Interleukin-1 (IL-1)-induced TAK1-dependent Versus MEKK3-dependent NFκB Activation Pathways Bifurcate at IL-1 Receptor-associated Kinase Modification*

Interleukin-1 (IL-1) receptor-associated kinase (IRAK) is phosphorylated after it is recruited to the receptor, subsequently ubiquitinated, and eventually degraded upon IL-1 stimulation. Although a point mutation changing lysine 134 to arginine (K134R) in IRAK abolished IL-1-induced IRAK ubiquitination and degradation, mutations of serines and threonines adjacent to lysine 134 to alanines ((S/T)A (131–144)) reduced IL-1-induced IRAK phosphorylation and abolished IRAK ubiquitination. Through the study of these IRAK modification mutants, we uncovered two parallel IL-1-mediated signaling pathways for NFκB activation, TAK1-dependent and MEKK3-dependent, respectively. These two pathways bifurcate at the level of IRAK modification. The TAK1-dependent pathway leads to IKKα/β phosphorylation and IKKβ activation, resulting in classical NFκB activation through IκBα phosphorylation and degradation. The TAK1-independent MEKK3-dependent pathway involves IKKγ phosphorylation and IKKα activation, resulting in NFκB activation through IκBα phosphorylation and subsequent dissociation from NFκB but without IκBα degradation. These results provide significant insight to our further understanding of NFκB activation pathways.


Interleukin-1 (IL-1) receptor-associated kinase (IRAK) is phosphorylated after it is recruited to the receptor, subsequently ubiquitinated, and eventually degraded upon IL-1 stimulation. Although a point mutation changing lysine 134 to arginine (K134R) in IRAK abolished IL-1-induced IRAK ubiquitination and degradation, mutations of serines and threonines adjacent to lysine 134 to alanines ((S/T)A (131-144)) reduced IL-1-induced IRAK phosphorylation and abolished IRAK ubiquitination.
Through the study of these IRAK modification mutants, we uncovered two parallel IL-1-mediated signaling pathways for NFB activation, TAK1-dependent and MEKK3-dependent, respectively. These two pathways bifurcate at the level of IRAK modification. The TAK1-dependent pathway leads to IKK␣/␤ phosphorylation and IKK␤ activation, resulting in classical NFB activation through IB␣ phosphorylation and degradation. The TAK1-independent MEKK3-dependent pathway involves IKK␥ phosphorylation and IKK␣ activation, resulting in NFB activation through IB␣ phosphorylation and subsequent dissociation from NFB but without IB␣ degradation. These results provide significant insight to our further understanding of NFB activation pathways.
Chen and co-workers (23,24) showed that protein ubiquitination plays an important role in TRAF6-mediated TAK1 and IKK activation. TRAF6, a Ring domain protein, has been shown recently to function as a ubiquitin protein ligase E3, and TRAF6 itself is the target of ubiquitination, which has been shown to activate TAK1 through an unknown mechanism. Once activated, TAK1 leads to the phosphorylation of IKK␤ and MKK6, resulting in the activation of both the JNK and NFB signaling pathways. In addition to TAK1, MEKK2 and MEKK3 have also been implicated in the activation of IKK and MAPK, leading to the activation of NFB and JNK (25)(26)(27)(28). The detailed signaling mechanism is not clear. Recently genetic studies have provided further evidence for an essential role of TAK1 in IL-1 signaling. Two groups (29,30) independently reported that TAK1 deficiency results in defects in IL-1 signaling. Intriguingly, although IL-1-induced JNK activation was completely abolished, NFB activation was only partially impaired in TAK1-deficient cells, implicating an additional NFB activation mechanism for the IL-1 pathway.
Here, through the study of IRAK modification, we uncovered two parallel IL-1-mediated signaling pathways for NFB activation, TAK1-dependent and MEKK3-dependent pathways, respectively (Fig. 1). These two pathways bifurcate at the level of IRAK modification. The TAK1-dependent pathway leads to IKK␣/␤ phosphorylation and IKK␤ activation, resulting in classical NFB activation through IB␣ phosphorylation and degradation. The TAK1-independent MEKK3-dependent pathway involves IKK␥ phosphorylation and IKK␣ activation, resulting in NFB activation through IB␣ phosphorylation and subsequent dissociation from NFB but without IB␣ degradation. These results provide significant insight to our further understanding of NFB activation pathways.
Recombinant Plasmids and Transfection-IRAK phosphorylation and ubiquitination mutants were generated by sitedirected mutagenesis and cloned into FLAG-tagged pCMV vector (Sigma). Oligonucleotides used for making the phosphorylation mutants were as follows: AGCCGCAGCCGCCGCC-TTCCTCGCCCCAGCTTTTCCAGGCGCCCAGACCCATT and GGGCGCCTGGAAAAGCTGGGGCGAGGAAGGCG-GCGGCTGCGGCTGGCAACTT. Oligonucleotides used for making the ubiquitination mutants were as follows: GCCCCC-GGAGGTTGCCATCCTCAGCCTCC and GGATGGCAAC-CTCCGGGGGCTCCAGGCCTC. Transfection of the indicated plasmids by FuGENE 6 transfection reagents was done as recommended by the manufacturer (Roche Diagnostics).

TAK1-Versus MEKK3-dependent NFB Activation Pathways
buffered saline at a 1:1 ratio). After incubation, the beads were washed three or four times with lysis buffer and resuspended in 20 l of lysis buffer. For immunoblotting, the immunoprecipitates were separated by 10% SDS-PAGE, transferred to Immobilon-P membranes (Millipore), and analyzed by immunoblotting.
Kinase Assays-Cell lysates were immunoprecipitated with anti-IKK␣/␤ (for IB kinase assay) or anti-TAK1 (for TAK1 kinase assay) and collected on protein A-Sepharose beads. Kinase reactions were performed in 50 l of buffer containing 20 mM HEPES (pH 7.0), 20 mM MgCl 2 , 1 mM ATP, 10 mCi of [␥-32 P]ATP at 30°C for 30 min. The substrate for IB kinase assay was 2 g of glutathione S-transferase (GST)-IB, residues (1-54 amino acids) (J. DiDonato, Cleveland Clinic Foundation, Cleveland, OH), whereas the substrate for TAK1 kinase assay was 2 g of His-MKK6. Samples were analyzed by 10% SDS-PAGE, followed by autoradiography. For IRAK kinase assay, cell lysates were immunoprecipitated with anti-IRAK antibody. Purified recombinant IRAK4 (Upstate) was added to the immunoprecipitates, followed by kinase assay as instructed by the manufacturer.
Virus Infections-MEFs were infected with adenovirus (provided by Frank Mercurio at Celgene (San Diego)) expressing either green fluorescent protein (control), IKK␤, or kinase-inactive IKK␤. The titer of each adenovirus stock used for infection was as follows: green fluorescent protein ϭ 3 ϫ 10 8 , IKK␤ ϭ 5 ϫ 10 8 , and kinase-inactive IKK␤ ϭ 1 ϫ 10 9 virus particles/ml. After 18 h, the infected cells were stimulated with IL-1 or were unstimulated. The production of retrovirus and infection has been described previously (35).
IRAK-deficient cells stably transfected with DD ϩ UD ϩ C1, (S/T)A (131-144) and K134R were immunoprecipitated with anti-IRAK, followed by a kinase assay using purified recombinant IRAK4. E, IRAK lysine mutant was generated in kinase inactive full-length IRAK. IRAK-deficient cells stably transfected with kinase inactive full-length (IRAK1 KD mt) and lysine mutant (IRAK1 KD ubmt) were either untreated or treated with IL-1 (10 ng/ml). The cells were then lysed and analyzed by immunoblot with anti-IRAK and anti-actin (loading control). F, cell lysate from IL-1-stimulated IRAK-deficient cells transfected with lysine mutant (IRAK1 KD ubmt) was treated with calf intestinal phosphatase (CIP). G, IRAK phosphorylation mutant was generated in kinase-inactive full-length IRAK. IRAK-deficient cells stably transfected with kinase-inactive full-length (IRAK1 KD mt) and phosphorylation mutant (IRAK1 KD pmt) were either untreated or treated with IL-1 (10 ng/ml). The cells were then lysed and analyzed by immunoblot with anti-IRAK and anti-actin (loading control).

TAK1-Versus MEKK3-dependent NFB Activation Pathways
been used effectively to study the function of IRAK in IL-1-dependent signaling (11,15,37,39). NFB and JNK activation were greatly reduced in IL-1-treated IRAK-deficient cells, but these responses are restored in IRAK-deficient cells transfected with IRAK, indicating that IRAK is required for both. Analysis of IRAK deletion mutants showed that the truncated IRAK pro-tein DD ϩ UD ϩ C1 (containing the DD, UD, and C1 domains) is sufficient to restore IL-1-induced NFB and JNK activation in IRAK-deficient cells. Like the full-length IRAK, DD ϩ UD ϩ C1 is phosphorylated after it is recruited to the receptor, subsequently ubiquitinated, and eventually degraded upon IL-1 stimulation (11,15,37,37,39). Treatment with calf intestinal phosphatase leads to loss of several IL-1-induced shifted IRAK bands, confirming that they are phosphorylated forms (11). IL-1-induced IRAK ubiquitination was confirmed by immunoprecipitation with anti-IRAK antibody followed by Western analysis with anti-ubiquitin antibody (39). To investigate the role of IRAK modification in IL-1 signaling, much effort has been devoted to map IRAK phosphorylation and ubiquitination sites. Because protein ubiquitination occurs on lysine residues, we mutated all three lysines in DD ϩ UD ϩ C1 ( Fig. 2A, lysines at 32, 134, and 179) individually and in combinations. These mutants were then transfected into IRAK-deficient cells and tested for their modification and degradation in response to IL-1 stimulation. Although DD ϩ UD ϩ C1 was phosphorylated, ubiquitinated, and degraded (degradation is complete 6 h after stimulation (data not shown)), a point mutation changing lysine 134 to arginine completely abolished IL-1induced IRAK ubiquitination and degradation (Fig. 2, B and C). Importantly, the K134R mutant was still phosphorylated upon IL-1 stimulation (Fig. 2D). We have shown previously that the kinase-inactive mutant of IRAK is still phosphorylated, ubiquitinated, and degraded upon IL-1 stimulation (11). We mutated lysine 134 to arginine in the kinase-inactive full-length IRAK in order to mimic DD ϩ UD ϩ C1 (which does not contain the kinase domain), eliminating modification initiated by IRAK autophosphorylation. The fulllength ubiquitination mutant (IRAK1 KD ubmt) was modified upon IL-1 stimulation but with greatly reduced degradation as compared with the full-length IRAK (IRAK1 KD mt) (Fig. 2E). Treatment with calf intestinal phosphatase leads to loss of IL-1-

. The impact of mutation in IRAK modification on IL-1-induced NFB and JNK activation.
A, Western analysis. Cell lysates from wild type 293, IRAK-deficient (IRAK-null), or IRAK-deficient cells stably transfected with truncated IRAK constructs (DD ϩ UD ϩ C1, (S/T)A (131-144) and K134R) and full-length kinase-inactive IRAK constructs (IRAK KD, IRAK KDubmt, IRAK KD pmt) were analyzed by Western blot with anti-IRAK and anti-actin antibodies. B, gel shift assay. Wild type 293, IRAK-deficient (IRAK-null), and IRAKdeficient cells stably transfected with truncated IRAK constructs (DD ϩ UD ϩ C1, (S/T)A (131-144), and K134R) and IRAK kinase-inactive full-length constructs (IRAK KD mt, IRAK KD ubmt, and IRAK KD pmt) were either untreated or treated with IL-1 (10 ng/ml) for the indicated times. Cell lysates were analyzed by electrophoretic mobility shift assay with a NFB-specific probe. C-E, Western analyses. 293 wild type cells, IRAK-deficient cells, and IRAK-deficient cells stably transfected with IRAK modification mutants in the truncated IRAK (DD ϩ UD ϩ C1, (S/T)A (131-144) and K134R) or kinase-inactive full length (IRAK KD mt, IRAK KD pmt, and IRAK KD ubmt) were either untreated or treated with IL-1 (10 ng/ml) for the indicated times. Cell lysates were analyzed by Western blots with anti-pIB␣, anti-IB␣, anti-pJNK, anti-JNK, or anti-actin. D, cells were pretreated with cycloheximide (CHX) (20 g/ml) for 2 h before stimulation with IL-1. The degradation of IB␣ was analyzed by Scion Image 1.62C alias and presented as relative percentage of the untreated samples.

TAK1-Versus MEKK3-dependent NFB Activation Pathways
induced shifted IRAK bands of the full-length ubiquitination mutant (IRAK1 KD ubmt), confirming that they are phosphorylated forms (Fig. 2F). These results suggest that the full-length ubiquitination mutant (IRAK1 KD ubmt) is probably still phosphorylated but deficient in ubiquitination in response to IL-1 stimulation, indicating that lysine 134 is an important site for IL-1-induced IRAK ubiquitination and subsequent degradation.
Protein ubiquitination occurs on conserved lysine residues near the phosphoacceptor serines or threonines, and protein phosphorylation precedes ubiquitination (40,41). Previous studies indeed suggested that phosphorylation of IRAK is probably required for its ubiquitination and degradation (42). To identify the phosphorylation sites, we mutated serines and threonines adjacent to K134R to alanines ((S/T)A (131-144), Ser 131 , Ser 137 , Ser 138 , Ser 140 , Thr 141 and Ser 144 ) in the truncated form of IRAK (DD ϩ UD ϩ C1) (Fig. 2, B and D). The mutation of these sites reduced IL-1-induced IRAK phosphorylation (Fig.  2D) and abolished IRAK ubiquitination and degradation (Fig.  2B). We then mutated the same sites in the kinase-inactive fulllength IRAK. The full-length phosphorylation mutant (IRAK1 KD pmt) altered the modification pattern of the full-length IRAK and greatly reduced its ubiquitination and degradation in response to IL-1 stimulation (Fig. 2G).
The Impact of IRAK Modification on IL-1-induced NFB and JNK Activation-To investigate the role of IRAK modification in IL-1 signaling, the above IRAK mutants defective in IRAK modification were examined for their ability to restore IL-1induced NFB and JNK activation in IRAK-deficient cells. For all of the following experiments, pools of IRAK-deficient cells stably transfected with IRAK modification mutants were used and the expression levels of the IRAK mutants were shown in Fig. 3A. Both ubiquitination (K134R) and phosphorylation ((S/ T)A (131-144)) mutants in the truncated form (DD ϩ UD ϩ C1) or the full-length IRAK fully restored IL-1-induced NFB activation in IRAK-deficient cells, as measured by NFB DNA binding activity (Fig.  3B), IB␣ phosphorylation ( Fig.  3C), and NFB-dependent luciferase reporter assay (data not shown). However, IL-1-induced IB␣ degradation was greatly reduced in IRAKdeficient cells transfected with these IRAK modification mutants (Fig.  3C), implicating a specific role of IRAK modification in the NFB activation pathway. Because IB␣ expression is induced upon IL-1 stimulation, we used the protein synthesis inhibitor cycloheximide to inhibit the new synthesis of IB␣ so that IL-1-induced IB␣ degradation can be more easily detected. We confirmed the lack of IL-1-induced IB␣ degradation in the presence of protein synthesis inhibition in IRAK-deficient cells transfected with the IRAK modification mutants (Fig. 3D). Furthermore, these IRAK mutants only partially restored IL-1-induced JNK activation in IRAK-deficient cells, indicating the importance of IRAK modification in IL-1 signaling (Fig. 3E).
IRAK Modification Mutants Failed to Activate TAK1-We then carefully examined the impact of these IRAK mutants on IL-1-induced intermediate signaling leading to NFB activation. Previous studies showed that IL-1-induced TAK1 activation plays a critical role in IL-1 signaling (15,17,29,30). As shown in Fig. 4A, whereas modified IRAK forms a complex with TAK1 upon IL-1 stimulation in wild type 293 cells and IRAKdeficient cells transfected with the truncated IRAK (DD ϩ UD ϩ C1), IL-1-induced interaction between IRAK and TAK1 was abolished in IRAK-deficient cells transfected with IRAK phosphorylation and ubiquitination mutants, indicating that IRAK modification is required for the interaction of IRAK with TAK1. Previous studies showed that IL-1 stimulation leads to TAK1 phosphorylation and TAK1 activation in an IRAK-dependent manner (22). Because the IRAK mutants failed to interact with TAK1 upon IL-1 stimulation, we suspect that these mutants have also lost the ability to activate TAK1. As shown in Fig. 4, B-D, the IRAK modification mutants indeed failed to activate the TAK1 in response to IL-1 stimulation, including TAK1 phosphorylation (Fig. 4B) and TAK1 kinase activity (Fig. 4D). IL-1-induced TAK1 phosphorylation was confirmed by treatment with calf intestinal phosphatase (Fig.  4C). These results suggest that IRAK modification is required for the interaction of IRAK with TAK1, thereby mediating the activation of TAK1.
IRAK Modification Mutants Formed Complex with TRAF6 and MEKK3-Whereas IRAK modification mutants displayed defects in IL-1 signaling, including abolished TAK1 activation,

TAK1-Versus MEKK3-dependent NFB Activation Pathways
reduced JNK activation, and IL-1-induced IB␣ degradation, they are still capable of mediating IL-1-induced IB␣ phosphorylation and NFB activation, implicating that the IRAK mutants mediate an alternative NFB activation pathway. Therefore, we carefully examined the interaction of IRAK mutants with other intermediate signaling components of the IL-1 pathway. TRAF6 is an immediate downstream component of IRAK in IL-1 signaling pathway (14,15,39). Although the IRAK mutants failed to interact with TAK1 upon IL-1 stimulation, they were still capable of forming a complex with TRAF6 (Fig. 4E), indicating that these specific IRAK modification mutations do not affect the ability of IRAK to interact with TRAF6. Although previous studies suggested that MEKK3 plays an important role in IL-1-induced NFB activation (28), we recently observed that IL-1 stimulation leads to interaction of IRAK and TRAF6 with MEKK3 (Fig. 4, F and G). Interestingly, the IRAK modification mutants retained the ability to interact with MEKK3 (Fig. 4F) and were also able to mediate the interaction between TRAF6 and MEKK3 (Fig. 4G), implicating a specific role of MEKK3 in IL-1-induced IRAK-mediated signaling. Taken together, the above results suggest that the IRAK modification mutants may mediate the alternative NFB activation pathway through their interaction with MEKK3 and TRAF6. It is important to note that although endogenous IRAK and TRAF6 formed a complex with both TAK1 (Fig. 4A) (15) and MEKK3 (Fig. 4, F and G), the interaction between endogenous TAK1 and MEKK3 was not detected upon IL-1 treatment (Fig. 4F), suggesting that IRAK-TRAF6-TAK1 and IRAK-TRAF6-MEKK3 are probably in two separate IL-1-induced complexes.
IL-1-induced NFB Activation Is Mediated by Both TAK1 and MEKK3-Although IRAK modification mutants are defective in several aspects of IL-1 signaling, including TAK1 activation, these mutants were still able to mediate IL-1-induced IB␣ phosphorylation and NFB activation, suggesting a TAK1-independent NFB activation pathway. Recently, TAK1-deficient mice were generated by gene targeting using the Cre-loxP system (29) and gene trapping strategy (30). Although TAK1-deficient MEFs have reduced IL-1-induced NFB activation (29,30), it is important to note that IL-1-induced NFB activation is not completely abolished in TAK1deficient cells (29,30). In this study we used TAK1-deficient (Map3k7 Ϫ/Ϫ ) MEFs generated by in vitro transduction of Cre in MEFs homologous for loxP-flanked (floxed) Map3k7 flox/flox , which were confirmed by genomic Southern, RT-PCR, and Western analysis (29). As shown in Fig. 5A, whereas IL-1-induced JNK activation was greatly reduced in TAK1-deficient MEFs as compared with that in control MEFs, IL-1-induced NFB DNA binding activity was partially defective in TAK1deficient MEFs. Interestingly, although IL-1-induced IB␣ degradation was abolished in TAK1-deficient MEFs, IL-1 can lead to significant levels of IB␣ phosphorylation in these cells. We used the protein synthesis inhibitor cycloheximide to inhibit the new synthesis of IB␣ to better assess IL-1-induced IB␣ degradation. In the presence of cycloheximide, IL-1 stimulation clearly led to IB␣ degradation in wild type MEFs, whereas IL-1-induced IB degradation was still not detectable in TAK1deficient cells (Fig. 5B). The phenotype of the TAK1-deficient MEFs for the IL-1-induced signaling echoes our results described above for the IRAK-deficient cells transfected with IRAK modification mutants, including abolished IL-1-induced TAK1 activation, reduced JNK activation, and IL-1-induced IB degradation, retained IL-1-induced IB␣ phosphorylation, and NFB activation. Consistent with the phenotype of the TAK1-deficient MEFs, a TAK1 inhibitor (5Z-7-oxozeaenol (43)) that specifically inhibits TAK1 kinase activity abolished IL-1-induced JNK activation and IB degradation and retained IL-1-induced IB phosphorylation and NFB activation in wild type 293 cells (Fig. 5C and data not shown). The fact that IL-1induced NFB activation was reduced and JNK activation impaired in both TAK1-deficient MEFs and 293 cells treated with the TAK1 inhibitor reflects the levels of IL-1-mediated signaling committed to the TAK1-dependent pathway in these cells. However, we noted that IRAK modification mutants were able to mediate similar levels of IL-1-induced NFB activation as the wild type IRAK and partially restored the JNK activation without activating the TAK1-dependent pathway in IRAK-deficient cells derived from 293 parental cells (Fig. 3, B and E, and Fig. 4, B-D). These results suggest that mutations in IRAK modification probably shift the commitment of all of the IRAK protein to the TAK1-independent pathway, which is able to compensate for the loss of the TAK1-dependent NFB activation and partially for JNK activation.
The important question is as follows: what is this IRAK-mediated, TAK1-independent, IL-1-induced NFB activation pathway? Although previous studies reported an important role of MEKK3 in IL-1-induced NFB activation (28), we now observe that IL-1 stimulation leads to interaction of IRAK and TRAF6 with MEKK3 (Fig. 4, F and G). Importantly, the IRAK modification mutants did retain the ability to interact with MEKK3 ( Fig. 4F) and were also able to mediate the interaction between TRAF6 and MEKK3 (Fig. 4G), implicating a specific role of MEKK3 in IL-1-induced IRAK-mediated signaling. Therefore, we examined the possibility for MEKK3 to be an important player in the TAK1-independent NFB activation pathway. We used the TAK1 inhibitor (5Z-7-oxozeaenol) to abolish the TAK1 activity in both wild type and MEKK3-deficient MEFs. In wild type MEFs, although the TAK1 inhibitor abolished IL-1-induced JNK activation, it had little effect on IL-1-induced IB phosphorylation and partial inhibition on NFB activation. However, this inhibitor completely impaired IL-1-induced IB␣ phosphorylation and NFB activation in addition to the abolishment of IL-1-induced JNK activation in the MEKK3-deficient MEFs (Fig. 5D). These results suggest  MARCH 2, 2007 • VOLUME 282 • NUMBER 9 that IL-1-induced IB␣ phosphorylation is completely abolished only when both TAK1 and MEKK3 are impaired, suggesting that MEKK3 may play a critical role in the IL-1-induced TAK1-independent NFB activation pathway.

TAK1-Versus MEKK3-dependent NFB Activation Pathways
IRAK Modification Is Required for IL-1-induced TAK1-dependent IKK␣/␤ Phosphorylation-Because the above results suggest that IRAK modification plays an essential role in mediating the TAK1-dependent versus MEKK3-dependent NFB activation pathway, it is important to examine the impact of IRAK modification on the IKK complex. Previous studies have shown that IL-1-induced TAK1 activation leads to IKK␣/␤ phosphorylation (23). IL-1-induced IKK␣/␤ phosphorylation was completely abolished in IRAK-deficient cells transfected with IRAK phosphorylation and ubiquitination mutants (Fig.  6A). These results suggest that IRAK modification is required for IL-1-induced IKK␣/␤ phosphorylation, probably through its impact on TAK1 activation (23). Consistent with these find-ings, in TAK1 Ϫ/Ϫ MEFs, IL-1-induced IKK␣/␤ phosphorylation was greatly reduced (Fig. 6B).
We next examined whether mutations in IRAK modification have any impact on IL-1-induced activation of the IKK kinase activity. As shown in Fig. 6C, although IL-1 stimulation leads to IKK activation in 293 cells, IL-1-induced IKK activation was abolished in IRAK-deficient cells. Importantly, the IRAK ubiquitination mutant (K134R) and phosphorylation mutant ((S/T)A (131-144)) restored IL-1-induced IKK activation in cells to the similar level as DD ϩ UD ϩ C1 (the truncated IRAK with wild type phenotype) (Fig. 6C). Taken together, these results showed that although mutation in IRAK modification impaired TAK1 activation and IL-1induced IKK␣/␤ phosphorylation, the kinase activity of IKK was still activated in IRAK-deficient cells transfected with IRAK phosphorylation and ubiquitination mutants, confirming that IKK can be activated through a TAK1-independent NFB activation pathway.
IKK␣ Not IKK␤ Is Required for IL-1-induced TAK1-independent NFB Activation Pathway-We then examined the relative contribution of IKK␣ versus IKK␤ in the TAK1-dependent and MEKK3-dependent NFB activation pathways. Interestingly, as observed in TAK1-deficient MEFs (Fig. 5A), IB␣ was phosphorylated, but not degraded, in IKK␤ Ϫ/Ϫ MEFs in response to IL-1 stimulation (Fig. 7, A and B), implying that IKK␤ is required for the TAK1-dependent but not for the TAK1-independent NFB activation pathway. Importantly, although wild type IKK␤ restored IL-1-induced IB degradation, the kinase-inactive mutant of IKK␤ did not, indicating that the kinase activity of IKK␤ is required for the IL-1-induced IB␣ degradation (Fig.  7, A and B). The fact that IL-1-induced IB␣ phosphorylation FIGURE 5. IL-1-induced NFB activation is mediated by both TAK1 and MEKK3. A and B, wild type or TAK1-deficient MEFs were either untreated or treated with IL-1 (10 ng/ml) for the indicated times. B, cells were pretreated with cycloheximide (20 g/ml) for 2 h prior to IL-1 stimulation. Cell lysates were analyzed by Western blots via anti-pJNK, JNK, pIB, IB, actin, or NFB gel shift assay. C, 293 cells were untreated or pretreated with 600 nM of TAK1 inhibitor for 3 h prior to IL-1 stimulation (10 ng/ml). Cell lysates were analyzed by Western blots via anti-pIB, IB, actin, or NFB gel shift assay. The degradation of IB␣ was analyzed by Scion Image 1.62C alias and presented as relative percentage of the untreated samples. D, wild type or MEKK3deficient MEFs were pretreated with 600 nM of TAK1 inhibitor for 3 h prior to IL-1 treatment (10 ng/ml). Cell lysates were analyzed by Western blots with anti-pJNK, pIB, or actin.
NFB Is Activated through IB Phosphorylation and Dissociation without IB Degradation in the TAK1-independent NFB Activation Pathway-The dogma for NFB activation is that signal-induced phosphorylation of IB␣ targets this inhibitor of NFB for ubiquitination and subsequent degradation, thus allowing NFB to enter the nucleus to turn on the target gene (45). As presented above, our results showed that whereas IL-1 stimulation leads to IB␣ phosphorylation and degradation in wild type cells, IL-1 treatment causes IB␣ phosphorylation and NFB activation but not IB␣ degradation in IRAK-deficient cells transfected with IRAK modification mutants (Fig. 3, A-D) and TAK1-deficient MEFs (Fig. 5, A-C). One possibility is that the IL-1-induced IB␣ phosphorylation mediated by the TAK1-independent pathway directly leads to dissociation of IB␣ from NFB without IB␣ degradation. The dissociated NFB then migrates to the nucleus to activate gene transcription. To test this hypothesis, we examined the NFB-IB complex in wild type and TAK1-deficient MEFs upon IL-1 stimulation. NFB was immunoprecipitated from cell lysates of wild type and TAK1-deficient MEFs with and without IL-1 stimulation, followed by Western analysis with antibodies against IB␣ and NFB. As expected, IL-1 stimulation induced IB␣ phosphorylation, IB␣ degradation, and liberation of NFB in wild type MEFs (Fig. 8A). Interestingly, in TAK1-deficient cells, whereas IL-1 led to IB␣ phosphorylation but not IB␣ degradation, IB␣ was dissociated from NFB upon IL-1 stimulation (Fig.  8A). Similarly, although IL-1 stimulation did not lead to IB␣ degradation in IRAK-deficient cells transfected with IRAK modification mutants, IB␣ was dissociated from NFB in these cells upon IL-1 stimulation (Fig. 8B). Taken together, these results indicate that NFB can also be activated through IB␣ phosphorylation and subsequent dissociation from NFB without IB␣ degradation.
The important questions for this unique mechanism of NFB activation are how the phosphorylated IB␣ mediated by the MEKK3-dependent pathway escapes the ubiquitination FIGURE 6. The impact of IRAK modification and TAK1 deficiency on IL-1-induced IKK activation and IKK␣/␤ phosphorylation. A and C, 293 wild type cells, IRAK-deficient cells and IRAK-deficient cells stably transfected with IRAK modification mutants in the truncated IRAK (DD ϩ UD ϩ C1, (S/T)A (131-144), and K134R) were either untreated or treated with IL-1 (10 ng/ml) for the indicated times. Cell lysates were either analyzed via Western blots with anti-p-IKK␣/␤ and anti-IKK␣/␤ (〈) or immunoprecipitated with anti-IKK␣/␤, followed by a kinase assay using GST-IB-(1-54) as a substrate (C). B, wild type or TAK1-deficient MEFs were either untreated or treated with IL-1 (10 ng/ml) for the indicated times. Cell lysates were analyzed via Western blots with anti-p-IKK␣/␤ and IKK␤. D, 293 wild type cells and IRAK-deficient cells stably transfected with IRAK modification mutants in the truncated IRAK (DD ϩ UD ϩ C1, and (S/T)A (131-144)) were either untreated or treated with IL-1 (10 ng/ml) for the indicated times. Cell lysates were analyzed via Western blots with anti-p-IKK␥, anti-IKK␥, and anti-actin. E, human synoviocytes were pretreated with 600 nM TAK1 inhibitor for 3 h prior to IL-1 stimulation. Cell lysates were analyzed by Western blots with anti-p-IKK␥, anti-IKK␥, and anti-GRB2. MARCH 2, 2007 • VOLUME 282 • NUMBER 9

TAK1-Versus MEKK3-dependent NFB Activation Pathways
and subsequent degradation and how the phosphorylated IB␣ dissociates from NFB without degradation. Ubiquitination of IB is carried out by a ubiquitin-conjugating enzyme of the UBC4/5 family and SCF-␤TrCP E3 ligase. IL-1 stimulation leads to IB␣ phosphorylation on both Ser 32 and Ser 36 . ␤TrCP E3 ligase binds specifically to the phosphorylated form of IB (DS(PO 3 )GLDS(PO 3 )) and brings it the SCF complex for polyubiquitination, which is then selectively degraded by the 26 S proteasome. Interestingly, we found that the phosphorylated IB␣ in I1A cells transfected with IRAK modification mutants failed to bind to ␤-TrcP, suggesting that IB␣ might be phosphorylated differently by the TAK1-independent pathway (Fig.  8C). To address this issue, we performed IKK kinase assay in wild type and TAK1-deficient MEFs using wild type IB␣ and IB␣ mutants (S32A, S36A, and S32A/S36A) as substrates (Fig.  8D). Interestingly, although the IKK complex from both the wild type and TAK1-deficient MEFs can phosphorylate wild type and S32A mutant IB␣, S36A mutant IB␣ can only be phosphorylated by the IKK complex precipitated from wild type but not TAK1-deficient MEFs. As expected, S32A/S36A cannot be phosphorylated by IKK from either wild type or TAK1-deficient MEFs. Taken together, the above results suggest that Ser 36 is probably the major phosphorylation site for the IKK complex in the absence of TAK1.
Alternative NFB Activation Pathway in Primary Intestinal Epithelial Cells-Through the manipulation of the IRAK molecule, we observed two distinct IL-1-induced NFB activation pathways (Fig. 1). The TAK1-dependent pathway leads to IKK␣/␤ phosphorylation and IKK␤ activation, resulting in classical NFB activation through IB␣ phosphorylation and degradation. Our results suggest the presence of a TAK1-independent MEKK3dependent pathway that involves IKK␥ phosphorylation and IKK␣ activation, resulting in NFB activation through IB␣ phosphorylation but without IB degradation. Interestingly, this alternative NFB activation pathway was observed in primary epithelial cells. As shown in Fig. 9, A and B, IL-1 and LPS stimulation leads to IB␣ phosphorylation and degradation in bone marrow-derived macrophages, whereas the same stimuli could only lead to IB␣ phosphorylation but not degradation in primary epithelial cells. These results suggest that the regulation of these two NFB activation pathways may modulate important physiological functions, because the two pathways were preferentially utilized by different primary cells. Although it is unclear how these two pathways are regulated in vivo, we did observe differential usage of the two pathways when cells were stimulated with different concentrations of IL-1. When 293 cells were stimulated with a low concentration of IL-1 (0.1 ng/ml), IB␣ was phosphorylated but with reduced degradation as compared with the cells stimulated with a high concentration of IL-1 (10 ng/ml), implicating regulation of the two pathways at the receptor complex (Fig. 9C).

DISCUSSION
This study reports the discovery of the co-existence of the two parallel IL-1-mediated TAK1-dependent and MEKK3-de- FIGURE 7. The role of IKK␣ and IKK␤ in TAK1-dependent and -independent NFB activation pathways. A and B, wild type, IKK␤-deficient MEFs, and IKK␤-deficient MEFs infected with adenovirus containing wild type IKK␤ or kinase-dead IKK␤ were either untreated or treated with IL-1 (10 ng/ml) for the indicated times. Cell lysates were analyzed with anti-pIKK␣/␤, pIB␣, IB␣, actin, or pJNK. The degradation of IB␣ was analyzed by Scion Image 1.62C alias and presented as relative percentage of the untreated samples. C, IKK␣ is required for IL-1-induced IB␣ phosphorylation mediated by the TAK1-independent pathway. Wild type or IKK␣-deficient MEFs were either untreated or treated with IL-1 (10 ng/ml), with or without TAK1 inhibitors (600 nM). Cell lysates were analyzed with anti-pIB␣, pJNK, and actin. D, wild type or IKK␣/␤-deficient MEFs were either untreated or treated with IL-1 (10 ng/ml) for the indicated times. Cell lysates were analyzed with anti-pJNK, pIB␣. pendent signaling pathways for NFB activation (Fig. 1). These two pathways are regulated at the level of IRAK modification. The TAK1-dependent pathway causes IKK␣/␤ phosphorylation and IKK␤ activation, leading to classical NFB activation through IB␣ phosphorylation and degradation. The TAK1-independent MEKK3-dependent pathway induces IKK␥ phosphorylation and IKK␣ activation, resulting in NFB activation through IB␣ phosphorylation and subsequent dissociation from NFB but without IB␣ degradation. It is important to note that we recently found that TLR8-mediated NFB and JNK activation are TAK1-independent and MEKK3-dependent (46), suggesting a regulatory mechanism at the level of receptor complexes that determines the usage of TAK1-dependent versus MEKK3dependent pathways in IL-1R/TLR signaling.
Upon IL-1 stimulation, IRAK becomes heavily phosphorylated. Several phosphorylation sites have been identified in the kinase domain of IRAK, including Thr 209 and Thr 387 (36). Both of these phosphorylation sites are critical for the kinase activity of IRAK. Although Thr 209 phosphorylation regulates the opening of IRAK kinase domain, Thr 387 phosphorylation regulates full enzymatic activity of IRAK. In this study, we focused on the UD domain (undetermined domain, amino acids 103-198), also referred to as Pro-ST region (rich in proline, serine, and threonine). The UD domain (Pro-ST region) is the site of IRAK hyperphosphorylation (36,37). We identified a cluster of six serines and threonines (Ser 131 , Ser 137 , Ser 138 , Ser 140 , Thr 141 , and Ser 144 ) in this UD domain. Mutation of these serines and threonines ((S/T)A (131-144)) severely impaired IL-1-induced IRAK phosphorylation, ubiquitination, and degradation. Previous studies showed that IRAK phosphorylation is required for IRAK ubiquitination (42). The phosphorylation of these sites ((S/T) (131-144)) is probably required for IL-1-induced ubiquitination, because we identified the adjacent lysine 134 as a critical ubiquitination site. Point mutation of lysine 134 to arginine indeed abolished IL-1-induced IRAK ubiquitination. Point mutation of serine 138 to alanine in the truncated form IRAK showed a partial defect in IL-1-induced IRAK phosphorylation. 3 Future effort is required to mutate these serines and threonines ((S/T) (131-144)) individually or in combination to identify the specific site required for mediating IRAK ubiquitination. Because the IRAK phosphorylation mutants were all created in the kinase-inactive IRAK molecule, these sites are probably phosphorylated by IRAK4. Our previous studies showed that the kinase activity of IRAK and IRAK4 might be redundant in the IL-1 pathway, suggesting that serines and threonines ((S/T) (131-144)) might also be autophosphorylated by IRAK (12). Nevertheless, it is also possible that different sites are autophosphorylated in the UD domain.
The fact that the IRAK ubiquitination mutants are only defective in IL-1-induced IRAK ubiquitination but not IRAK phosphorylation suggests that it is the lack of ubiquitination not phosphorylation responsible for the abolished IL-1-induced signaling events. The defect in IL-1 signaling observed with the IRAK phosphorylation mutants ((S/T)A (131-144), IRAK1KD pmt) is likely due to the lack of ubiquitination of these mutants, because that IRAK phosphorylation is required for its ubiquitination. The fact that the IRAK ubiquitination mutants retained their interaction with TRAF6, but failed to interact with TAK1 and activate TAK1, suggests that IRAK ubiquitination is necessary for the recruitment of TAK1 to the IRAK-TRAF6 complex, and the formation of IRAK-TRAF6-TAK1 complex is necessary for the activation of TAK1. Polyubiquitination of a target protein with the ubiquitin linked through Lys 48 is recognized by proteasome and ultimately degraded. However, polyubiquitination chains linked through Lys 63 of ubiquitin do not target the substrate for proteasome-mediated degradation, mediating protein-protein interaction and cell signaling instead. It has been reported that TRAF6-mediated Lys 63 polyubiquitination plays an essential role in the activation of TAK1, IKK, and NFB. However, TRAF6 is probably not the E3 for IRAK ubiquitination, because IRAK is still ubiquitinated and degraded in TRAF6-deficient MEFs. 3 We have found that polyubiquitination chains on IRAK are linked through both Lys 48 and Lys 63 . 3 We and others recently found that Pellinos proteins can ubiquitinate IRAK, although the precise function of Pellinos in IL-1 signaling is still unclear (47). 3 Because Pellinos do not lead to the degradation of IRAK, they are unlikely the ubiquitin protein ligase E3s responsible for the specific IRAK ubiquitination that leads to IRAK degradation (47). 3 At the moment, it is unknown which ubiquitin protein ligase E3(s) is responsible for the IL-1induced ubiquitination of IRAK required for the activation of the TAK1-dependent NFB activation pathway. Although our results suggest that IL-1-induced ubiquitination and degradation of IRAK are required for the TAK1-dependent NFB activation pathway, it is also possible that only the IL-1-induced IRAK ubiquitination is the necessary biochemical modification required for the activation of the TAK1-dependent pathway, and the IL-1-induced IRAK degradation is the consequence of activation. On the other hand, we found that the proteasome inhibitor MG-132 blocked IL-1-induced TAK1 activation, suggesting that IL-1-induced IRAK degradation might be a necessary step in the activation of the TAK1-dependent pathway. 3 One possible role for IL-1-induced IRAK degradation in the activation of TAK1 is to release the TAK1 complex from the membrane-associated IRAK-TRAF6-TAK1-TAB2/3 complex, because we have previously shown that modified IRAK is always associated with the membrane, and only the TAK1 complex is released to the cytosol (15).
Another important finding of this study is the identification of the IL-1-induced TAK1-independent NFB activation pathway. Although the IRAK modification mutants are defective in the TAK1-mediated NFB activation pathway, these IRAK mutants are still capable of mediating IL-1-induced IB␣ phosphorylation and NFB activation but not IB␣ degradation, implicating a TAK1-independent NFB activation pathway. Consistent with this conclusion, in TAK1 Ϫ/Ϫ MEFs, IL-1 stimulation could lead to IB␣ phosphorylation and a significant level of NFB activation without IL-1-induced IKK␣/␤ phosphorylation and a lack of IB␣ degradation, confirming a TAK1-independent NFB activation pathway. Importantly, we found that the IRAK modification mutants retained the interaction not only with TRAF6 but also with MEKK3, implicating TRAF6 and MEKK3 in the TAK1-independent NFB activa-tion pathway. Whereas Blonska et al. (48) recently reported that TAK1-MEKK3 forms a complex to mediate tumor necrosis factor-␣-induced NFB activation, direct interaction between endogenous TAK1 and MEKK3 was not detected upon IL-1 stimulation. Although endogenous IRAK and TRAF6 interact with both endogenous TAK1 and MEKK3 in 293 cells in response to IL-1 stimulation, we did not detect any interaction between TAK1 and MEKK3. Our results suggest that IRAK-TRAF6-TAK1 and IRAK-TRAF6-MEKK3 are two independent IL-1-induced complexes. In support of this, we found that the impairment of both TAK1 and MEKK3 completely abolished IL-1-induced NFB activation, by using a TAK1 inhibitor in MEKK3-deficient cells and knockdown of MEKK3 in TAK1-deficient cells with MEKK3 small interfering RNA. The fact that IL-1-induced NFB activation is completely abolished in TRAF6-deficient cells indicates that TRAF6 is probably required for both TAK1-dependent and MEKK3-dependent NFB activation pathways. Taken together, our results imply that IL-1 mediates TAK1-dependent and MEKK3-dependent NFB activation pathways, which bifurcate at the level of IRAK modification.
The differences between IL-1-induced TAK1-dependent and TAK1-independent pathways are also reflected at the level of the IKK complex. Previous studies suggest that ligand-induced phos-phorylation of IKK␣/␤ plays a critical role in IKK activation, especially in the activation of IKK␤ (49). In TAK1-deficient MEFs, IL-1-induced IKK␣/␤ phosphorylation is impaired, suggesting that IKK␣/␤ phosphorylation plays an important role in TAK1-mediated IKK activation. Interestingly, although IL-1 treatment does not lead to IKK␣/␤ phosphorylation in TAK1-deficient MEFs, IL-1 still activates the kinase activity of IKK in these cells. These results suggest that IKK can be activated in the absence of TAK1 through a differential mechanism. Furthermore, using IKK␣-, IKK␤-, and IKK␣/␤-deficient MEFs, we showed that IKK␣ but not IKK␤ is required for IL-1-induced TAK1independent NFB activation pathways, presumably through the MEKK3-dependent pathway. It is intriguing that in the absence of TAK1 activation, IL-1-induced IKK␥ phosphorylation can still take place, suggesting a potential role of IKK␥ phosphorylation in the TAK1-independent IKK activation cascade. It is possible that MEKK3 mediates the phosphorylation of IKK␥, which in turn alters the conformation of the IKK complex and leads to its activation. Because IKK␣ but not IKK␤ has been implicated in the MEKK3-dependent pathway, it is possible that MEKK3 only specifically activates an IKK␥-IKK␣-I⌲⌲␣ complex. Future studies are required to delineate this unique IKK activation mechanism.
Thedogma for NFB activation is that signal-induced phosphorylation of IB␣ targets this inhibitor of NFB for ubiquitination and subsequent degradation, thus allowing NFB to enter the nucleus to turn on the target gene (45). We report here that IL-1-induced IB␣ phosphorylation mediated by the TAK1-independent MEKK3-dependent pathway directly leads to dissociation of IB␣ from NFB without IB␣ degradation. Our results reveal an alternative mechanism to activate NFB. We found that the phosphorylated IB␣ mediated by the TAK1-independent MEKK3-dependent pathway failed to interact with ␤TrCP E3 ligase, which is probably why this phosphorylated IB␣ is not degraded. . Dissociation of IB␣ from NFB in the TAK1-independent pathway. A, cell lysates from wild type and TAK1-deficient MEFs untreated or treated with IL-1 for the indicated times were immunoprecipitated (IP) with anti-p65 (NFB), followed by Western analysis with anti-IB and anti-p65 (NFB) antibodies. The dissociation of IB␣ was analyzed by Scion Image 1.62C alias and presented as relative percentage of the untreated samples. Cell lysates were also analyzed directly by Western blots with anti-IB, anti-p65 (NFB), and anti-actin antibodies. The degradation of IB␣ was analyzed by Scion Image 1.62C alias and presented as relative percentage of the untreated samples. B, cell lysates from IRAK-deficient cells transfected with DD ϩ UD ϩ C1, (S/T)A (131-144), and K134R untreated or treated with IL-1 for the indicated times were immunoprecipitated with anti-p65 (NFB), followed by Western analysis with anti-IB and anti-p65 (NFB) antibodies. The dissociation of IB␣ was analyzed by Scion Image 1.62C alias and presented as relative percentage of the untreated samples. Cell lysates were also analyzed directly by Western blots with anti-IB, anti-p65 (NFB), and anti-actin antibodies. The degradation of IB␣ was analyzed by Scion Image 1.62C alias and presented as relative percentage of the untreated samples. C, IRAK-deficient cells stably transfected with MyC-␤TrCP and IRAK modification mutants in the truncated IRAK (DD ϩ UD ϩ C1, (S/T)A (131-144), and K134R) were either untreated or treated with IL-1 (10 ng/ml) for the indicated times. Cell lysates were immunoprecipitated with anti-MyC and probed with anti-pIB␣ and anti-MyC. D, cell lysates from wild type (WT) (ϩ/ϩ) and TAK1-deficient (Ϫ/Ϫ) MEFs were immunoprecipitated with anti-IKK␣/␤, followed by a kinase assay using GST-IB-(1-54) or GST-IB␣-(1-54) mutants (S36A, S32A, S32A/S36A) as substrates. Interestingly, our results also showed that Ser 36 is the major phosphorylation site of IB␣ mediated by the TAK1-independent MEKK3-dependent pathway. Although phosphorylation on Ser 36 but not on Ser 32 of IB␣ probably prevents this unique form of phosphor-IB␣ from interaction with ␤TrCP E3, the Ser 36 -phosphorylated IB␣ might fold into a different conformation, leading to its dissociation from NFB. Our results showed that the kinase activity of IKK␤ is required for IL-1-induced IB degradation (Fig. 7B), suggesting the requirement of IKK␤ for IB␣ phosphorylation on Ser 32 . Previous studies showed that phosphorylation of IB␣ in the C-terminal PEST region is also critical for IB␣ degradation (45,50). Future studies are required to carefully compare the phosphorylation patterns of IB␣ mediated by the TAK1-dependent versus MEKK3-dependent pathway, which will help us to investigate the detailed molecular mechanisms of these two parallel IL-1-mediated NFB activation pathways. Future studies are also required to understand the regulatory mechanism at the level of receptor complexes that determines the usage of TAK1-dependent versus MEKK3-dependent pathways. The physiological significance of this novel TAK1-independent MEKK3-dependent pathway still needs to be further investigated. We found that this novel pathway is predominantly present in primary intestine/colon epithelial cells. Therefore, it is possible that this unique pathway plays a critical role in maintaining the homeostasis of epithelium.