TRAF2 Phosphorylation Modulates Tumor Necrosis Factor Alpha-Induced Gene Expression and Cell Resistance to Apoptosis

Abstract

TRAF2 is an adaptor protein that regulates the activation of the c-Jun N-terminal kinase (JNK) and IκB kinase (IKK) signaling cascades in response to tumor necrosis factor alpha (TNF-α) stimulation. Although the downstream events in TNF-α signaling are better understood, the membrane-proximal events are still elusive. Here, we demonstrate that TNF-α and cellular stresses induce TRAF2 phosphorylation at serine 11 and that this phosphorylation is required for the expression of a subset of NF-κB target genes. Although TRAF2 phosphorylation had a minimal effect on the TNF-α-induced rapid and transient IKK activation, it was essential for secondary and prolonged IKK activation. Consistent with this, TRAF2 phosphorylation is not required for its recruitment to the TNFR1 complex in response to TNF-α stimulation but is required for its association with a cytoplasmic complex containing RIP1 and IKK. In addition, TRAF2 phosphorylation was essential for the full TNF-α-induced activation of JNK. Notably, TRAF2 phosphorylation increased both basal and inducible c-Jun and NF-κB activities and rendered cells resistant to stress-induced apoptosis. Moreover, TRAF2 was found to be constitutively phosphorylated in some lymphomas. These results unveil a new, finely tuned mechanism for TNF-α-induced IKK activation modulated by TRAF2 phosphorylation and suggest that TRAF2 phosphorylation contributes to elevated levels of basal NF-κB activity in certain human cancers.

Tumor necrosis factor (TNF) receptor (TNFR)-associated factors (TRAFs) are characterized by the presence of a TRAF domain at the C terminus. Currently, six members of this family are known (TRAR1 to TRAR6), and all TRAFs, except TRAF1, contain N-terminal RING finger domains followed by five or seven zinc finger motifs (5, 31). TRAF2 is a prototypical member of the TRAF family and regulates signals from TNFR superfamily members.

One of the well-characterized members of the TNFR superfamily is TNFR1. Currently, it is believed that TNFR1 activation by soluble TNF-α induces the sequential formation of two complexes with opposing effects on cell fate. In the first step, TNFR1 recruits the TNFR1-associated death domain (TRADD) protein, which in turn recruits TRAF2 and receptor-interacting protein 1 (RIP1) to form the membrane-bound prosurvival complex I. This leads to the sequential activation of mitogen-activated protein kinase (MAPK) kinase kinase (such as MEKK1/3), MAPK kinase (e.g., MKK4/7), and MAPK (e.g., c-Jun N-terminal kinase [JNK]) as well as in activation of transforming growth factor β-activated kinase 1 (TAK1), RIP1, and IκB kinase (IKK). In the next step, the TRADD/RIP1/TRAF2 complex dissociates from TNFR1 and associates with the Fas-associated death domain protein and caspase-8 to form the cytoplasmic proapoptotic complex II (23). The balance between these pathways is usually tipped by two events. One is the expression of cFLIP in response to NF-κB activation by complex I; cFLIP then binds to complex II and inhibits apoptosis by interfering with caspase-8 activation. The other is the TRAF2-mediated recruitment of inhibitor-of-apoptosis (IAP) family members (e.g., cIAP1/2), which inhibits apoptosis by interfering with the activation of the effector caspases that act downstream of the initiator caspases (18, 25). Thus, TNF-α triggers apoptosis only when new protein synthesis or signaling through the IKK pathway is inhibited.

JNK and IKK activate AP-1 (c-Jun/ATF2) and NF-κB transcription factors, respectively, and these transcription factors in turn induce the expression of genes that are involved in inflammation, the immune response, cell proliferation, and cell differentiation as well as genes that suppress death receptor- and stress-induced apoptosis (3, 30). Although the signaling mechanisms from IKK to NF-κB and from MKK4/7 to AP-1 are better understood, the receptor-proximal events that determine the TRAF2-dependent activation of IKK versus MKK4/7 still remain largely elusive. Genetic knockout studies have revealed that TRAF2 is essential for TNF-α-induced JNK, but not IKK, activation (36). Whereas TRAF5 null mouse embryonic fibroblasts (MEFs) respond normally with respect to the TNF-α-induced activation of JNK and IKK, TRAF2 and TRAF5 double-knockout (TRAF2/5 DKO) MEFs exhibit an almost complete defect in TNF-α-induced NF-κB activation (28). These studies suggest that TRAF2 and TRAF5 have redundant roles in TNF-α-induced IKK activation.

Using a functional in vitro screening assay, Deng et al. previously showed that TRAF6 acts as an E3 ligase and generates noncanonical K63-linked polyubiquitin chains, which appear to be necessary for the TRAF6-mediated activation of both the IKK and JNK signaling pathways (10). Recently, an increasing number of studies have shown that TRAF2 possesses E3 ligase activity and that it can catalyze the formation of noncanonical K63-linked polyubiquitin chains on itself as well as on its likely substrate, RIP1; those studies have also shown that the K63-linked polyubiquitin chains are essential for TNF-α-induced IKK and JNK activation (8, 13, 34). Ubc13/UEV1A is the only E2 ubiquitin-conjugating enzyme complex currently known to bind to TRAF2 and TRAF6 and to catalyze the K63-linked ubiquitination of these proteins (8). We have demonstrated that TRAF2's own RING-dependent self-ubiquitination causes its translocation to membrane rafts, resulting in the selective activation of JNK but not of IKK (14). In that same study, we also showed that the RNA interference-mediated knockdown of Ubc13 impairs TNF-α-induced JNK, but not IKK, activation (14). Consistent with these findings, the fusion of a membrane-targeting myristoylation peptide to either a TRAF3 or a TRAF2 whose RING domain had been deleted enabled its translocation to membrane rafts, resulting in the concomitant activation of the JNK, but not the IKK, signaling pathway (1, 9). Moreover, in both B cells and MEFs, the conditional ablation of Ubc13 resulted in considerably impaired JNK, but normal IKK, activation in response to a variety of stimuli. This suggests that either the K63-linked polyubiquitination of TRAF2 and TRAF6 is not essential for IKK activation or IKK can also be activated by an alternative, ubiquitination-independent mechanism (35).

TRAF2 has been reported to be phosphorylated at serine residues (7). However, the TRAF2 phosphorylation sites have not been identified. In order to gain a better understanding of the molecular mechanisms by which TRAF2 activates the IKK and JNK pathways, we set out to map TRAF2 phosphorylation sites using a classical phosphopeptide-mapping approach. Here, we report that TNF-α induces TRAF2 phosphorylation at Ser-11 and that this phosphorylation is essential for TNF-α-induced secondary and prolonged IKK activation and for the expression of a subset of NF-κB target genes.

MATERIALS AND METHODS

Cell lines, plasmids, and reagents.

HeLa, 293T, NIH 3T3, and MEF (wild type [WT] and TRAF2/5 DKO) cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% bovine calf serum and antibiotics. Antibodies and reagents were purchased as follows: anti-TRAF2, anti-JNK1, anti-IKKγ, anti-IKKβ, and anti-TNFR1 antibodies were obtained from Santa Cruz (Santa Cruz, CA); anti-phospho-JNK antibody was obtained from Promega (Madison WI); anti-Bcl-X_L and anti-IκBα antibodies were obtained from Cell Signaling (Danvers, MA); anti-cFLIP antibody was obtained from Upstate (Lake Placid, NY); anti-cIAP antibody was obtained from R&D Systems (Minneapolis, MN); mouse TNF-α (mTNF-α) and human TNF-α (hTNF-α) were obtained from Roche (Indianapolis, IN); anti-Flag antibody, hydroxyurea, and etoposide were obtained from Sigma (St. Louis, MO); and Halt cocktails of protease and phosphatase inhibitors were obtained from Pierce (Rockford, IL). Constructs encoding Flag-TRAF2 and the 2× NF-κB-Luc and Jun2-Luc reporter genes were described previously (14). Mutations in the Flag-TRAF2 expression vector were introduced using the QuikChange site-directed mutagenesis kit (Stratagene) and were confirmed by DNA sequencing. Various truncated forms of Flag-TRAF2 were generated by PCR amplification and the insertion of an amplicon into pCDNA3. Retroviral vectors for the transduction of Flag-TRAF2 constructs were generated by subcloning the TRAF2 cDNA into a pBabe-puro plasmid.

In vivo [³²P]orthophosphate labeling.

Cells cultured in six-well or 100-mm culture plates were transfected with 1.0 μg or 2.0 μg of Flag-TRAF2. Thirty-six hours after transfection, cells were incubated for 1 h in phosphate-free DMEM containing 10% dialyzed fetal bovine serum, followed by incubation for 90 min in the same medium supplemented with 0.5 mCi/ml of [³²P]orthophosphate. Cells were then washed with ice-cold Tris-buffered saline and lysed in TNE buffer (20 mM Tris-HCl [pH 7.5], 150 mM NaCl, 10 mM NaF, 1.0% NP-40, 2 mM EDTA, 1 mM dithiothreitol [DTT], 0.5 mM phenylmethylsulfonyl fluoride, 1× Halt protease, and phosphatase inhibitor cocktail) on ice for 30 min, followed by centrifugation in screw-cap tubes at 12,500 × g for 20 min at 4°C. ³²P-labeled Flag-TRAF2 was immunoprecipitated with anti-Flag antibody, separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE), and transferred onto a polyvinylidene difluoride (PVDF) membrane. The membrane was then stained with Ponceau S and exposed to X-ray film. The same membrane was then immunoblotted with anti-Flag antibody to visualize the level of Flag-TRAF2.

Two-dimensional separation of phosphoamino acids on a TLC plate.

293T cells (two 100-mm plates) were transfected with 2.0 μg of Flag-TRAF2 and labeled with [³²P]orthophosphate (1.0 mCi/ml) for 90 min. ³²P-labeled Flag-TRAF2 was immunoprecipitated, separated by SDS-PAGE, and transferred onto a PVDF membrane. The ³²P-labeled Flag-TRAF2 band was then excised and digested in 200 μl of 6 N HCl in a screw-cap tube at 110°C for 1 h, followed by SpeedVac drying for 2 h. Hydrolyzed and washed ³²P-labeled Flag-TRAF2 amino acids were dissolved in 8 μl of pH 1.9 buffer containing phosphoamino acid standards and separated (horizontally with pH 1.9 buffer and vertically with pH 3.5 buffer) on a glass-backed thin-layer cellulose chromatography (TLC) plate using a Hunter thin-layer peptide-mapping electrophoresis system (HTLE-7002; CBS Scientific, Del Mar, CA). The TLC plate was baked at 60°C for 20 min and sprayed with 0.25% (wt/vol) ninhydrin in acetone, followed by rebaking at 60°C for 15 min to visualize the position of phosphoamino acid standards. The TLC plate was then exposed to X-ray film for a week.

CNBr cleavage of ³²P-Flag-TRAF2.

³²P-labeled Flag-TRAF2 was immunopurified and separated by SDS-PAGE, and the gel slices containing the ³²P-labeled Flag-TRAF2 band was excised and homogenized. ³²P-labeled Flag-TRAF2 was then eluted in elution buffer (50 mM ammonium bicarbonate, 1.0% β-mercaptoethanol, and 1.0% SDS) by boiling for 10 min in a screw-cap tube. Eluate was then precipitated with 15% trichloroacetic acid, washed twice with acetone, dried, and digested in 200 μl of 0.25 M CNBr in 70% formic acid for 24 h at room temperature in a screw-cap tube that was completely covered with aluminum foil. The CNBr-cleaved ³²P-labeled Flag-TRAF2 was then dried by SpeedVac, washed twice with distilled water, and separated on a 15% Tris-Tricine gel. The gel was then transferred on to a 0.2-μm PVDF membrane, autoradiographied, and immunoblotted with anti-TRAF2 or anti-Flag antibody.

Luciferase reporter gene assays.

HeLa and MEF cells cultured in six-well plates were transfected with either an NF-κB or c-Jun firefly luciferase reporter plasmid (NF-κB-Luc or Jun2-Luc) (0.2 μg), together with a control Renilla luciferase reporter plasmid (pRL-TK) (0.01 μg) and WT or phosphomutant Flag-TRAF2 (0.2 μg) using Lipofectamine 2000 reagents according to the manufacturer's protocol. Thirty-six hours after transfection, test cells were treated with hTNF-α (10 ng/ml) or mTNF-α (5 ng/ml), and protein samples were prepared at 6 h (HeLa cells) or 4 h (MEFs) after treatment. The firefly and Renilla luciferase activities were then measured using a dual-luciferase assay system according to the manufacturer's instructions (Promega).

Preparation of retroviral supernatants and infection of TRAF2/5 DKO cells.

293T cells at 60 to 70% confluence were cotransfected with 2 μg of pMD.OGP (encoding Gag-Pol), 2 μg of pMD.G (encoding vesicular stomatitis virus G protein), and 2 μg of pBabe-puro-Flag-TRAF2 by a standard calcium phosphate precipitation method. Forty-eight hours after transfection, the viral supernatant was collected and filtered through a 0.45-μm filter. This supernatant was diluted two-, three-, four-, and fivefold with 10% fetal bovine serum-DMEM and then immediately used for infection of TRAF2/5 DKO MEFs in the presence of 4 μg/ml polybrene for 6 h. Forty-eight hours after infection, cells were selected with puromycin (2.0 μg/ml) for 6 days, and resistant cells were pooled. Cells that expressed Flag-TRAF2 at a physiological level (transduced with retroviral supernatant at a fourfold dilution) were used for the functional experiments within a month after selection (see Fig. S4 in the supplemental material).

JNK and IKK immunokinase assays.

MEFs were treated with mTNF-α (10 ng/ml), and protein samples were extracted using kinase lysis buffer (20 mM HEPES [pH 7.4], 350 mM NaCl, 0.5% Triton X-100, 1 mM DTT, 1 mM EDTA, 20% glycerol, and a cocktail of protease and phosphatase inhibitors). Endogenous JNK1 or IKK complexes were immunoprecipitated using anti-JNK1 or anti-IKKγ antibody and then subjected to in vitro kinase assays in which glutathione S-transferase (GST)-Jun^1-87 (for JNK) or GST-IκBα^1-55 (for IKK) served as a substrate, as described previously (14).

Phosphoantibody and immunoblotting.

Phosphopeptide (VTPPGpS¹¹LELLQC) synthesis, rabbit immunization, and antibody purification were performed by Abgent Envision Proteomics (San Diego, CA). For the detection of TRAF2 phosphorylation, cells were treated as indicated above, and protein samples were extracted with TNE lysis buffer (20 mM Tris-HCl [pH 7.5], 150 mM NaCl, 10 mM NaF, 1.0% NP-40, 2 mM EDTA, 1 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride, 1× Halt cocktails of protease and phosphatase inhibitors) for 30 min on ice. Thirty micrograms of cleared lysates was separated by SDS-PAGE and transferred onto nitrocellulose membranes. The blots were blocked with Tween 20-Tris-buffered saline containing 3% bovine serum albumin for 4 h and incubated with TRAF2 phosphoantibody overnight at 4°C. The phosphorylation status of TRAF2 was then assessed using horseradish peroxidase-labeled secondary antibody and ECL solution. The same membranes were then stripped and reprobed with anti-TRAF2 antibody.

Real-time PCR.

MEF cells were treated with mTNF-α (10 ng/ml), and total RNA was prepared using the RNeasy minikit (Qiagen). Five micrograms of total RNA was treated with RQ1 RNase-free DNase for 30 min at 37°C and then reverse transcribed using an oligo(dT) primer. The resulting cDNA was subjected to quantitative real-time PCR using the Power Sybr green AB master mix and an ABI Prism 7700 sequence detector (Applied Biosystems). Mouse glyceraldehyde-3-phosphate dehydrogenase (GAPDH)-specific primers were used to generate an internal control, and the average threshold cycle for samples in triplicate was used in subsequent calculations. Relative expression levels of NF-κB target genes were calculated as the ratio with respect to GAPDH levels. The means ± standard errors for four independent experiments were considered to be statistically significant at a P value of <0.05. Real-time PCR products were also separated on an agarose gel to confirm the presence of single bands (see Fig. S7A in the supplemental material).

RESULTS

TRAF2 phosphorylation occurs on N-terminal serine residues.

To identify TRAF2 phosphorylation sites, we first analyzed the phosphorylation of exogenously expressed Flag-TRAF2 in 293T and NIH 3T3 cells by an in vivo [³²P]orthophosphate labeling method. In 293T cells cultured under standard culture conditions, Flag-TRAF2 was highly phosphorylated, and TNF-α treatment did not elevate TRAF2 phosphorylation further (Fig. 1A). In contrast, in NIH 3T3 cells, TNF-α treatment led to a greater-than-twofold increase in levels of TRAF2 phosphorylation (Fig. 1A). We first attempted to map the TRAF2 phosphorylation sites by purifying Flag-TRAF2 expressed in 293T cells and subjecting it to liquid chromatography-tandem mass spectrometry analysis. However, our attempts failed repeatedly, as most of the TRAF2 peptides were not detected by liquid chromatography-tandem mass spectrometry (data not shown). Therefore, we proceeded to perform classical in vivo [³²P]orthophosphate labeling and phosphopeptide-mapping experiments. First, ³²P-labeled and purified Flag-TRAF2 from 293T cells was hydrolyzed by treatment with 6 N HCl, and amino acids were then separated on a glass-backed TLC plate by two-dimensional electrophoresis. The migration of TRAF2-derived phosphoamino acids showed that TRAF2 phosphorylation in 293T cells takes place on serine residues (Fig. 1B). The same experiments revealed that TRAF2 is also phosphorylated on serine residues in NIH 3T3 cells before and after TNF-α stimulation (data not shown).

FIG. 1.

TRAF2 is phosphorylated at serine residues of its N-terminal region. (A) 293T or NIH 3T3 cells were transfected with 2.0 μg of Flag-TRAF2, labeled with [³²P]orthophosphate, and mock treated or treated with TNF-α for 10 min. ³²P-labeled Flag-TRAF2 was immunoprecipitated, separated by SDS-PAGE, and transferred onto a PVDF membrane. The membrane was then stained with India ink (bottom) and exposed to X-ray film for 6 h (top). IgG, immunoglobulin G. (B) Purified ³²P-labeled Flag-TRAF2 was hydrolyzed in 6 N HCl, after which its amino acids were separated on a TLC plate by two-dimensional electrophoresis and exposed to X-ray film for 7 days. (C) Purified ³²P-labeled Flag-TRAF2 was digested with CNBr, after which its peptides were separated on a Tris-Tricine gel, transferred onto a PVDF membrane, and exposed to X-ray film for 3 days. The membrane was then sequentially immunoblotted (IB) (with stripping between each probe exposure) with three distinct antibodies: anti-TRAF2 H249 antibody (that recognizes the TRAF2 N terminus), anti-TRAF2 C-20 antibody (that recognizes the TRAF2 C terminus), and anti-Flag antibody. (D) Diagram showing the fragments that could theoretically be generated by CNBr cleavage of Flag-TRAF2. (E) NIH 3T3 cells were transfected with Flag-TRAF2, labeled with [³²P]orthophosphate, and pretreated with staurosporine (0.2 μM) or AG18 (50 μM), as indicated, for 1 h before being treated with mTNF-α (10 ng/ml) for 15 min. ³²P-labeled Flag-TRAF2 was then immunoprecipitated, separated by SDS-PAGE, and transferred onto a PVDF membrane. The membrane was then exposed to X-ray film overnight (top). The membrane was subsequently probed with anti-Flag antibody (bottom). (F) NIH 3T3 cells were cotransfected with Flag-TRAF2 and either a constitutively active form of Akt (Myr-Akt), PKCα (CA-PKCα), or PKCɛ (CA-PKCɛ) or a dominant negative form of PKCα (DN-PKCα) or PKCɛ (DN-PKCɛ). Thirty-six hours after transfection, the phosphorylation states of Flag-TRAF2 were detected as described above (E).

To identify the regions that contain these phosphoserines, purified ³²P-labeled Flag-TRAF2 was digested with CNBr, which should theoretically generate 11 peptides (Fig. 1D). The peptides produced were separated on a Tris-Tricine gel, transferred onto a PVDF membrane, and exposed to X-ray film (Fig. 1C). The same membrane was sequentially immunoblotted with a polyclonal antibody (H249; Santa Cruz) that recognizes the TRAF2 N terminus, a polyclonal antibody (C20; Santa Cruz) that recognizes the TRAF2 C terminus, and an anti-Flag antibody that recognizes the N-terminal Flag epitope (Fig. 1C), with stripping between each application of antibody. The results revealed that TRAF2 phosphorylation occurs on the N-terminal region encompassing amino acids (aa) 1 to 246.

TRAF2 is phosphorylated at Ser-11.

Computational analysis using the NetPhos 2.0 program revealed that TRAF2 contains several consensus phosphomotifs for protein kinase C (PKC)-, CKI-, CKII-, GSK3β-, PKA-, and Akt-mediated phosphorylation (data not shown). Also, in vivo [³²P]orthophosphate labeling in the presence or absence of pharmacological inhibitors revealed that staurosporine, a general inhibitor of PKC isoforms, completely inhibits both basal and inducible TRAF2 phosphorylation in NIH 3T3 cells, whereas AG18, a pan-tyrosine kinase inhibitor, has a minimal effect on TRAF2 phosphorylation (Fig. 1E). Consistent with this finding, the coexpression of a constitutively active form of PKCα (CA-PKCα) or Akt (Myr-Akt) increased TRAF2 phosphorylation in NIH 3T3 cells (Fig. 1F).

For the sake of simplicity, we mutated all the consensus serine residues within the TRAF2 N-terminal region at aa 1 to 246. Surprisingly, none of the single mutations inhibited TRAF2 phosphorylation (see Fig. S1A in the supplemental material), nor did double and triple mutant combinations (see Fig. S1B in the supplemental material). To clearly identify the TRAF2 phosphorylation domain, we constructed several new expression vectors encoding truncated forms of TRAF2. As shown in Fig. 2A, the deletion of 86 aa from the N terminus of TRAF2 almost completely abolished TRAF2 phosphorylation in 293T cells (lane T2-87-501). In line with this, a TRAF2 N-terminal fragment comprising aa 1 to 128 was strongly phosphorylated in vivo (Fig. 2B, lane T2-1-128). As the TRAF2-1-128 fragment is relatively short and strongly phosphorylated in vivo, we next mutated all the consensus serine residues in this fragment and analyzed their phosphorylation in vivo. Surprisingly, we again failed to detect a decrease in TRAF2 phosphorylation for any mutant (see Fig. S2A in the supplemental material). Our analysis up to this point had left only three serine residues (Ser-11, -20, and -64) in the TRAF2-1-128 fragment untested (see Fig. S2B in the supplemental material). We next mutated these three serines in the Flag-TRAF2-1-128 fragment and found that the mutation of Ser-11 significantly decreased TRAF2 phosphorylation in vivo (Fig. 2C). The mutation of Ser-11 in full-length TRAF2 confirmed that Ser-11 is the major phosphorylation site (Fig. 2D).

FIG. 2.

TRAF2 is phosphorylated at serine 11. (A) The full-length form (T2-1-501) and various truncated forms (T2-1-352, T2-87-501, T2-230-501, and T2-309-501) of Flag-TRAF2 were expressed in 293T cells, and their phosphorylation was assessed as described in the legend of Fig. 1A. (B) Shorter fragments of Flag-TRAF2 (T2-1-128, T2-352-501, and T2-87-230) were expressed in 293T cells, and their phosphorylation was assessed as described in the legend of Fig. 1E. (C) WT and mutant Flag-TRAF2-1-128 fragments were expressed in 293T cells, and their phosphorylation was assessed as described in the legend of Fig. 1E. (D) WT and mutant forms of full-length Flag-TRAF2 were expressed in 293T cells, and their phosphorylation was assessed as described in the legend of Fig. 1E.

TRAF2 phosphorylation of Ser-11 increases basal and inducible c-Jun and NF-κB activities.

To assess the role of TRAF2 phosphorylation in TNF-α-induced c-Jun and NF-κB activation, we generated phosphomutant TRAF2 plasmids: TRAF2-S11A, in which Ser-11 is mutated to alanine to abolish phosphorylation, and TRAF2-S11D, in which Ser-11 is mutated to aspartic acid to mimic phosphorylation. As shown in Fig. S3A and S3B in the supplemental material, whereas the expression of TRAF2-S11A in HeLa cells decreased TNF-α-induced NF-κB and c-Jun activation by 25 to 30% relative to that in TRAF2-WT-transfected cells, the expression of TRAF2-S11D increased NF-κB and c-Jun activities by 20 to 30%. To test these constructs in the absence of interference from endogenous TRAF2 and TRAF5, we performed luciferase reporter gene assays in TRAF2/5 DKO MEFs as well as in WT MEFs. Consistent with the results from HeLa cells, the expression of TRAF2-S11A in WT MEFs partially decreased NF-κB and c-Jun activities relative to the levels exhibited in cells transfected with TRAF2-WT, whereas the expression of TRAF2-S11D partially increased the basal and inducible NF-κB and c-Jun activities (Fig. 3A and B). Notably, in TRAF2/5 DKO MEFs, TRAF2-S11D expression significantly increased both the basal and inducible NF-κB and c-Jun activities compared to that of TRAF2-S11A expression (Fig. 3C and D). These data suggest that TRAF2 phosphorylation does contribute to the TNF-α-induced activation of NF-κB and c-Jun.

FIG. 3.

TRAF2 phosphorylation increases NF-κB and c-Jun activities. (A to D) WT and TRAF2/5 DKO MEFs were cotransfected with NF-κB-Luc or Jun2-Luc, pRL-TK and pCDNA3, TRAF2-WT, TRAF2-S11A, or TRAF2-S11D. Thirty-six hours after transfection, cells were treated with or without mTNF-α (5 ng/ml) for 4 h, after which the NF-κB-Luc or Jun2-Luc activity was measured and normalized to pRL-TK activity. Data shown are the means ± standard deviations of three experiments that were done in triplicate. An “*” indicates a P value of <0.05. (E and F) TRAF2/5 DKO cells reconstituted with TRAF2-WT (pBa-T2-WT), TRAF2-S11A (pBa-T2-S11A), or TRAF2-S11D (pBa-T2-S11D) were treated with mTNF-α (10 ng/ml) as indicated. The IKK complex or JNK1 was immunoprecipitated with anti-IKKγ or anti-JNK1 antibody and subjected to in vitro kinase assays in which GST-IκBα^1-55 served as a substrate for IKK and GST-Jun^1-87 served as a substrate for JNK1. Reaction mixtures were separated by SDS-PAGE, transferred onto a nitrocellulose membrane, and exposed to X-ray film for 6 h (³²p-G-IκBα or ³²p-G-jun). The same membrane was stained with Ponceau S (G-IκBα or G-jun) and then immunoblotted with anti-IKKβ or anti-JNK1 antibody (IKKβ or JNK1).

TRAF2 phosphorylation is essential for the second phase of TNF-α-induced IKK activation.

To examine the role of TRAF2 phosphorylation in TNF-α-induced JNK and IKK activation under physiological conditions, we established TRAF2/5 DKO cell lines that stably express Flag-TRAF2-WT (pBa-T2-WT), Flag-TRAF2-S11A (pBa-T2-S11A), or Flag-TRAF2-S11D (pBa-T2-S11D) at physiological levels (see Fig. S4 in the supplemental material). Immunofluorescence staining confirmed that over 95% of the transfected cells expressed TRAF2 (data not shown). To avoid the change in TRAF2 expression levels due to possible artificial selection during cell culture, we performed all experiments within a month after the establishment of these cell lines. As expected, the expression of TRAF2-WT in TRAF2/5 DKO cells restored TNF-α-induced immediate/transient as well as secondary/prolonged IKK activation (Fig. 3E). In contrast, the expression of TRAF2-S11A restored immediate but not secondary IKK activation (Fig. 3E, compare lanes 9 and 10 with lanes 4 and 5). This suggests that TRAF2 phosphorylation at Ser-11 is essential for TNF-α-induced secondary, but not immediate, IKK activation, which explains why the expression of TRAF2-S11A partially inhibits NF-κB activity. Intriguingly, the expression of TRAF2-S11D not only increased basal and inducible IKK activity but also altered the oscillation of IKK activity following TNF-α stimulation (Fig. 3E, compare lanes 3 to 5 with lanes 13 to 15). This further supports the notion that TRAF2 phosphorylation at Ser-11 regulates secondary IKK activation. Similarly, the expression of TRAF2-S11A decreased TNF-α-induced JNK activation, whereas the expression of TRAF2-S11D increased it (Fig. 3F). However, we did not observe the clear oscillation of JNK activity that we saw for IKK activity in MEFs following TNF-α stimulation. In vitro IKK and JNK kinase assays were repeated three times, and average kinase activities are summarized in Fig. S5 in the supplemental material. In a separate experiment, we established pBa-T2-WT and pBa-T2-S11A cell lines using retroviral supernatants that were prepared independently and examined JNK1 phosphorylation and IκBα degradation by Western blotting. As shown in Fig. S6 in the supplemental material, the expression of TRAF2-S11A attenuated but did not completely inhibit TNF-α-induced JNK1 phosphorylation and IκBα degradation.

TRAF2 phosphorylation is essential for TNF-α-induced expression of a subset of NF-κB target genes.

To further assess the role of TRAF2 phosphorylation in TNF-α-induced NF-κB activation, we analyzed the expression of several NF-κB target genes in pBa-T2-WT, pBa-T2-S11A, and pBa-T2-S11D cell lines by real-time reverse transcription (RT)-PCR. As shown in Fig. 4, although the expression levels of IκBα, IP-10, and Mn-superoxide dismutase did not differ substantially among these cell lines, either before or after TNF-α stimulation, the expression levels of ICAM-1, RANTES, interleukin-6 (IL-6), and Cox-2 were significantly dampened in pBa-T2-S11A cells at the 3- and 6-h time points compared to those in pBa-T2-WT and pBa-T2-S11D cells. These data indicate that TRAF2 phosphorylation is essential for the efficient expression of certain NF-κB target genes. To confirm our results, we designed a second set of primers (see Table S2 in the supplemental material) for these NF-κB target genes and performed conventional semiquantitative RT-PCR analysis. Consistent with the real-time RT-PCR results, this conventional RT-PCR analysis also showed that the mutation of Ser-11 significantly decreases the TNF-α-induced expression of RANTES, ICAM-1, IL-6, and COX-2 (see Fig. S7B in the supplemental material).

FIG. 4.

TRAF2 phosphorylation is essential for TNF-α-induced expression of RANTES, ICAM-1, IL-6, and COX-2 in MEFs. pBa-T2-WT, pBa-T2-S11A, and pBa-T2-S11D cells were treated with mTNF-α (10 ng/ml) as indicated, and the expression levels of IκBα, IP-10, RANTES, ICAM-1, IL-6, Mn-superoxide dismutase (SOD), and COX-2 were determined by real-time PCR. The relative expression level of each gene is presented as the ratio between it and a reference gene, the GAPDH gene, as an average from four independent experiments. An * represents a P value of <0.05.

TRAF2 phosphorylation protects cells from stress-induced cell death by increasing the expression of antiapoptotic proteins.

TNF-α stimulation of TRAF2/5 DKO cells has been reported to cause the accumulation of reactive oxygen species and prolong JNK activation, both of which ultimately lead to necrotic and apoptotic cell death (26). We also observed that treatment with TNF-α alone causes cell death within 48 h in over 90% of TRAF2/5 DKO MEFs. The stable expression of TRAF2-WT, TRAF2-S11A, or TRAF2-S11D in TRAF2/5 DKO MEFs completely inhibited TNF-α-induced cell death (Fig. 5A), indicating that the phosphorylation of TRAF2 is not required for its inhibition of TNF-α-induced cell death. On the other hand, the stable expression of TRAF2-S11A in TRAF2/5 DKO cells sensitized the cells to H₂O₂-induced cell death (Fig. 5B). As expected, pBa-T2-S11D cells exhibited significant resistance to H₂O₂-induced cell death compared to pBa-T2-S11A cells. Colony formation assays also revealed that pBa-T2-S11D cells are more resistant and that pBa-T2-S11A cells are more sensitive to hydroxyurea- and H₂O₂-induced apoptosis than are pBa-T2-WT cells (Fig. 5C and see Fig. S8 in the supplemental material). Although pBa-T2-S11D cells were also found to be more resistant to etoposide-induced apoptosis than were their pBa-T2-S11A counterparts, this difference was not statistically significant. Bcl-X_L, cIAP1, and cFLIP are among the antiapoptotic proteins whose expressions are regulated by NF-κB activity (2, 6). Western blot analysis revealed that the TNF-α-induced expression of these proteins was reduced in pBa-T2-S11A cells compared to that in pBa-T2-S11D cells. Notably, the basal expression levels of these proteins were about twofold elevated in pBa-T2-S11D cells compared to levels in pBa-T2-S11A cells (Fig. 5D). Together, these data suggest that TRAF2 phosphorylation protects cells from apoptosis trigged by certain types of stress by increasing the levels of expression of antiapoptotic proteins.

FIG. 5.

TRAF2 phosphorylation inhibits stress-induced cell death. (A and B) WT, TRAF2/5 DKO, pBa-T2-WT, pBa-T2-S11A, and pBa-T2-S11D cells were treated with mTNF-α (10 ng/ml) or H₂O₂ (0.075 mM) as indicated. At 24 or 48 h after treatment, total cell death was assessed via a trypan blue exclusion assay, and data shown represent the averages of three experiments performed in triplicate. An * indicates a P value of <0.05. (C) pBa-T2-WT, pBa-T2-S11A, and pBa-T2-S11D cells cultured in six-well plates (500 cells/well) were left untreated or treated with H₂O₂ (0.1 mM), etoposide (5 μM), or hydroxyurea (0.4 mM) for 6 h. Fourteen days later, colonies containing more than 50 cells were counted. The averages are presented as means ± standard deviations. An * indicates a P value of <0.05. (D) TRAF2/5 DKO MEFs reconstituted with Flag-TRAF2-WT (pBa-TRAF2), Flag-TRAF2-S11A (pBa-T2-S11A), or Flag-TRAF2-S11D (pBa-T2-S11D) were treated with mTNF-α (10 ng/ml) for 1 and 3 h as indicated, and the levels of expression of Bcl-X_L, cIAP1, and cFLIP were monitored by Western blotting using the corresponding antibodies. Relative expression levels of these proteins under unstimulated conditions were quantified by densitometry (Dens.) and normalized to β-actin.

TNF-α and cellular stresses induce TRAF2 phosphorylation at Ser-11.

To analyze TNF-α-induced TRAF2 phosphorylation, we generated a phosphoantibody (pTRAF2-Ser¹¹) using a synthetic phosphopeptide as an antigen. As shown in Fig. 6A, pTRAF2-Ser¹¹ recognized TRAF2-WT but not TRAF2-S11A expressed in NIH 3T3 cells. Unexpectedly, the transient expression of TRAF2 in NIH 3T3 or 293T (data not shown) cells resulted in its constitutive phosphorylation at Ser-11, and this phosphorylation was not increased further by TNF-α stimulation or by the coexpression of CA-PKCα or Myr-Akt (Fig. 6A). Treatment of immunoprecipitated Flag-TRAF2-WT with calf intestinal alkaline phosphatase completely blocked the recognition of TRAF2-WT by pTRAF2-Ser¹¹ antibody (see Fig. S8 in the supplemental material), confirming that this antibody recognizes only phosphorylated TRAF2. In contrast to transiently expressed TRAF2, endogenous TRAF2 in HeLa cells was not phosphorylated when the cells were cultured under standard conditions, and stimulation with TNF-α immediately induced TRAF2 phosphorylation, which peaked 30 min after stimulation (Fig. 6B). TNF-α stimulation also induced TRAF2 phosphorylation in pBa-T2-WT cells, with kinetics similar to those in HeLa cells (Fig. 6B).

FIG. 6.

TNF-α induces TRAF2 phosphorylation, and such phosphorylation occurs constitutively in FaDu and HL cells. (A) Flag-TRAF2-WT or Flag-TRAF2-S11A was cotransfected into NIH 3T3 cells with Myr-Akt or CA-PKCα. At 36 h after transfection, cells were treated with or without mTNF-α (10 ng/ml) as indicated, and TRAF2 phosphorylation was detected by Western blotting using a phospho-specific antibody (pTRAF2-Ser¹¹). The same membrane was then stripped and reprobed with anti-Flag antibody. (B and C) HeLa, FaDu, PC3, and MDA-MB-231 cells as well as TRAF2/5 DKO MEFs reconstituted with Flag-TRAF2 (pBa-TRAF2) or empty vector (pBa-C) were treated with mTNF-α (10 ng/ml) or hTNF-α (20 ng/ml) as indicated, and the phosphorylation of TRAF2 was monitored by Western blotting using TRAF2 phosphoantibody. (D) Protein samples from HL, tonsils, and HeLa cells treated with or without hTNF-α were separated by SDS-PAGE, and the phosphorylation of TRAF2 was monitored by Western blotting using TRAF2 phosphoantibody. (E) HeLa cells were left untreated or treated with cytokine (hTNF-α), growth factor (IGF-I), tumor promoter (TPA), or a DNA-damaging agent (etoposide, hydroxyurea, camptothecin, or UV) as indicated, and TRAF2 phosphorylation was monitored by Western blotting using phosphoantibody. (F) HeLa cells were treated with or without 30 J/m² of UVC, and cells were harvested at the indicated time points. TRAF2 phosphorylation was then monitored by Western blotting using phosphoantibody.

TRAF2 is constitutively phosphorylated in FaDu cells and HL.

The human head and neck squamous cell carcinoma cell line FaDu has been reported to exhibit elevated basal NF-κB activity, and this elevated NF-κB activity seems to be induced by the TNFR1/TRADD/TRAF2/RIP1 signaling pathway (17). We thus wanted to assess the correlation between TRAF2 phosphorylation and NF-κB activation in cancer cells. To this end, we first examined the phosphorylation of TRAF2 in well-established cancer cell lines, including FaDu, PC3 (prostate cancer), and MDA-MB-231 (breast cancer) cells. As shown in Fig. 6C, TRAF2 was constitutively phosphorylated in FaDu cells, and the treatment of these cells with TNF-α led to a further increase in TRAF2 phosphorylation. Interestingly, TRAF2 was highly expressed in FaDu and MDA-MB-231 cells compared to HeLa and PC3 cells (Fig. 6C). NF-κB is also constitutively activated in Hodgkin's lymphoma (HL) and Hodgkin/Reed-Sternberg cell lines, and the expression of dominant negative TRAF2 in Hodgkin/Reed-Sternberg cells decreases basal NF-κB activity and sensitizes them to TNF-α-induced cell death (16). Thus, to assess the correlation between TRAF2 phosphorylation and NF-κB activation in tumor tissues, we examined TRAF2 phosphorylation in HL tissues from patients. As shown in Fig. 6D, TRAF2 was constitutively phosphorylated in five out of six HL samples and not at all in normal tonsil samples. To further examine the phosphorylation status of TRAF2 in other normal and cancer tissues, we examined four additional normal samples (two tonsils and two thyroids) and four non-HLs (two chronic lymphocytic leukemias and two small lymphocytic lymphomas). Interestingly, TRAF2 was constitutively phosphorylated in three out of four non-HLs (see Fig. S9B in the supplemental material). Unexpectedly, TRAF2 was also constitutively phosphorylated in one of two thyroids. At present, it is not clear why TRAF2 is constitutively phosphorylated in some normal thyroids. Nevertheless, these data demonstrate that constitutive TRAF2 phosphorylation was very common in both HLs and non-HLs. As the stable expression of TRAF2-S11D increases both basal and inducible NF-κB activity in TRAF2/5 DKO cells, the constitutive phosphorylation of TRAF2 in FaDu cells and human lymphomas implicates TRAF2 phosphorylation in the elevation of basal NF-κB activity in some cancers.

Cellular stresses also induce TRAF2 phosphorylation at Ser-11.

Apart from inflammatory cytokines, growth factors and many types of stress activate NF-κB. To examine TRAF2 phosphorylation in response to additional types of stimuli, we treated HeLa cells with growth factors (insulin-like growth factor I [IGF-I] and 12-O-tetradecanoylphorbol-13-acetate), DNA-damaging agents (etoposide, hydroxyurea, camptothecin, and UV), H₂O₂ (oxidative stress), and tunicamycin (endoplasmic reticulum stress) and monitored TRAF2 phosphorylation by Western blotting. As shown in Fig. 6E and in Fig. S10A in the supplemental material, IGF-I and TPA induced TRAF2 phosphorylation only very weakly, whereas all stress-inducing agents used in these experiments clearly induced TRAF2 phosphorylation, albeit at various levels. Notably, UV induced TRAF2 phosphorylation as strongly as TNF-α did. Time course analysis revealed that UV-induced TRAF2 phosphorylation takes place within 5 min and reaches a peak at 30 min (Fig. 6F), a pattern that resembles that of TNF-α-induced TRAF2 phosphorylation.

TRAF2 phosphorylation occurs in both the TNFR1-associated and cytoplasmic complexes.

TRAF2 is a cytoplasmic protein, and a subpopulation is recruited to the TNFR1 signaling complex upon TNF-α stimulation (31). To determine where in cells TRAF2 is phosphorylated, we stimulated HeLa cells with TNF-α and UV and immunoprecipitated the TNFR1 complex with anti-TNFR1 antibody. As shown in Fig. 7A, TRAF2 was recruited to and phosphorylated in the TNFR1 complex upon TNF-α stimulation. The levels of TRAF2 protein associated with TNFR1 at the 10- and 30-min time points were comparable; however, TRAF2 phosphorylation was clearly higher at the 30-min time point, suggesting that TRAF2 is phosphorylated after it is recruited to TNFR1. Although UV treatment also efficiently induced TRAF2 phosphorylation, TRAF2 was not found in the TNFR1 complex after UV treatment. These data suggest that TRAF2 can be phosphorylated in the TNFR1 complex in response to TNF-α stimulation and in the cytoplasm in response to UV treatment.

FIG. 7.

TRAF2 phosphorylation stabilizes the TRAF2/RIP1/IKK complex. (A) HeLa cells were mock treated or treated with hTNF-α or UV as indicated, and the TNFR1 complexes were immunoprecipitated (IP) with anti-TNFR1 antibody. The phosphorylation of TRAF2 within the receptor complex as well as in the cytoplasm was monitored by Western blotting using TRAF2 phosphoantibody. IgG, immunoglobulin G. (B) pBa-T2-WT and pBa-T2-S11A cells were treated with or without mTNF-α (10 ng/ml), and the TNFR1 complexes were immunoprecipitated with anti-TNFR1 antibody. The recruitment of Flag-TRAF2, RIP1, and IKK to the receptor complex was monitored by Western blotting with the corresponding antibodies. (C) TRAD2/5 DKO cells reconstituted with Flag-TRAF2-WT (pBa-T2-WT) or Flag-TRAF2-S11A (pBa-T2-S11A) were treated with or without mTNF-α (10 ng/ml) as indicated. Flag-TRAF2 was then immunoprecipitated with anti-Flag antibody, and the recruitment of TNFR1, RIP1, and IKK to TRAF2 was monitored by Western blotting with the corresponding antibodies. (D) Proposed model. Upon TNF-α treatment, TRAF2 is recruited to the membrane-bound complex I, which contains TNFR1, TRADD, RIP1, and IKK. There, it induces rapid IKK activation, and this primary IKK activation does not require TRAF2 phosphorylation. Thereafter, TRADD, TRAF2, RIP1, and IKK dissociate from TNFR1 as a complex to form cytoplasmic complex II. TRAF2 induces the secondary phase of IKK activation in complex II, which is dependent on TRAF2 phosphorylation at Ser-11.

TRAF2 phosphorylation is essential for its association with complex II.

TRAF2-mediated RIP1 ubiquitination is currently thought to play a critical role in TNF-α-induced IKK activation (8). To examine the role of TRAF2 phosphorylation in TNF-α-induced RIP1 ubiquitination, we extensively analyzed the ubiquitination of RIP1 in pBa-T2-WT and pBa-T2-S11A cells both by immunoprecipitating TNFR1 and immunoblotting with anti-RIP1 antibody and by immunoprecipitating RIP1 and immunoblotting with antiubiquitin antibody. However, we did not observe any difference in RIP1 ubiquitination between pBa-T2-WT and pBa-T2-S11A cells (Fig. 7B and data not shown). Both RIP1 and IKK were equally recruited to TNFR1 within 10 min after TNF-α stimulation in these cells (Fig. 7B). However, when TRAF2 was immunoprecipitated, we repeatedly observed that TRAF2-WT remained associated with RIP1 and IKK until 120 min after TNF-α stimulation, whereas TRAF2-S11A associated with RIP1 and IKK at early time points (10 and 30 min) but dissociated from RIP1 and IKK at later time points (Fig. 7C). TNFR1 was present in both complexes pulled down by TRAF2-WT and TRAF2-S11A 10 min poststimulation but dissociated from the complexes 30 min after stimulation. The complexes that were pulled down by TRAF2-WT at later time points contained RIP1 and IKK but were devoid of TNFR1, indicating that these complexes correspond to cytoplasmic complex II (23). This suggests that TRAF2 phosphorylation is required for its association with complex II in cytoplasm. Kinetically, this association correlates very well with the second phase of IKK activation, which is dependent upon TRAF2 phosphorylation. Collectively, these data suggest that TRAF2 phosphorylation is not essential for the TNF-α-induced recruitment of TRAF2, RIP1, and IKK to the TNFR1-containing complex (complex I) at the plasma membrane but that it is required for the retention of TRAF2 in complex II and the subsequent induction of the second phase of IKK activation.

DISCUSSION

TRAF2 has been reported to be phosphorylated at Thr-117 following TNF-α stimulation (22). However, in spite of our extensive phosphoamino acid analysis of TRAF2, before and after TNF-α stimulation as well as after the coexpression of Flag-TRAF2 with Myr-Akt or CA-PKC in 293T and NIH 3T3 cells (Fig. 1B and data not shown), only serine was found to be phosphorylated.

TRAF2 phosphorylation at Ser-11 is unexpected; no phosphoprediction software available on the internet predicted a direct phosphorylation of TRAF2 at Ser-11 by any kinase. Although the coexpression of TRAF2 with either CA-PKCα or Myr-Akt1 increased overall TRAF2 phosphorylation as assessed by ³²P_i labeling in NIH 3T3 cells (Fig. 1F), Western blot analysis using our Ser-11-specific phosphoantibody revealed that neither CA-PKCα nor Myr-Akt1 increased TRAF2 phosphorylation at Ser-11 under the same conditions (Fig. 6A). An in vitro kinase assay revealed that purified PKC does phosphorylate the purified GST-TRAF2-1-128 fusion protein; however, the mutation of Ser-11 to alanine did not attenuate TRAF2 phosphorylation by PKC (see Fig. S10B in the supplemental material). This suggests that PKC and Akt1 phosphorylate TRAF2 at different sites. Interestingly, the transient expression of TRAF2 in NIH 3T3 cells resulted in its constitutive phosphorylation at Ser-11, and this phosphorylation was not increased by TNF-α stimulation, although TNF-α did induce endogenous TRAF2 phosphorylation at Ser-11 in HeLa cells (Fig. 6A and B). This suggests that either transient overexpression causes nonspecific TRAF2 phosphorylation or DNA transfection itself activates a kinase involved in TRAF2 phosphorylation at Ser11. Another possibility is that the kinases responsible for TRAF2 phosphorylation are constitutively active in some cells. We are currently conducting a series of experiments to identify the kinases that mediate TRAF2 phosphorylation at Ser-11.

Ligation of TNFR1 initiates the formation of membrane-bound complex I, which consists of at least TNFR1, TRADD, RIP1, TRAF2, and IKK. This complex I induces the immediate activation of the JNK and NF-κB signaling pathways. Thereafter, the TRADD/RIP1/TRAF2/IKK complex dissociates from TNFR1 and recruits the Fas-associated death domain protein and caspase-8 to form cytoplasmic complex II (23). TRAF2 and RIP1 can be recruited to complex I independently, whereas IKK recruitment to complex I and its full activation require the presence of both TRAF2 and RIP1 (11). With respect to the phosphorylation of IκBα, membrane-bound IKK was reported to be less effective than the IKK recovered from the whole-cell lysates, indicating that the IKK complex retains kinase activity when it is dissociated from TNFR1 (37). The TRAF2 RING domain plays an essential role in IKK recruitment to TNFR1; TRAF2-deficient MEFs reconstituted with the RING domain-deleted form of TRAF2 fail to recruit IKK to TNFR1 in response to TNF-α treatment (11). TRAF2 Ser-11 is adjacent to the TRAF2 RING domain, raising the possibility that TRAF2 Ser-11 phosphorylation might regulate IKK recruitment to TNFR1 upon TNF-α stimulation. However, we did not observe any difference between TRAF2-WT- and TRAF2-S11A-expressing cells with respect to RIP1 and IKK recruitment to TNFR1 (Fig. 7B). On the other hand, analysis of TRAF2 association with RIP1 and IKK in response to TNF-α stimulation revealed that whereas TRAF2-WT remained associated with RIP1 and IKK (complexes I and II) until 120 min after TNF-α stimulation, TRAF2-S11A associates with RIP1 and IKK at early time points (complex I) but quickly dissociates from them (complex II) at later time points (Fig. 7B and C). Although complex II has been regarded as an apoptosis-inducing complex, it does not cause cell death in WT cells unless protein synthesis or the NF-κB pathway is blocked. Thus, the physiological function of complex II has been unclear. TRAF2-S11A-expressing cells are still sensitive to TNF-α-induced cell death in the presence of cycloheximide (data not shown), indicating that TRAF2 phosphorylation is not required for TNF-α-induced complex II formation. Likewise, TRAF2 expression is not essential for complex II formation, as TNF-α-induced cell death is not impaired but is augmented in TRAF2-deficient cells. This suggests that TRAF2 phosphorylation is essential for its association with complex II but not with complex I. Kinetically, TRAF2 association with complex II correlates well with the TRAF2 phosphorylation-dependent induction of secondary IKK activation. This suggests that the physiological function of complex II is most likely to trigger the second phase of IKK activation in a TRAF2 phosphorylation-dependent manner. However, we cannot rule out the possibility that TRAF2 phosphorylation induces the formation of a new complex (containing TRAF2, RIP1, and IKK) in the cytoplasm that differs from complex II.

Ligand-induced receptor internalization and the biphasic activation of downstream signaling pathways have been reported for many types of plasma membrane receptors. For example, numerous studies have shown that the internalized and endosomally localized epidermal growth factor receptor associates with SHC and GRB2 and that these interactions lead to a prolonged activation of the MAPK cascade (21). A study of ASK1-deficient MEFs revealed that ASK1 is essential for sustained but not immediate JNK activation by TNF-α, suggesting that even sustained JNK activation is subjected to regulation (29). Together, these published findings and the data presented here suggest that the phosphorylation of TRAF2 at Ser-11 mediates the second phase of IKK activation in response to TNF-α stimulation and that it does so by retaining TRAF2 itself in complex II.

NF-κB transactivates the expression of more than 300 genes, and the expression of these genes is controlled by fine-tuned mechanisms that are regulated by the strength and duration of the NF-κB activity (15). Analysis of the expression of well-known NF-κB target genes by real-time PCR revealed that TRAF2 phosphorylation is not essential for TNF-α-induced expression of IκBα and IP-10 (Fig. 4). The expression of these genes by TNF-α is very quick, peaking within 1 h after stimulation. This suggests that these genes have NF-κB-sensitive promoters and that even a slight increase in NF-κB activity drives their expression to maximal levels in a short time period. Therefore, the transient activation of IKK that occurs in the absence of TRAF2 phosphorylation is sufficient to trigger the efficient expression of these genes. On the other hand, the expression of RANTES, ICAM-1, IL-6, and COX-2 in TRAF2-S11A-expressing cells was significantly dampened compared to that in TRAF2-WT-expressing cells (Fig. 4). The expression of these genes after TNF-α stimulation was relatively slow, reaching a peak at 3 h or continuously increasing over the time course tested (e.g., RANTS). These genes seem to require prolonged NF-κB activity for their efficient expression. This suggests that the second phase of IKK activation regulated by TRAF2 phosphorylation plays a critical role in the TNF-α-induced expression of a subset of NF-κB target genes.

A variety of stimuli other than inflammatory cytokines also activate NF-κB. These include growth factors, B- and T-cell receptors, double-stranded RNA, and DNA-damaging agents. Several serine/threonine kinases such as Akt, PKCα, PKCβ, PKCθ, PKR, and DNA-PK have been implicated in NF-κB activation in response to these stimuli (3, 12, 18, 27). Although these stimuli activate the same NF-κB pathway, they nevertheless induce different patterns of gene expression and thus elicit different cellular responses (15, 33). For example, PKCζ and TRAF2-associated kinase (T2K/TBK1/NAK) have both been implicated in TNF-α-induced IKK activation (24). However, the targeted disruption of PKCζ results in reduced NF-κB activation without abolishing TNF-α-induced IKK activation (20). Although a T2K knockout likewise has no effect on TNF-α-induced immediate IKK activation, it significantly reduces TNF-α-induced expression of NF-κB target genes such as TLR2 and ICAM-1 (4). Also, as shown by Kuai et al., T2K is recruited to the TNFR1 complex in a TNF-α-dependent manner, and its expression is essential for the TNF-α-induced expression of RANTES (19). Collectively, these studies suggest that although these serine/threonine kinases do not directly activate IKK, they are involved in the regulation of NF-κB activity and in the expression of certain NF-κB target genes. Here, we demonstrate that TRAF2 phosphorylation at Ser-11 is essential for the TNF-α-induced secondary activation of IKK and also for the expression of some NF-κB target genes such as RANTES and ICAM-I. This suggests that TRAF2 phosphorylation represents a new layer of pathway and that it regulates the expression of NF-κB target genes by linking certain serine/threonine kinases to secondary IKK activation. It will next be crucial to identify the kinases that phosphorylate TRAF2 at Ser-11 as well as to map the sites of phosphorylation by CA-PKCα and Myr-Akt.

TRAF2 recruits antiapoptotic cIAP1 and cIAP2 to the TNFR1 signaling complex through its C-terminal TRAF-N domain and thereby inhibits the activation of effector caspases by TNF-α (25). TRAF2/5 DKO cells are very sensitive to TNF-α-induced cell death, which has been proposed to be due to an accumulation of reactive oxygen species and to prolonged JNK activation as a consequence of impaired NF-κB activation (26). In the study presented here, we also found that TRAF2 DKO cells are sensitive to TNF-α-induced cell death (Fig. 5A). The stable expression of WT or phosphomutant TRAF2 completely inhibited TNF-α-induced cell death, suggesting that TRAF2 phosphorylation is not required for its inhibition of TNF-α-induced cell death. On the other hand, TRAF2-S11D-expressing cells displayed significant resistance to the induction of apoptosis by H₂O₂ and hydroxyurea in comparison to TRAF2-S11A-expressing cells (Fig. 5B to D). Inducible NF-κB activation is essential for a normal immune response, but persistent NF-κB activation causes chronic inflammatory disease and contributes to cancer development, metastasis, and resistance to drug-induced apoptosis (3, 32). Although elevated levels of basal NF-κB activity have been found in many types of cancers (18), the mechanisms that underlie its activity in cancer cells are not yet defined clearly. Our study shows that basal and inducible NF-κB activities are elevated in TRAF2-S11D-expressing cells and that these cells are resistant to stress-induced apoptosis (Fig. 5). In line with this, basal expression levels of antiapoptotic proteins, such as Bcl-X_L, cIAP1, and cFLIP, are also elevated about twofold in TRAF2-S11D-expressing cells compared to TRAF2-S11A-expressing cells (Fig. 5D). We also show that TRAF2 phosphorylation at Ser-11 is induced not only by TNF-α but also by DNA-damaging agents (Fig. 6E). Moreover, TRAF2 is constitutively phosphorylated in FaDu cells and some human lymphoma tissues that exhibit constitutive NF-κB activities. Collectively, these data suggest that TRAF2 phosphorylation is one of the events that are responsible for the elevated basal NF-κB activity in some human cancers.

In sum, our data provide evidence for the existence of a new layer of regulation for the TNF-α-induced activation of the JNK and IKK signaling pathways and that this depends on TRAF2 phosphorylation at Ser-11. This phosphorylation induces secondary IKK activation in response to TNF-α stimulation by retaining TRAF2 itself in complex II. Importantly, we show that this secondary phase of IKK activation is essential for the efficient expression of certain NF-κB target genes and for cellular resistance to stress-induced cell death.