Research ArticleImmunology

Two Mechanistically and Temporally Distinct NF-κB Activation Pathways in IL-1 Signaling

Science Signaling  20 Oct 2009:
Vol. 2, Issue 93, pp. ra66
DOI: 10.1126/scisignal.2000387

Abstract

The cytokine interleukin-1 (IL-1) mediates immune and inflammatory responses by activating the transcription factor nuclear factor κB (NF-κB). Although transforming growth factor–β–activated kinase 1 (TAK1) and mitogen-activated protein kinase (MAPK) kinase kinase 3 (MEKK3) are both crucial for IL-1–dependent activation of NF-κB, their potential functional and physical interactions remain unclear. Here, we showed that TAK1-mediated activation of NF-κB required the transient formation of a signaling complex that included tumor necrosis factor receptor–associated factor 6 (TRAF6), MEKK3, and TAK1. Site-specific, lysine 63–linked polyubiquitination of TAK1 at lysine 209, likely catalyzed by TRAF6 and Ubc13, was required for the formation of this complex. After TAK1-mediated activation of NF-κB, TRAF6 subsequently activated NF-κB through MEKK3 independently of TAK1, thereby establishing continuous activation of NF-κB, which was required for the production of sufficient cytokines. Therefore, we propose that the cooperative activation of NF-κB by two mechanistically and temporally distinct MEKK3-dependent pathways that diverge at TRAF6 critically contributes to immune and inflammatory systems.

Introduction

Interleukin-1 (IL-1) is one of the principal cytokines responsible for the induction of mediators that orchestrate various immune and inflammatory responses by activating the transcription factors nuclear factor κB (NF-κB) and activating protein 1 (AP-1) (1). IL-1 signaling is initiated by ligand–induced formation of a receptor complex that consists of the IL-1 receptor (IL-1R) and IL-1R accessory protein (IL-1RAcp). Subsequently, the cytosolic adaptor protein myeloid differentiation marker 88 (MyD88) is recruited to this complex (2). MyD88, in turn, recruits IL-1R–associated kinase–1 (IRAK1) and IRAK4 through interactions with their death domains (3). IRAK1 is presumably activated through its phosphorylation by IRAK4 (4) and interacts with tumor necrosis factor (TNF) receptor–associated factor 6 (TRAF6) (5). TRAF6 then moves from the membrane to the cytosol, where it associates with transforming growth factor–β (TGF-β)–activated kinase 1 (TAK1), a member of the mitogen-activated protein kinase (MAPK) kinase kinase (MAP3K, or MEKK) family (6). TRAF6, which contains a really interesting new gene (RING) domain, functions as an E3 ubiquitin ligase that conjugates Lys63-linked polyubiquitin chains to TRAF6 itself, to IRAK1, and to NF-κB essential modulator [NEMO, also known as inhibitor of NF-κB (IκB) kinase γ (IKKγ)], a subunit of the IKK complex (79). Preferential binding of conserved zinc finger domains in TAK1-binding protein 2 (TAB2) and TAB3 to the polyubiquitin chains of TRAF6 and the binding of NEMO to the polyubiquitination chains of IRAK1 are thought to result in stimulation-dependent recruitment of the TAK1-TAB2-TAB3 and IKK complexes to TRAF6. This then leads to the activation of TAK1 and the IKK complex (911).

Several lines of evidence have shown that another MAP3K, MEKK3, functions downstream of TRAF6 to activate the IKK complex, p38 MAPK, and c-Jun N-terminal kinase (JNK), consequently leading to the activation of NF-κB and AP-1 (12, 13). Despite their indispensability in IL-1 signaling, the mechanisms by which TAK1 and MEKK3 are activated by TRAF6 and whether they physically and functionally interact with each other remain unclear. Here, we show critical roles for TAK1 and MEKK3 in IL-1–induced TRAF6 signaling and propose a model for IL-1 signaling in which two mechanistically and temporally distinct MEKK3-dependent pathways that diverge at TRAF6 cooperatively activate NF-κB to produce proinflammatory cytokines.

Results

TAK1 is ubiquitinated at Lys209 in response to IL-1

TRAF6 reportedly catalyzes Lys63-linked polyubiquitination of TAK1 at Lys34 in response to TGF-β, which leads to the activation of p38 and JNK (14). However, whether TAK1 is ubiquitinated in response to physiological stimuli that activate NF-κB and, if so, how the ubiquitination of TAK1 is involved in signal transduction remains unclear. To address this question, we first determined the ubiquitin linkage of polyubiquitin chains on TAK1 and assessed whether polyubiquitination was facilitated by TRAF6. Human embryonic kidney (HEK) 293T cells were transfected with various combinations of expression vectors encoding FLAG-tagged TAK1 (FLAG-TAK1), FLAG-TRAF6, or hemagglutinin (HA)-tagged ubiquitin (Ub). TAK1 was immunoprecipitated after the cell lysates were boiled to avoid coimmunoprecipitation of any proteins noncovalently attached to TAK1, and samples were analyzed by Western blotting with an antibody against HA (Fig. 1A). Polyubiquitination of TAK1 was dramatically facilitated by TRAF6, but not by T6ΔR, a ligase-deficient mutant of TRAF6, and was also observed when the ubiquitin mutant K63, which contains only one lysine residue at position 63, was used. However, polyubiquitination of TAK1 was markedly diminished when the ubiquitin mutant R63, which is unable to form polyubiquitin chains at Lys63, was used. Furthermore, polyubiquitination of TAK1 was eliminated in the presence of the protein cylindromatosis (CYLD), which removes Lys63-linked polyubiquitin chains (15).

Fig. 1

IL-1 induces Lys63-linked polyubiquitination of TAK1 at Lys209. (A) TRAF6 facilitates Lys63-linked polyubiquitination of TAK1. HEK 293T cells were transfected with the indicated expression plasmids. Cell lysates were first boiled to remove noncovalently attached proteins and then subjected to immunoprecipitation (IP) with an antibody against TAK1. Ubiquitination of TAK1 was detected with an antibody against HA. Whole-cell lysates (WCLS) were analyzed for the presence of TRAF6 and CYLD. (B) IL-1 induces Lys63-linked polyubiquitination of TAK1. WT-MEFs stably transfected with a plasmid expressing HA-K63Ub were stimulated with IL-1 (10 ng/ml). Ubiquitination of TAK1 was analyzed as described in (A). (C) TRAF6 is required for the IL-1–induced polyubiquitination of TAK1. Traf6−/− MEFs and those expressing exogenous TRAF6-WT were treated with IL-1 (10 ng/ml). Cell lysates were boiled and then immunoprecipitation of TAK1 was performed. Ubiquitination of TAK1 was detected with an antibody against Ub. (D) Lys209 of TAK1 is the predominant polyubiquitination site. HEK 293T cells were transfected with the indicated expression plasmids. Cell lysates were subjected to immunoprecipitation, followed by Western blotting analysis as described in (A). (E) IL-1 induces Lys63-linked polyubiquitination of TAK1 at Lys209. Tak1−/− MEFs expressing HA-K63Ub and exogenous TAK1-WT or one of the TAK1 KR mutants were stimulated with IL-1. Ubiquitination of TAK1 was analyzed as described in (A). The results shown are representative of three independent experiments.

We then addressed whether endogenous TAK1 was conjugated to a Lys63-linked polyubiquitin chain after a physiological stimulus. When wild-type mouse embryonic fibroblasts (MEFs) that contained HA-K63Ub were stimulated with IL-1, endogenous TAK1 became polyubiquitinated (Fig. 1B). Moreover, IL-1–induced polyubiquitination of TAK1 required TRAF6 (Fig. 1C). Therefore, the polyubiquitin chains on TAK1 were primarily linked through Lys63 of ubiquitin, and the generation of polyubiquitin chains was induced by IL-1–mediated activation of TRAF6.

To identify the role of the ubiquitination of TAK1, we first mapped the polyubiquitination site. Alignment of the amino acid sequences of TAK1 from six different species highlighted seven conserved lysine residues at positions 52, 150, 158, 172, 209, 257, and 520 of human and mouse TAK1 (fig. S1). Each of these lysine residues was then mutated to arginine to generate KR mutants, and the extent of TRAF6-induced Lys63-linked polyubiquitination of these TAK1 mutants was tested. TAK1-K209R was less ubiquitinated than the other mutants (Fig. 1D). To assess whether Lys209 was required for the polyubiquitination of TAK1 under physiological conditions, we analyzed the extent of IL-1–stimulated ubiquitination of TAK1 in Tak1−/− MEFs retrovirally transduced with either wild-type TAK1 (TAK1-WT) or one of the KR mutants of TAK1. (Hereafter, Tak1−/− MEFs that were transduced with the TAK1-WT–encoding virus are named TAK1-WT-MEFs and Tak1−/− MEFs that were transduced with the TAK1-K209R–encoded virus are named K209R-MEFs.) The K209R mutation abolished IL-1–induced, Lys63-linked polyubiquitination of TAK1, whereas mutations at other lysine residues had little effect (Fig. 1E and fig. S2). These results indicate that Lys209 was the primary polyubiquitination site on TAK1 in response to IL-1.

Polyubiquitination of TAK1 is involved in the IL-1–induced activation of NF-κB

To determine whether polyubiquitination of TAK1 was required for the activation of TAK1 and IKK, Tak1−/− MEFs transduced with retroviruses expressing TAK1-WT or the KR mutants of TAK1 were stimulated with IL-1. Phosphorylation of TAK1 and degradation of IκBα were severely reduced in K209R-MEFs compared to those in TAK1-WT-MEFs but were barely affected in K172R-MEFs (Fig. 2A). The antibody against phosphorylated TAK1 (pTAK1) recognized TAK1 phosphorylated at Thr187, a critical phosphorylation site for the activation of TAK1. Moreover, the DNA binding activity of NF-κB was largely inhibited by the K209R mutant TAK1 (Fig. 2B). Lys63-linked polyubiquitination of TAK1 was also induced when cells were stimulated with TNF-α (Fig. 2C). The K209R mutant TAK1 abolished both TNF-α– and lipopolysaccharide (LPS)-induced activation of NF-κB as judged by analyses of the phosphorylation of IκBα (Fig. 2D) and the DNA binding activity of NF-κB (Fig. 2E). In contrast, the K34R mutation in TAK1 affected neither IL-1– nor TNF-α–induced activation of NF-κB (Fig. 2, F and G), although polyubiquitination of TAK1 at Lys34 is essential for TGF-β–induced activation of TAK1 (14). Moreover, the K209R mutation, but not the K34R mutation, impaired the IL-1– or LPS-induced expression of Il6, an NF-κB target gene (Fig. 2H), and resulted in TNF-α–induced apoptosis, which occurs when TNF-α–induced activation of NF-κB is blocked (Fig. 2I). Together, these results indicated that polyubiquitination of TAK1 at Lys209 was involved in the activation of TAK1 and in TAK1-induced activation of IKK in response to IL-1, TNF-α, and LPS.

Fig. 2

Polyubiquitination of TAK1 at Lys209 is required for the activation of TAK1 in response to IL-1, TNF-α, and LPS. (A and B) IL-1–induced activation of TAK1 and NF-κB is reduced by the K209R mutation. Tak1−/− MEFs transduced with retroviruses expressing TAK1-WT or the indicated TAK1 KR mutants were treated with IL-1 (10 ng/ml). Western blotting (A) and EMSA (B) analyses were performed. A filled circle in (A) denotes nonfunctional TAK1Δ expressed in Tak1−/− MEFs (33). (C) TNF-α induces Lys63-linked polyubiquitination of TAK1. WT-MEFs transduced with a retrovirus expressing HA-K63Ub were stimulated with TNF-α (10 ng/ml). Cell lysates were boiled and analyzed for the ubiquitination of TAK1. (D and E) TNF-α- and LPS-induced activation of NF-κB is impaired by the K209R mutation. TAK1-WT-MEFs or K209-MEFs were treated with TNF-α or LPS (2 μg/ml). Western blotting (D) and EMSA (E) analyses were performed. (F and G) IL-1– or TNF-α–induced activation of NF-κB is not affected by the K34R mutation. TAK1-WT-MEFs or K34R-MEFs [(F), right panel] were treated with IL-1 (upper left) or TNF-α (lower left). Western blotting [(F), left panel] and EMSA (G) analyses were performed. (H and I) The K209R, but not the K34R, mutation impairs the expression of NF-κB target gene. TAK1-WT-MEFs, K209-MEFs, or K34R-MEFs were stimulated with IL-1 or LPS (H), or with TNF-α (I). Real-time RT-PCR was performed to measure the abundance of Il6 mRNA (H). Trypan blue staining was performed to measure cell death, and cell lysates were analyzed for the processing of caspase-3 (I). The results shown in (A) to (G) and the lower panel of (I) are representative of three independent experiments. The results shown in (H) and the upper panel of (I) indicate the mean ± SD of triplicate determinations and are representative of two independent experiments.

Polyubiquitination of TAK1 is essential for the formation of the TRAF6-MEKK3-TAK1 complex

The association of TAK1 with TAB2 and TAB3 is required for the activation of TAK1 (10, 16). Therefore, we first assessed the effect of the K209R mutation on the association of TAK1 with TAB2. Immunoprecipitation of TAB2 and subsequent Western blotting analysis with an antibody against TAK1 revealed that wild-type TAK1 and TAK1-K209R were both constitutively bound to TAB2 (fig. S3). We then examined the effect of the K209R mutation on the recruitment of TAK1 by TRAF6. The IL-1–dependent association of TAK1 with TRAF6 was impaired by the K209R mutation (Fig. 3A), indicating that polyubiquitination of TAK1 was essential for IL-1–induced recruitment of TAK1 by TRAF6.

Fig. 3

Polyubiquitination of TAK1 at Lys209 is required for IL-1–induced formation of the TRAF6-MEKK3-TAK1 signaling complex. (A) Lys209 of TAK1 is required for the association of TAK1 with TRAF6. Tak1−/− MEFs transduced with retroviruses expressing TAK1-WT or TAK1-K209R were treated with IL-1 (10 ng/ml). (B) Mekk3−/− MEFs transduced with retroviruses expressing MEKK3-WT or the kinase-defective (KD) mutant of MEKK3 (upper panel) were treated with IL-1 (lower panel). (C and D) IL-1 induces formation of the TRAF6-MEKK3-TAK1 signaling complex. WT-MEFs were treated with IL-1 and the cell lysates were subjected to immunoprecipitation with antibodies against MEKK3 (C) or TAK1 (D). (E) TRAF6 is required for the association between MEKK3 and TAK1. Traf6−/− MEFs and those expressing TRAF6-WT were treated with IL-1. (F) MEKK3 is required for the association between TRAF6 and TAK1. WT-MEFs and Mekk3−/− MEFs were treated with IL-1. (G) Lys209 of TAK1 is required for the association between TRAF6 and MEKK3. Tak1−/− MEFs transduced with retroviruses expressing TAK1-WT or TAK1-K209R were treated with IL-1. (H) Lys209 of TAK1 is required for the association between MEKK3 and TAK1. Tak1−/− MEFs transduced with retroviruses expressing TAK1-WT or TAK1-K209R were treated with IL-1. The results shown are representative of three independent experiments.

We hypothesized that MEKK3 was required for the association between TAK1 and TRAF6 because MEKK3 is essential for IL-1–induced activation of NF-κB and it associates with TRAF6 in response to IL-1 (12). To test this hypothesis, we first assessed whether MEKK3 was essential for IL-1–induced activation of TAK1. Whereas TAK1 was activated in response to IL-1 in the presence of MEKK3, the activation of TAK1 was severely reduced in Mekk3−/− MEFs and those transduced with a retrovirus that expressed the kinase-defective (KD) form of MEKK3 (13) (Fig. 3B). This strongly suggested that MEKK3 was an upstream activator of TAK1, which led us to test whether MEKK3 formed a complex with TAK1 and TRAF6 in a stimulation-dependent manner. In MEFs, both TRAF6 and TAK1 associated with MEKK3 within 5 min of stimulation with IL-1 and were substantially dissociated from MEKK3 30 min after stimulation (Fig. 3C). When TAK1 rather than MEKK3 was immunoprecipitated, we observed a similar transient association of TAK1 with TRAF6 and MEKK3 (Fig. 3D). Although the kinetics of the activation of TAK1 and that of the formation of the complex among TRAF6, MEKK3, and TAK1 during the first 30 min of IL-1 stimulation varied subtly from experiment to experiment, the kinetics of both events always matched within any given experiment (Fig. 3, C and D).

To further characterize how this complex was formed, we assessed whether any one of the three proteins was required for the IL-1–induced association of the other two proteins. TAK1 did not associate with MEKK3 in the absence of TRAF6 (Fig. 3E), and TRAF6 did not associate with TAK1 in the absence of MEKK3 (Fig. 3F). Finally, TRAF6 did not bind to MEKK3 in Tak1−/− MEFs (Fig. 3G). These results indicated that the signaling complex simultaneously including TAK1, MEKK3, and TRAF6 was formed in an IL-1–dependent manner. The K209R mutation abrogated the association between TRAF6 and MEKK3, as well as that between TAK1 and MEKK3 (Fig. 3, G and H). These three proteins may loosely associate with one another even before stimulation; the overabundance of any pair of the three proteins resulted in their association (Fig. 6B and fig. S4), and TAK1 has been found in MEKK3-containing complexes isolated from unstimulated cells by tandem affinity purification (17). However, the physiological formation of the TRAF6-MEKK3-TAK1 complex requires stimulation-dependent conjugation of a Lys63-linked polyubiquitin chain to Lys209 of TAK1, suggesting that the polyubiquitination of TAK1 is involved in the stabilization of the complex. Moreover, phosphorylation of IKKα, IKKβ, and TAK1 and formation of the complex were observed over a similar time course (Fig. 3, C and D), suggesting that formation of the complex was a prerequisite for the activation of TAK1 and thereby for TAK1-induced activation of IKK.

To rule out the possibility that the K209R mutation abrogated the intrinsic kinase activity of TAK1, we took advantage of the ability of TAB1 to activate TAK1 in transiently transfected cells. TAB1 is dispensable for the activation of TAK1 in response to physiological stimuli, including IL-1 and TNF-α (18), whereas TAB1 can activate TAK1 when both are overexpressed in transfected cells. Moreover, MEKK3 and ubiquitination of TAK1, both of which are essential for TAK1-mediated activation of NF-κB induced by IL-1 or TNF-α, were dispensable for TAB1-induced activation of TAK1 (fig. S5), indicating that TAB1-induced activation of TAK1 is a suitable method for checking the intrinsic kinase activity of TAK1 mutants. On the basis of the autophosphorylation state of TAK1 and the band shifts observed for TAK1 and TAB1, we inferred that the kinase activity of TAK1-K209R was similar to that of wild-type TAK1 (fig. S6), indicating that the K209R mutation of TAK1 did not alter its kinase activity.

Ubc13 is required for the IL-1–induced polyubiquitination of TAK1

To elucidate the molecular mechanism responsible for the polyubiquitination of TAK1, we searched for the responsible E2 ubiquitin–conjugating and E3 ubiquitin ligase enzymes. TRAF6 is an E3 ubiquitin ligase that functions together with the E2 enzyme Ubc13-Uev1A to catalyze the synthesis of Lys63-linked polyubiquitin chains (8). We observed that the polyubiquitination of TAK1 was impaired in the absence of TRAF6 (Fig. 1C), suggesting that TRAF6 acted as the E3 enzyme responsible for ubiquitination of TAK1. We then investigated the polyubiquitination of TAK1 in the absence of Ubc13. The Cre gene was introduced into Ubc13fl/fl MEFs, which were derived from mice homozygous for loxP-flanked Ubc13 alleles (19). In the absence of Ubc13, ubiquitination of TAK1 was remarkably impaired (Fig. 4A), but the ubiquitination of TRAF6 was hardly affected (Fig. 4B), which strongly suggested that another E2 protein was involved in its ubiquitination (supplementary text and fig. S7). Furthermore, Ubc13 was required for activation of TAK1, degradation of IκBα, and activation of JNK and p38 (Fig. 4C). Ubc13 was also essential for the stimulation-dependent association of TAK1 with MEKK3 (Fig. 4D). This was consistent with the requirement for Lys209 in the formation of the TRAF6-MEKK3-TAK1 complex. Together, these results showed that TRAF6 and Ubc13 were likely to function as E3 and E2 enzymes, respectively, during stimulation-induced Lys63-linked polyubiquitination of TAK1. In contrast to our current observation, it has previously been reported that Cre-dependent deficiency in Ubc13 does not affect the degradation of IκBα or the amount of phosphorylated TAK1, although it does delay the kinetics of phosphorylation of TAK1 (19). The discrepancy between this finding and our data could be due to the different conditions used to select Cre-expressing MEFs (Supplementary text and fig. S8).

Fig. 4

Ubc13 is required for the IL-1–induced polyubiquitination of TAK1. (A) WT-MEFs and Ubc13fl/fl MEFs expressing Cre were treated with IL-1. Cell lysates were boiled and then immunoprecipitation of TAK1 was performed. (B) Ubc13 is not essential for IL-1–induced ubiquitination of TRAF6. WT-MEFs and Ubc13fl/fl MEFs expressing Cre and HA-K63Ub were treated with IL-1. Cell lysates were boiled and then immunoprecipitation of TRAF6 was performed. Ubiquitinated TRAF6 was detected with an antibody against HA. (C) Ubc13 is required for IL-1–induced phosphorylation of TAK1. Cells were treated as described in (A). Cell lysates were then analyzed by Western blotting. (D) Ubc13 is required for the IL-1–induced association between TAK1 and MEKK3. Cells were treated as described in (A), followed by analysis of the extent of the association between TAK1 and MEKK3. The results shown are representative of three independent experiments.

Cooperation between the RING and Zinc pathways results in the sustained activation of NF-κB in response to IL-1

In addition to the NF-κB activation pathway that depends on the RING domain of TRAF6 (the RING pathway), we have previously identified a RING domain–independent, zinc finger–dependent NF-κB activation pathway (hereafter, the Zinc pathway) on the basis of the analysis of Traf6−/− MEFs that expressed TRAF6-WT, T6ΔR, or T6ΔRZ (20) (fig. S9). Interestingly, the Zinc pathway depended entirely on MEKK3 but did not require TAK1 (fig. S10). These results are consistent with our other observations that the lack of either TAK1 or Ubc13 resulted in a substantial, but partial, reduction in the phosphorylation of IκBα and the activation of NF-κB (Figs. 2, A, B, and F, and 4C) whereas deficiency in MEKK3 resulted in a severe reduction in IL-1–induced activation of NF-κB (fig. S11), as previously reported (12). To further characterize both pathways, we introduced amino acid substitutions into each zinc finger domain to search for a TRAF6 mutant that dominantly activated the RING pathway without substantially activating the Zinc pathway (fig. S12A). The T6mZ5 mutant, in which the fifth zinc finger was mutated, was identified as such a mutant because stimulation of T6mZ5-MEFs with IL-1 resulted in the phosphorylation of IκBα in a TAK1-dependent manner (fig. S12, B to D) and the normal activation of MAPKs, an activity unique to the RING pathway (fig. S12E).

We then compared the degree of IL-1–induced activation of NF-κB in TRAF6-WT-MEFs, T6ΔR-MEFs, and T6mZ5-MEFs by electrophoretic mobility shift assays (EMSAs). Nuclear NF-κB appeared 15 min after stimulation with IL-1 in both TRAF6-WT-MEFs and T6mZ5-MEFs (Fig. 5A). However, although the abundance of nuclear NF-κB after 30 min of stimulation was increased compared with that at 15 min in TRAF6-WT-MEFs, in T6mZ5-MEFs it was decreased. Sixty minutes after stimulation, a slightly reduced but substantial amount of nuclear NF-κB was still detected in TRAF6-WT-MEFs but not in T6mZ5-MEFs. In T6ΔR-MEFs, nuclear NF-κB was undetectable 15 min after stimulation with IL-1, but its abundance increased thereafter. These results strongly suggest that the RING pathway is mainly involved in the early phase (0 to 30 min after stimulation), whereas the Zinc pathway is mainly involved in the late phase (30 to 60 min after stimulation) of activation of NF-κB.

Fig. 5

Cooperation of the RING and Zinc pathways is required for the efficient expression of NF-κB target genes in response to IL-1. (A) Cooperation of the RING and Zinc pathways in the persistent activation of NF-κB. The indicated cells were treated with IL-1 and EMSA analyses were performed. (B) Time-dependent phosphorylation of IκBα by the RING and Zinc pathways in the presence of a proteasome inhibitor. The indicated cells were cultured with MG132 (30 μM) for 15 min and then treated with IL-1. (C) MEKK3, but not TAK1, is required for the Zinc pathway. After siRNA-mediated knockdown (fig. S10A) of MEKK3 (upper) or TAK1 (lower), the indicated cells were analyzed as described in (B). Cont, control. (D) The activity of TRAF6 is lost by inactivating both the RING and the Zinc pathways. The indicated cells were analyzed as described in (B) (upper). After siRNA-mediated knockdown of MEKK3 or TAK1 (lower right), Traf6−/− MEFs expressing T6D57K were analyzed as described in (B) (lower left). (E) Dependency of the expression of NF-κB target genes on the Zinc pathway. Traf6−/− MEFs transduced with retroviruses expressing TRAF6-WT or T6mZ5 were treated with IL-1. The expression of NF-κB target genes was assessed by real-time RT-PCR analysis. (F) The Zinc pathway is required for the production of IL-6. Traf6−/− MEFs expressing TRAF6-WT or the indicated TRAF6 mutants were treated with IL-1. The amount of IL-6 in the medium was measured by ELISA (enzyme-linked immunosorbent assay). The results shown in (A) to (D) are representative of three independent experiments. The results shown in (E) and (F) indicate the mean ± SD of triplicate determinations and are representative of two independent experiments.

To gain further evidence of the time-dependent roles of the two pathways, the degree of phosphorylation of IκBα was analyzed in the presence of the proteasome inhibitor MG132 to largely block the degradation of IκBα, thereby preventing loss of its phosphorylated form (Fig. 5B). In TRAF6-WT-MEFs, the phosphorylated form of IκBα appeared 5 min after stimulation with IL-1 and thereafter accumulated. In T6ΔR-MEFs, a band corresponding to phosphorylated IκBα was faintly detectable 5 min after stimulation and it became more apparent 15 min after stimulation and thereafter. The phosphorylated form of IκBα appeared 5 min after stimulation with IL-1 in T6mZ5-MEFs, as was observed in TRAF6-WT-MEFs, whereas it subsided thereafter. The MEKK3-dependence and TAK1-independence of the Zinc pathway were confirmed in the presence of MG132 (Fig. 5C).

The dependence of the early phase of the activation of NF-κB on TAK1 was also confirmed in Tak1−/− MEFs (fig. S13). In addition to T6ΔR, the full-length, ligase-defective mutant T6D57K (21), in which the Ubc13-binding residue Asp57 was mutated to a lysine, induced the late phase of activation of NF-κB, which was blocked by small interfering RNA (siRNA)–mediated silencing of MEKK3, but not of TAK1 (Fig. 5D, supplementary text, and fig. S14). Furthermore, T6ΔR-mZ5 and T6D57K-mZ5, which carry mutations to both the RING and the Zinc domains, barely induced the phosphorylation of IκBα (Fig. 5D).

Because inactivation of the Zinc pathway resulted in a faster decay with a shorter oscillation period in the DNA binding activity of NF-κB during the first 24 hours after stimulation (fig. S15), we analyzed the expression of NF-κB target genes by real-time, reverse transcription–polymerase chain reaction (RT-PCR) assays (Fig. 5E). Expression of Tnfα, Ccl2, and Cxcl10 was strongly dependent on the Zinc pathway. Although not as pronounced, the expression of Il6 and Irf1 still showed substantial dependence on the Zinc pathway, whereas the expression of Icam1 was unaffected. The absence of late-phase activation resulted in a 30% reduction in the production of IL-6 24 hours after stimulation with IL-1 (Fig. 5F), which may reflect the reduction in the amount of Il6 messenger RNA (mRNA) that had accumulated by this point because of the faster decay in the activation of NF-κB in T6mZ5-MEFs. p38 MAPK was required for the expression of Il6, Icam1, and Irf1 in response to TNF-α, but not for the expression of Tnfα and Ccl2 (22). Therefore, the varied extents of Zinc pathway dependence exhibited by various NF-κB target genes may reflect the extent of the dependence of the expression of each gene on MAPK. Together, these results imply the physiological importance of cooperation between the RING and the Zinc pathways in the IL-1–induced production of cytokines.

TRAF6-mediated oligomerization of MEKK3 induces its activation through phosphorylation of Ser526

We next searched for a domain within TRAF6 that was required for its activation of MEKK3. We took advantage of transient transfection experiments that allowed us to characterize the TRAF6-MEKK3 interaction in the absence of exogenous TAK1. Transient transfection of HEK 293T cells with plasmids encoding HA-MEKK3 and FLAG-TRAF6 or its mutants (Fig. 6A) and subsequent coimmunoprecipitations revealed that full-length TRAF6 and T6ΔR efficiently bound to MEKK3, whereas T6ΔRZ, T6ΔCT, and T6ΔT did not (Fig. 6B). This indicated that the RING domain was dispensable for the binding of TRAF6 to MEKK3, whereas the zinc finger and TRAF-C domains were required.

Fig. 6

Possible involvement of the TRAF6-mediated oligomerization of MEKK3 in the activation of MEKK3. (A) Schematic representation of various deletion mutants of TRAF6. (B) The RING and Zinc-finger domains of TRAF6 are involved in the activation of MEKK3. HEK 293T cells were cotransfected with plasmids encoding FLAG-TRAF6 or one of the FLAG-tagged TRAF6 mutants described in (A) with a plasmid encoding HA-MEKK3. The cell pool was then divided into two portions. Whole-cell lysate (WCL) was prepared from one portion and then analyzed by Western blotting with antibodies against HA and pMEKK3 to assess the extent of activation of MEKK3. The other portion of cells was subjected to immunoprecipitation with an antibody against FLAG, followed by Western blotting with antibodies against HA and FLAG to analyze the binding of MEKK3 to TRAF6 or its mutants. (C) TRAF6 induces oligomerization of MEKK3. HEK 293T cells were cotransfected with plasmids encoding FLAG-TRAF6 or one of the FLAG-tagged TRAF6 mutants described in (A) together with HA-MEKK3 and FLAG-MEKK3. The cell pool was then divided into two portions. WCL was prepared from one portion and then analyzed by Western blotting with an antibody against FLAG. The other portion of cells was subjected to immunoprecipitation with an antibody against HA, followed by Western blotting analysis with antibodies against FLAG and HA to analyze the binding of HA-MEKK3 to FLAG-MEKK3. The results shown are representative of three independent experiments.

To identify the TRAF6 domain required for the activation of MEKK3, cell lysates were analyzed for the abundance of MEKK3 phosphorylated at Ser526, a critical phosphorylation site for its activation (13). Phosphorylation of MEKK3 was facilitated by TRAF6 and T6ΔR, albeit to a lesser extent than TRAF6, but not by T6ΔRZ, T6ΔCT, or T6ΔT (Fig. 6B). These results indicated that the RING and zinc finger domains of TRAF6 were involved in the activation of MEKK3. Oligomerization of MEKK3 results in its activation (23), probably as a consequence of autophosphorylation, and TRAF6 forms homotrimers to transduce signals (24). Therefore, we tested whether TRAF6 was involved in the oligomerization of MEKK3. HEK 293T cells were cotransfected with plasmids encoding FLAG-MEKK3 and HA-MEKK3, and their oligomerization was analyzed in the absence or presence of TRAF6 or its mutants. Oligomerization of MEKK3 was efficiently augmented by TRAF6 or T6ΔR, whereas T6ΔRZ and T6ΔT had no effect (Fig. 6C). Together, these results strongly suggest that TRAF6-induced activation of MEKK3 could involve oligomerization of MEKK3. Although it remains unclear why TRAF6 activated MEKK3 to a greater extent than did T6ΔR (Fig. 6B), it should be noted that overexpression of TRAF6 and stimulation with IL-1 resulted in Lys63-linked polyubiquitination of MEKK3, a process that depended on the RING domain of TRAF6 (fig. S16, A and B) and Ubc13 (fig. S16C). Therefore, RING domain–induced Lys63-linked polyubiquitination of MEKK3 may be involved in its activation.

Discussion

Here, we showed that IL-1 induced the transient formation of a signaling complex that contained TRAF6, TAK1, and MEKK3 and that formation of the complex was essential for the activation of TAK1, which led to TAK1-dependent activation of NF-κB and MAPKs. We also showed that the catalytic activity of MEKK3 was essential for the activation of TAK1, which strongly suggested that MEKK3 was an upstream activator of TAK1. Therefore, MEKK3 and TAK1 physically and functionally interact with each other to transduce the IL-1 signal.

Another finding showed herein is that site-specific Lys63-linked polyubiquitination of TAK1 at Lys209, presumably catalyzed by TRAF6 and Ubc13, was essential for the formation of the TRAF6-MEKK3-TAK1 signaling complex. Lys124 of TRAF6 becomes conjugated to the Lys63-linked polyubiquitin chain in response to IL-1, and TAB2 associates with the polyubiquitin chain of TRAF6 (10, 25). Moreover, Lys134 and Lys180 of IRAK1 are conjugated to Lys63-linked polyubiquitin chains in response to IL-1, and NEMO associates with the polyubiquitin chain of IRAK1, but not with that of TRAF6 (9). Because TRAF6 associates with IRAK1 upon stimulation (26), one may speculate that the Lys63-linked polyubiquitin chains of TRAF6 and IRAK1 mediate stimulation-dependent recruitment of the TAK1-TAB2-TAB3 and IKK complexes to TRAF6, respectively (Fig. 7). How then is the Lys63-linked polyubiquitination of TAK1 described herein involved in IL-1 signaling? Given that ubiquitination of TAK1 was required for the formation of the TRAF6-MEKK3-TAK1 signaling complex, one possibility is that the polyubiquitin chains of TAK1 may mediate the interaction between TAK1 and MEKK3 through the preferential binding of MEKK3, or of its associated proteins, to the polyubiquitin chain of TAK1 (Fig. 7).

Fig. 7

A model illustrating the IL-1–induced activation of NF-κB. The RING pathway transduces signals by forming a signaling complex consisting of TRAF6, MEKK3, and TAK1. The binding of TAB2 and TAB3 to the Lys63-linked polyubiquitin chain conjugated to Lys124 of TRAF6 (9, 10) and the binding of MEKK3 or its associated protein (designated as X) to the Lys63-linked polyubiquitin chain conjugated to Lys209 of TAK1 may contribute to the stabilization of the signaling complex. In addition, binding of NEMO to the Lys63-linked polyubiquitin chain conjugated to Lys134 and Lys180 of IRAK1 may result in the recruitment of the IKK complex to the signaling complex (9). These polyubiquitin reactions are thought to be catalyzed by TRAF6. Such polyubiquitin chain–mediated protein complex formation may trigger the MEKK3-TAK1-IKK kinase cascade, thereby leading to the activation of NF-κB. Lys63-linked polyubiquitin chains conjugated to Lys392 of NEMO are involved in TAK1-induced activation of MAPKs (19, 44). On the other hand, the Zinc pathway is independent of TAK1 but dependent on MEKK3, and the pathway is preceded by the RING pathway during the first hour after stimulation with IL-1. Cooperation of the two pathways is required for the production of sufficient amounts of inflammatory cytokines. Note that the requirement for the polyubiquitination of TRAF6 at Lys124 for the activation of NF-κB is controversial (31). Linear polyubiquitin chains conjugated to Lys285 and Lys309 of NEMO are involved in the activation of IKK complex by TNF-α and IL-1 (45). Further analyses are needed to understand the critical relationships between Lys63-linked and linear polyubiquitinations.

Although MEKK3 does not have any known ubiquitin-binding domains, our preliminary data showed that p62, which binds to the Lys63-linked polyubiquitin chain and is involved in IL-1 signaling (27), associates with MEKK3 in a PB1 domain–dependent manner (fig. S17). Given that the PB1 domain is thought to play a critical role in protein-protein interactions (28), this result suggests that p62 is likely to be involved in the stimulation-dependent recruitment of TAK1 to MEKK3 by associating with both MEKK3 and the polyubiquitin chain of TAK1. An alternative possibility could be that the polyubiquitination of TAK1 results in a conformational change in TAK1 that causes Thr178, Thr184, and Thr187, critical phosphorylation sites in the activation loop of TAK1 (29, 30), to be available to some unidentified kinase that directly phosphorylates and activates TAK1, because Lys209 is juxtaposed to the activation loop. Our data suggest that MEKK3 may directly phosphorylate and activate TAK1. It has been reported that polyubiquitination of TRAF6 is dispensable for the activation of TAK1 (31), which suggests that stimulation-dependent polyubiquitination of TAK1 may play a critical role in the recruitment of the TAK1-TAB2-TAB3 complex to the TRAF6-MEKK3 complex.

In addition to the RING pathway, we also characterized the Zinc pathway, which was mediated by MEKK3, but not by TAK1 (Fig. 7). The Zinc pathway was activated by the ligase-deficient mutant of TRAF6, T6ΔR, but not by T6mZ5. This suggested that MEKK3 could be activated by the zinc finger of TRAF6, consistent with the finding that T6ΔR induced the oligomerization of MEKK3, a possible mechanism for the activation of MEKK3. Although MEKK3 was activated by IL-1 in the Zinc pathway in Tak1−/− MEFs, stimulation-dependent association of endogenous TRAF6 with MEKK3 was not detected in the absence of TAK1. The weak or short-lasting interaction between TRAF6 and MEKK3, which is undetectable by immunoprecipitation, may be sufficient to activate MEKK3.

The RING and Zinc pathways diverged at TRAF6 and activation of NF-κB by the RING pathway preceded that by the Zinc pathway during the first hour after stimulation with IL-1. Although the molecular mechanisms of a temporal switch in the signaling proteins used downstream of TRAF6 remain to be elucidated, one may speculate that deubiquitination of the Lys63-linked polyubiquitin chains might be involved, because the RING pathway, but not the Zinc pathway, requires Ubc13 and the catalytic activity of TRAF6. Cooperation of these two mechanistically and temporally distinct pathways resulted in fully sustained activation of NF-κB. This cooperation is physiologically relevant because lack of the Zinc pathway resulted in a substantial reduction in the production of various cytokines and chemokines, a reduction that would likely affect inflammatory responses in vivo.

Two parallel TAK1-dependent and MEKK3-dependent signaling pathways have been previously proposed for the IL-1–induced activation of NF-κB (32). Such a MEKK3-dependent pathway may be similar to the Zinc pathway described here. However, the TAK1-dependent pathway is distinct from our RING pathway because the former is MEKK3-independent but the latter requires MEKK3. Further studies are required to explain this discrepancy.

TAK1 activation is required for multiple signaling pathways, including the TLR (Toll-like receptor), IL-1R, TNFR, T cell receptor, nucleotide-binding oligomerization domain 2 (NOD2), TGF-β receptor, and osmotic stress signaling pathways (3337). Here, we have shown that polyubiquitination of TAK1 at Lys209 was required for the activation of TAK1 in response to IL-1, TNF-α, and LPS. However, polyubiquitination of TAK1 at Lys34, an essential event for TGF-β–induced activation of TAK1 (14), was not required for IL-1– and TNF-α–induced activation. These results suggest that the specific ubiquitination site of TAK1 differs in response to distinct stimuli. Such differences may be required to form various signal complexes that consist of distinct components, which would lead to the generation of signals with distinct qualities. IL-1 is a potent mediator of tumor development (38, 39) in addition to being involved in inflammation (1). Therefore, further studies are required to dissect the signaling mechanisms responsible for this phenomenon to potentially elucidate therapeutic targets for the various diseases promoted by IL-1.

Materials and Methods

Antibodies and plasmids

The following antibodies were used: antibodies against HA (sc-3792), TAK1 (M-579), TAB2 (K-20), IκBα (sc-371), TRAF6 (sc-7221, sc-33897, and sc-8409), ubiquitin (sc-8017), and p38 (sc-535) (Santa Cruz Biotechnology); FLAG (Sigma); pIκBα (9246), pIKKα/β (2681), pp38 (9211), pJNK (9251) pERK (9101), cleaved caspase-3 (9661) (Cell Signaling Technology); MEKK3 (611103) (BD Transduction Laboratories); tubulin (cp06) (Calbiochem); Ubc13 (Zymed Laboratories); UbcH7 (Chemicon); pMEKK2/3 (13), MEKK3 (12), and pTAK1 (29). pEF-HA-TAB1 and pEF-HA-TAB2 were provided by K. Matsumoto (Nagoya University, Japan). pcDNA3-HA-Ub and its variants were provided by K. Tanaka (Tokyo Metropolitan Institute of Medical Science, Japan). HA-MEKK3 (12), pMRX-Cre (19), pME-FLAG-CYLD (40), pME-FLAG-TRAF6, pME-FLAG-T6ΔR, and pME-FLAG-T6ΔRZ (20) were used as described. Complementary DNAs (cDNAs) encoding wild-type and various KR mutants of TAK1, wild-type and various deletion and point mutants of TRAF6, and wild-type and the kinase-deficient (KD) mutant of MEKK3 were generated by PCR with the appropriate primers. These cDNAs were inserted into the retrovirus vector pMXs or the expression plasmid pME with or without the FLAG tag sequence. The pMXs vector was provided by T. Kitamura (University of Tokyo, Japan).

Cell culture and transfections

HEK 293T cells, WT-MEFs, Traf6−/− MEFs (20), Tak1−/− MEFs (33), Mekk3−/− MEFs (41), and Ubc13fl/fl MEFs (19) were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS). Transfection of HEK 293T cells was performed by the calcium phosphate method. MEFs were transfected with siRNAs at a final concentration of 10 nM with RNAi (RNA interference) MAX reagents (Invitrogen) and with expression plasmids by Lipofectamine 2000 (Invitrogen). The target sequences were as follows: TAK1, UUUAGACCAACAACGAGUCAUCAGG; MEKK3, UAAAGAUGGUGAGGAUCUUCUCAGC.

Retrovirus-mediated gene transfer

The packaging Plat-E cell lines (4 × 106 cells) were transfected with 16 μg of the various retrovirus vectors. Virus stocks were prepared by collecting the culture medium 48 hours after transfection. MEFs (1 × 106 cells) were cultured overnight and then mock-infected or incubated with a virus stock medium containing polybrene (5 μg/ml; Sigma-Aldrich) for 8 hours. One day after the infection, puromycin (3 μg/ml) was added to the medium. Puromycin-resistant cell pools were used for further experiments.

Immunoprecipitation and Western blotting

For preparation of whole-cell lysates, cells were washed once with phosphate-buffered saline (PBS) and then lysed in sample buffer [67.5 mM tris-HCl (pH 7.2), 2.25% SDS, 10% glycerol, 5% β-mercaptoethanol]. For immunoprecipitations, cells were first lysed in TNE buffer [50 mM tris-HCl (pH 7.5), 250 mM NaCl, 1% NP-40, 1 mM EDTA, 1 mM Na3VO4, 1 mM NaF, 0.4 μM phenylmethylsulfonyl fluoride (PMSF), 1 mM dithiothreitol (DTT)] and then centrifuged at 17,000g for 15 min at 4°C to remove the nuclei. The resulting supernatant was mixed with 0.5 μg of antibody together with 10 μl of protein G–Sepharose (GE Healthcare). Immunoprecipitates were washed thrice with TNE buffer and then resuspended in sample buffer. For Western blotting analysis, either the immunoprecipitates or the whole-cell lysates were separated by SDS polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a polyvinylidenedifluoride (PVDF) membrane (Millipore). The membranes were incubated with the appropriate primary antibodies. Immunoreactive proteins were visualized with antibodies against rabbit or mouse immunoglobulin G conjugated to horseradish peroxidase (Amersham Biosciences), followed by processing with an ECL detection system (Amersham Biosciences). For detection of the ubiquitination of TAK1, TRAF6, and MEKK3, cell lysates prepared in TNE buffer were boiled for 10 min in the presence of 1% SDS to remove noncovalently attached proteins. The lysates were then diluted 10-fold in TNE buffer to reduce the SDS concentration to 0.1%. Immunoprecipitations were performed with antibodies against TAK1, TRAF6, or MEKK3 in the presence of protease inhibitors. Ubiquitin was then detected by Western blotting analysis.

Electrophoretic mobility shift assay

MEFs were stimulated with IL-1 (10 ng/ml) for the indicated times and then suspended in hypotonic buffer [10 mM Hepes (pH 7.9), 1.5 mM MgCl2, 10 mM KCl, 0.5 mM DTT, 0.4 μM PMSF]. The cell suspension was maintained on ice for 20 min, and the cells were then disrupted by pipetting. The supernatant was removed, and the pelleted nuclei were incubated with extraction buffer [20 mM Hepes (pH 7.9), 1.5 mM MgCl2, 420 mM NaCl, 0.2 mM EDTA, 0.5 mM DTT, 0.4 μM PMSF, 25% glycerol]. The suspension was incubated on ice for 20 min, and the nuclear extract was obtained from the supernatant. Equal amounts of extracts were incubated for 25 min at room temperature with 32P-labeled oligonucleotide containing the NF-κB binding site of the Igκ light-chain gene (5′-AGCTTCAGAGGGGACTTTCCGAGAGG-3′, 5′-TCGACCTCTCGGAAAGTCCCCTCTGA-3′) and poly(dI-dC) [poly(deoxyinosine-deoxycytidine), 0.05 μg/ml]. Binding reactions were carried out in the following buffer: 15 mM tris-HCl (pH 7.5), 75 mM NaCl, 1.5 mM EDTA, 1.5 mM DTT, 7.5% glycerol, 0.3% NP-40, and bovine serum albumin (BSA, 1 μg/ml). Electrophoresis was performed on 4% acrylamide gels at 150 V for 90 min. The gel was then dried and exposed to film (Kodak).

Real-time RT-PCR assays

Total RNA was isolated from cells with the Trizol Reagent (Invitrogen) and cDNA synthesis was performed with PrimeScript II (Takara Bio). Real-time RT-PCR analysis was performed with the 7300 system (Applied Biosystems) and SYBR Green (Roche). The extent of the expression of β-actin cDNA in each sample was used to standardize the data. The primers used for β-actin, Il6, Cxcl10, Tnfα, Icam1, Ccl2, and Irf-1 have been previously described (22, 42, 43).

Acknowledgments

We thank K. Shimizu, M. Hashimoto, J. Kuritani, and K. Semba for secretarial assistance. We are grateful to K. Matsumoto, K. Tanaka, and T. Kitamura for various expression plasmids. This work was supported by Grants-in-Aid for Scientific Research on Priority Areas (to J.I.), Young Scientists B (to J.G.), and a Medical Genome Science Program Fellowship (to K.Y.) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/2/93/ra66/DC1

Supplementary Text

Fig. S1. Amino acid sequence alignment of TAK1 from various species.

Fig. S2. IL-1 induces Lys63-linked polyubiquitination of TAK1 KR mutants.

Fig. S3. The K209R mutation of TAK1 does not affect the interaction between TAK1 and TAB2.

Fig. S4. Analysis of the association between TAK1, TRAF6, and MEKK3 in transiently transfected cells.

Fig. S5. The activation of TAK1 by TAB1 does not require MEKK3 or the Lys63-linked polyubiquitination of TAK1.

Fig. S6. The K209R mutation of TAK1 does not abolish the intrinsic kinase activity of TAK1.

Fig. S7. Possible involvement of UbcH7 in TRAF6-induced activation of NF-κB.

Fig. S8. Effect of a partial reduction in the abundance of Ubc13 on IL-1 signaling.

Fig. S9. RING domain–independent and zinc finger–dependent activation of NF-κB by IL-1.

Fig. S10. MEKK3, but not TAK1, is required for the Zinc pathway.

Fig. S11. Severe reduction in the IL-1–induced activation of NF-κB in the absence of MEKK3.

Fig. S12. The T6mZ5 mutant, in which the fifth zinc finger is inactivated, activates the RING pathway but not the Zinc pathway.

Fig. S13. Time-dependent activation of NF-κB in Tak1−/− MEFs.

Fig. S14. Time-dependent phosphorylation of IκBα induced by RING domain–defective mutants of TRAF6.

Fig. S15. Cooperation of the RING and Zinc pathways results in the prolonged activation of NF-κB in the nucleus.

Fig. S16. IL-1–induced polyubiquitination of MEKK3 depends on the RING domain of TRAF6 and Ubc13.

Fig. S17. MEKK3 associates with p62 in a PB1 domain–dependent manner.

References

References and Notes

View Abstract

Navigate This Article