates TNF-induced AP-1 activation (20). However, the mechanism which links TRAF2 to activation of the SAPK pathway has been unclear.
Three highly related serine threonine kinases termed germinal center kinase (GCK), germinal center-like kinase (GLK), and GCK related (GCKR, also termed kinase homologous to STE20) have been found to specifically activate the SAPK kinase pathway and to be TNF responsive (21, 22, 23, 24, 25). Recently a fourth GCK-related kinase, HGK, has been found to be TNF responsive and to potentially function in TNF-induced SAPK activation (26). Directly implicating GCKR as a mediator of TNF-induced SAPK activation, a kinase-inactive mutant of GCKR significantly impaired TRAF2-induced SAPK activation, and, more importantly, a GCKR antisense construct had a similar effect (25). These results suggest the involvement of GCKR as well as other GCK family members in TNF-induced SAPK activation. Here, we have examined whether TRAF2 and GCKR interact and show that TNF signaling results in the recruitment of GCKR into a TRAF2-containing complex.
Materials and Methods
Cell lines, plasmids, and constructs
The 293T cell line was obtained from Dr. O. Witte (Los Angeles, CA) following permission from Dr. D. Baltimore (Pasadena, CA). 293, HS-Sultan, and Jurkat cells were obtained from the American Tissue Culture Collection (Manassas, VA). RT-PCR was used to generate a 486-bp fragment of TRAF2 from Jurkat cell RNA (primers, TCIGGICTA/GAAIGCA/GAT and GTICTIAAT/CCGIGAA/GGTIGA). This PCR fragment was used to obtain a full-length TRAF2 cDNA clone from a human activated T cell cDNA library, which was cloned into pMT2T. The construct, which contains the RING finger and zinc finger domains of TRAF2 (amino acids 1–225) was created by inserting a NheI-XhoI fragment of TRAF2 in pMT2T. The sequence upstream of the NheI site was replaced with a linker encoding the first 4 aa. The construct containing the RING finger domain of TRAF2 (amino acids 1–105) was generated by inserting an EcoRI-EagI fragment in pMT2T. The construct containing the zing finger domain of TRAF2 (amino acids 76–282) was generated by PCR. The dominant negative TRAF2 (amino acids 87–501) and the TRAF domain of TRAF2 (amino acids 271–501) constructs were created by PCR using pMT2T TRAF2 as template and cloned into pCR3.1. The hemagglutinin (HA)-tagged full-length GCKR and its different mutants (amino acids 1–691, 1–599, 1–493, 1–396, and 386–846) were obtained by PCR with the appropriated restriction sites incorporated into the primers and using GCKR cDNA clone as a template and pcDNA-HA as a vector. The HA-GCKR (amino acids 1–283) construct was generated by releasing a XbaI restriction fragment using an endogenous XbaI site in the GCKR cDNA and a XbaI site in the polylinker of the pcDNA-HA-GCKR construct. This fragment was subcloned into the XbaI site in pcDNA-HA.
Metabolic labeling and immunoprecipitation of different TRAF2 constructs
293 cells (3 × 106) were transfected with expression vectors that direct the expression of TRAF2 and truncated versions of TRAF2. Two days later, the cell culture media was replaced with media that lacked cysteine and methionine, but contained [35S]cysteine and [35S]methionine (50 μCi of 35S in vitro cell-labeling mix per milliliter of media; Amersham, Arlington Heights, IL). One hour later, the labeled cells were washed with cold PBS and lysed at 4°C in 1 ml of 1% Triton X-100 lysis buffer containing 25 mM HEPES, pH 7.5, 150 mM NaCl, 10% glycerol, 1 mM EDTA, and protease inhibitors (Roche Molecular Biochemicals, Indianapolis, IN). Nuclear and cell debris was removed by centrifugation at 14,000 rpm for 10 min at 4°C. Cell extracts were incubated for 2 h at 4°C with 5 μl of anti-TRAF2 antiserum, and the immunoprecipitates were collected with 20 μl of protein A-Sepharose. The TRAF2 polyclonal antiserum raised against a GST fusion protein that contained the RING, and a portion of the zing fingers of TRAF2 1–225(1–225) was used throughout these experiments. Beads were washed four times with lysis buffer, boiled in SDS sample buffer, and the supernatant were subjected to SDS-PAGE and autoradiography.
In vitro kinase assays, coimmunoprecipitation, and immunoblotting
293T cells were cotransfected by calcium phosphate/DNA precipitation with pcDNA-HA-GCKR (1 μg), pMT3-HA-SAPK-p46 (1 μg), and constructs that direct either wild-type TRAF2 or various truncation mutants (2 μg). Transfected DNA levels were equalized using empty plasmid. Thirty-six hours following the transfection of 293T cells, HA immunoprecipitates (HA-GCKR or HA-SAPK) were subjected to in vitro kinase assays using myelin basic protein (MBP; Sigma, St. Louis, MO) or c-Jun 1–79, (Santa Cruz Biotechnology, Santa Cruz, CA) as substrates, respectively, as previously described (21, 25). To examine endogenous kinase activity, GCKR and SAPK immunoprecipitates were subjected to similar in vitro kinase assays. For these experiments, GCKR- and SAPK-specific antisera were used (25). For the coimmunoprecipitations 2 × 106 transfected 293T cells were extracted in 1 ml of lysis buffer (20 mM Tris, pH 8.0, 137 mM NaCl, 2 mM EDTA, 1% Triton X-100, 1 mM sodium orthovanadate, 1 mM DTT, plus protease inhibitors) for 30 min on ice. The HA mAb (Babco, Richmond, CA) and the TRAF2 antiserum were used for the coimmunoprecipitation experiments in 293 cells. The TRAF2- and GCKR-specific antisera were used to perform the coimmunoprecipitation experiments in Jurkat and HS-Sultan cells (10 × 106 cells). The immunoprecipitates were collected with the appropriate anti-Ig Ab coupled to magnetic beads (Dynal, Oslo, Norway). The beads were washed six times with lysis buffer. The bound proteins were eluted in SDS-sample buffer, fractionated by SDS-PAGE, and transferred to nitrocellulose using standard methodology. The signals were detected by enhanced chemiluminescence (Amershan, Arlington Heights, IL). For the coimmunoprecipitation following TNF signaling, 293T cells that had been transfected with HA-GCKR 2 days previously, HS-Sultan, or Jurkat cells were exposed to TNF (100 ng/ml, Endogen, Woburn, MA) for 15 min before cell lysis.
Results and Discussion
Considerable evidence indicates that the TNF-induced NF-κB and SAPK activation requires the recruitment of TRAF2 to the TNFR. Simple overexpression of TRAF2 leads to both NF-κB and SAPK activation. However, the TNF-induced signaling pathway that leads to NF-κB and SAPK activation diverges at the level of TRAF2. TRAF2-mediated NF-κB activation can be inhibited by a dominant negative form of NIK, and NIK overexpression leads to NF-κB activation, but not SAPK activation (11, 12, 13, 15, 16, 17). Conversely, TRAF2-mediated SAPK activation can be inhibited by a dominant negative form of GCKR, while NF-κB activation is unaltered. Furthermore, GCKR overexpression leads to SAPK activation, but not NF-κB activation (25). In TNFR signaling, the TRAF2 molecule seems to serve a scaffolding function, recruiting and activating downstream effector molecules. As such, we were interested to determine whether GCKR is able to interact with TRAF2. To examine this possibility, we overexpressed a HA-tagged version of GCKR and TRAF2 in 293T cells and examined whether they coimmunoprecipitated. For the TRAF2 immunoprecipitations and immunoblotting, we used a rabbit anti-TRAF2 antisera raised against the RING finger and a portion of the zinc finger region. We found that the HA-GCKR immunoprecipitates contained TRAF2 and TRAF2 immunoprecipitates contained HA-GCKR (Fig. 1). We were unable to verify the presence of TRAF2 in the TRAF2 immunoprecipitates because the TRAF2 band merged with the H chain of Ig.
To map the site of interaction of TRAF2 with GCKR, we made a series of constructs that direct the expression of TRAF2 deletion mutants. First, we determined whether the truncated proteins were expressed and that the TRAF2 antiserum immunoprecipitated them. Cell lysates were prepared from 293T cells that had been in vivo labeled with [35S]methionine and [35S]cysteine following transfection with constructs that directed expression of TRAF2, TRAF2 1–225(1–225), TRAF2 1–105(1–105), TRAF2 76–282(76–282), and TRAF2 98–501(98–501). TRAF2 immunoprecipitates were collected and analyzed by SDS-PAGE and autoradiography (Fig. 2). A strong band for each of the mutant proteins was observed, confirming that they were expressed and recognized by the TRAF2 antiserum. Next, we coexpressed HA-GCKR along with the TRAF2 or the truncated forms of TRAF2 in 293T cells and prepared HA immunoprecipitates. Neither TRAF2 1–225(1–225), TRAF2 1–105(1–105), or TRAF2 76–282(76–282) coimmunoprecipitated with GCKR. However, TRAF2 96–501(96–501), which lacks the TRAF2 RING finger and behaves as a dominant negative for TNF-induced SAPK and GCKR activation, readily coimmunoprecipitated with GCKR (Fig. 3, left). This suggests that the TRAF2/GCKR interaction depend upon the presence of the TRAF domain in TRAF2. In contrast, the TRAF1 protein, which fails to activate the SAPK pathway, did not coimmunoprecipitate with GCKR (C.-S.S., unpublished observations).
We also analyzed the various TRAF2 mutant proteins for their ability to activate GCKR and the SAPK pathway. GCKR activation was assessed by an in vitro kinase assay, which measures the ability of immunoprecipitated epitope-tagged GCKR to phosphorylate MBP (25). Similarly, SAPK activation was assayed by examining the ability of SAPK to phosphorylate a recombinant N-terminal fragment of c-Jun (18, 19). 293T cells were transfected with constructs that direct the expression of wild-type TRAF2 and the truncated version. All of the TRAF2 mutant proteins including TRAF2 98–501(98–501) had significantly impaired abilities to activate GCKR, although the constructs that expressed either the RING or zinc finger domains had low levels of activity in 293T cells (Fig. 3, right). The small increases in the basal GCKR activity noted following expression of the TRAF2-truncated proteins were not observed when a similar experiment was performed in 293 cells, suggesting that it results from the high expression levels obtained in 293T cells (C.-S.S. unpublished observations). Nevertheless, these small increases in GCKR activity were insufficient to result in SAPK activation in vivo as all the mutants were markedly impaired in their ability to activate SAPK. Only the wild-type TRAF2 significantly increased both GCKR and SAPK activity levels.
A prediction of our model that GCKR activation occurs downstream of TRAF2 activation is that the TRAF2 dominant negative mutant, which impairs TNF-α-induced SAPK activation (25), should not inhibit GCKR-induced SAPK activation. To verify this prediction, we cotransfected constructs that direct the expression of GCKR, the TRAF2 dominant negative mutant as well with the other TRAF2 mutants, and HA-SAPK. We found that neither TRAF2 98–501(98–501) nor the other TRAF2 mutant proteins significantly impaired GCKR-induced SAPK activation (Fig. 4). We verified that SAPK, GCKR, and the TRAF2 mutant proteins were appropriately expressed by immunoblotting. Thus, GCKR acts downstream of TRAF2 in TNF-induced SAPK activation.
GCKR has an N-terminal catalytic domain and a large C-terminal regulatory domain, which has several proline-rich regions, one of which behaves as a CrkL interaction site.4 The C-terminal portion of GCK (amino acids 679–819) and a region between amino acids 270 and 329, which spans a PEST sequence, were both required for binding of TRAF2 to the GCK (27). To map the site of interaction of GCKR with TRAF2, we coexpressed various C-terminal-truncated HA-tagged GCKR proteins with wild-type TRAF2 and examined TRAF2 immunoprecipitates for the presence of HA-tagged GCKR proteins. While we readily detected full-length GCKR, we did not detect any of the truncation mutants despite adequate expression levels (Fig. 5). Thus, similar to GCK the C-terminal portion of GCKR is required for the interaction with TRAF2. The GCKR C-terminal region is well conserved with the corresponding regions in GLK and GCK (∼60% identity between GCKR, GCK, and GLK over the C-terminal 110 aa), suggesting that that all three kinases may use this region to interact with TRAF2 (Fig. 6).
To examine whether the TRAF domain of TRAF2 and the regulatory domain of GCKR were sufficient to observe an interaction with GCKR, we coexpressed an N-terminal truncation of GCKR, HA-GCKR 386–846(386–846), with TRAF2 272–501(272–501) in 293T cells. This truncated version of GCKR lacks the region corresponding to the first PEST-like sequence in GCK, which was necessary for GCK and TRAF2 to interact. We observed that HA-GCKR 386–846(386–846) strongly bound TRAF2 272–501(272–501), as each protein readily coimmunoprecipitated with the other (Fig. 7 A). Similar to the other TRAF2 mutants, TRAF2 272–501(272–501) did not appreciably activate either GCKR or SAPK (data not shown). Thus, the first PEST-like sequence in GCKR is not necessary for its interaction with the TRAF domain of TRAF2.
Next, we sought to determine whether TNF signaling altered the interaction between TRAF2 and GCKR. We expressed HA-GKCR in 293T cells and stimulated the cells with TNF or not, immunoprecipitated the endogenous TRAF2, and examined the immunoprecipitates for the presence of HA-GCKR. TNF treatment induced in a significant increase in the amount of GCKR coimmunoprecipitating with TRAF2 (Fig. 7 B). Attempts to visualize the recruitment of endogenous GCKR to endogenous TRAF2 in 293T cells were unsuccessful. We suspect this was because of the relatively low levels of GCKR in these cells.
Because GCKR is well expressed in both T and B lymphocytes, we sought to determine whether TNF exposure activated GCKR in lymphocyte cell lines. We stimulated Jurkat and HS-Sultan cells with TNF, immunoprecipitated GCKR and SAPK with specific antisera, and subjected the immunoprecipitates to in vitro kinase assays using MBP or c-Jun 1–79, respectively (Fig. 8). We found that GCKR was rapidly activated in both Jurkat and HS-Sultan following TNF exposure. TNF also induced SAPK activation in these two cell lines, although it was slightly delayed compared with GCKR activation. Next, we determined whether we could detect endogenous GCKR associated with endogenous TRAF2 following TNF signaling. We found low levels of GCKR immunoprecipitated with TRAF2 before TNF stimulation; however, following TNF stimulation we detected a marked increase in the amount of GCKR coimmunoprecipitating with TRAF2.
We conclude from these experiments that TNF stimulation results in the recruitment and activation of GCKR by TRAF2, which leads to SAPK activation. The TRAF domain of TRAF2 is required for the recruitment of GCKR. However, the interaction of the TRAF domain with GCKR is insufficient for high-level GCKR activation for which the TRAF2 RING and perhaps the zinc fingers are required. How the RING and zinc fingers of TRAF2 contribute to GCKR activation is unclear. They may alter the conformation of GCKR triggering GCKR autophosphorylation or perhaps they induce the dissociation of an inhibitor. Alternatively, they may be necessary for interaction with other proteins that have a role in GCKR activation. It will be of interest to determine whether GCKR and NIK compete for the same interaction site on TRAF2 and, in particular, whether the WKI motif defined in TRAF2 (16), which is important for its interaction with NIK, is required for the TRAF2-GCKR interaction. However, suggesting that GCKR and NIK may interact with separate sites on TRAF2, the overexpression of NIK failed to interfere with the coimmunoprecipitation of GCKR with TRAF2, and, conversely, the overexpression of GCKR failed to inhibit the coimmunoprecipitation of NIK with TRAF2 (C.-S.S., unpublished observations).
Overall, our findings are consistent with the previously described bifurcation of TNF-induced NF-κB and SAPK activation at the level of TRAF2 and support the model that TRAF2 functions as a docking protein for additional signaling molecules that trigger nonredundant signaling cascades (16, 17). Here, we have shown that GCKR and likely by analogy GCK and GLK function as a downstream effector of TRAF2 to trigger the SAPK pathway. Why three highly related kinases should all link TRAF2 to the SAPK pathway remains to be determined, although with further study variations in tissue distribution and subtle differences in their regulation are likely to emerge. In addition, how the GCK family of kinases and ASK1, which has also been shown to have a role in TNF-induced SAPK activation, interface remains to be determined (28). Like GCKR ASK1 is TNF-inducible and recruited to the TNF-receptor following TNF-signaling. The C-terminal portion of ASK1 is necessary for the interaction of ASK1 and TRAF2 (28), but it does not share a significant degree of homology with the C-terminal region of GCK, GCKR, and GLK (C.-S.S., unpublished observations). A catalytically inactive form of either GCKR or ASK1 impairs TNF- and TRAF2-induced SAPK activation in 293 cells. However, GCKR-induced SAPK activation is not impaired by the expression of the ASK1 mutant protein, indicating that ASK1 is unlikely downstream from GCKR in the TNF-induced SAPK activation (25). Targeted inactivation of the various GCK family members as well as ASK1 in mice will likely be required to formally evaluate the relative importance of these kinases in TNF-induced SAPK activation in vivo.
We thank Mary Rust for her editorial assistance and Dr. Anthony Fauci for his continued support.
Abbreviations used in this paper: TNFR1, TNF receptor type 1; TRADD, TNFR-associated death domain; TRAF, TNFR-associated factor; RIP, receptor interacting protein; NIK, NF-κB-inducing kinase; SAPK, stress-activated protein kinase; JNK, Jun NH2-terminal kinase; GCK, germinal center kinase; GLK, germinal center-like kinase; GCKR, GCK related; HA, hemagglutinin; MBP, myelin basic protein; PEST, proline glutamic acid serine threonine; ASK1, apoptosis signal-regulating kinase 1.
C.-S. Shi. Submitted for publication.