Tumor Necrosis Factor-α-induced IKK Phosphorylation of NF-κB p65 on Serine 536 Is Mediated through the TRAF2, TRAF5, and TAK1 Signaling Pathway

The activation of NF-κB has been shown to be regulated by multiple phosphorylations of IκBs and the NF-κB p65 subunit. Here, we characterized the intracellular signaling pathway leading to phosphorylation of p65 on Ser-536 using a novel anti-phospho-p65 (Ser-536) antibody. The Ser-536 of endogenous p65 was rapidly phosphorylated in response to a wide variety of NF-κB stimulants including TNF-α in the cytoplasm and rapidly dephosphorylated in the nucleus. The TNF-α-but not IL-1β-induced Ser-536 phosphorylation was severely impaired in murine embryonic fibroblasts derived from traf2-/-traf5-/- mice. Bay 11-7082, an inhibitor of IκB phosphorylation, inhibited the TNF-α-induced phosphorylation in vivo. In addition, overexpression of TGF-β-activated kinase 1 (TAK1), IKKα and IKKβ stimulated the phosphorylation, and their dominant negative mutants blocked the TNF-α-induced phosphorylation. Moreover, small interfering RNAs (siRNAs) against TAK1, IKKα and IKKβ blocked the phosphorylation of endogenous p65. On the other hand, calyculin-A, a protein phosphatase inhibitor, blocked the dephosphorylation in the nucleus in vivo. These results indicate that similar signaling pathways were utilized for the phosphorylations of IκBα and p65, which further support the idea that both IκB and NF-κB are substrates for the IKK complex in the activation of NF-κB.

Recent attention has been focused on the molecular mechanisms for the transactivation of NF-B (27,28). While all of the Rel proteins share a Rel homology domain (RHD) controlling DNA-binding activity, only p65, RelB, and c-Rel contain COOH-terminal transactivation domains (TADs) (1). Therefore, the only TAD in the p65 subunit can regulate the transcriptional activity of the prototypical p65/p50 NF-B complex. In addition to stimulus-induced nuclear translocation of NF-B, several lines of evidence suggest that stimulus-induced phosphorylation of the p65 subunit plays a key role in the transcriptional activation after the nuclear translocation (29 -33). The sites of phosphorylation reported to date are Ser-276 in RHD (29,33) and Ser-529/Ser-536 in TAD (34 -36). Several candidate kinases that phosphorylate each serine residue have been identified, such as protein kinase A and mitogen-and stress-activated protein kinase-1 (MSK1) for Ser-276 (29,37), casein kinase II for Ser-529 (35), and IKK for Ser-536 (36). It has been shown that phosphorylation at Ser-276 is critical for p65 NF-B activation by regulating the interaction with coactivators p300/CBP or the histone deacetylase HDAC-1 (29,30). The phosphatidylinositol 3Ј-kinase (PI3K)-Akt pathway has been reported to regulate the TNF-␣-and IL-1-induced transcriptional activation (38 -41). Furthermore, NF-B transcriptional activity, but not the nuclear translocation, has been shown to be impaired in mouse embryonic fibroblasts deficient for glycogen synthase kinase (GSK)-3␤ (42), which was recently found to be able to phosphorylate p65 in vitro (43). Collectively, all this evidence further supports the importance of p65 phosphorylation in the nuclear function of NF-B. However, the molecular mechanism of p65 phosphorylation remains to be characterized.
In contrast to the nuclear function, the intracellular signaling pathway leading to p65 phosphorylation is still unclear. We previously demonstrated that TNF-␣ induces the phosphorylation by IKK of p65 on Ser-536 using in vitro kinase assay and overexpression experiments in vivo (36). In this study, we have investigated the regulatory mechanisms for the phosphorylation of p65 on Ser-536 under physiological conditions using a novel anti-phospho-p65 antibody specifically recognizing the Ser-536 phosphorylated form. We found that phosphorylation of Ser-536 was widely observed in accordance with the phosphorylation of IB␣ and the phosphorylated Ser-536 was rapidly dephosphoylated by certain protein phosphatases in the nucleus.
Immunoblotting-Cell lysates or immunoprecipitates were resolved by SDS-PAGE and transferred to Immobilon-P nylon membrane (Millipore). The membrane was treated with BlockAce (Dainippon pharmaceutical Co. Ltd, Suita, Japan) overnight at 4°C and probed with primary antibodies as described above. The antibodies were detected using horseradish peroxidase-conjugated anti-rabbit, anti-mouse, and anti-goat IgG (DAKO), and visualized with the ECL system (Amersham Biosciences).
In Vitro Kinase Assay-Recombinant human His 6 -tagged IKK␤ (hr-IKK␤) was expressed in Sf21 cells infected with recombinant Baculoviruses generated by co-transfection with BaculoGold DNA and FIG. 1. Specificity of anti-phospho-p65 (Ser-536) antibody to the phosphorylated Ser-536. A, HeLa cells were transfected with expression vectors for wild type or S536A-mutated HA-tagged p65 (1 g) with IB␣ (1 g). Total DNA amounts were adjusted to 2 g with an empty vector. Twenty-four hours after the transfection, cells were stimulated with or without TNF-␣ (20 ng/ml) for 5 min. Whole cell lysates were immunoprecipitated with an anti-HA tag antibody and the immunoprecipitates were immunoblotted with anti-phospho-p65 (S536P) (upper panel) and anti-p65 (lower panel) antibodies. B, p65 knockout MEFs mock-transfected and reconstituted with wild type or S536A-mutated p65 by retroviral vectors were stimulated with TNF-␣ (20 ng/ml) for 10 min. Whole cell lysates were immunoblotted with anti-phospho-p65 (upper panel) and anti-p65 (lower panel) antibodies.

Specificity of a Novel Anti-phospho-p65 Antibody-Previ-
ously, we demonstrated that TNF-␣-induced phosphorylation of p65 on Ser-536 in HeLa cells (36). In the present study, we have used a novel phospho-p65 antibody to characterize the phosphorylation of p65 on Ser-536 in vivo. First, we confirmed the specificity of this antibody. HeLa cells were transfected with wild type or S536A-mutated HA-p65 with IB␣ and stimulated with TNF-␣. HA-p65 was immunoprecipitated with an anti-HA tag antibody, and the immunoprecipitates were immunoblotted with the phospho-p65 antibody. TNF-␣ induced Ser-536 phosphorylation of wild type p65 but not S536A mutant ( Fig. 2A). To further explore the phosphorylation in vivo, we tried to test the phosphorylation in p65 Ϫ/Ϫ mouse embryo fibroblasts (MEFs) stably transfected with wild type p65 or S536A mutant using retroviral vectors (Fig. 1B). The expression levels of wild type and S536A mutant p65 were almost equivalent to the endogenous level of p65 ϩ/ϩ MEFs (33). TNF-␣-induced phosphorylation was detected in wild type p65transfected MEFs, but not in mock-transfected and p65-S536Atransfected MEFs (Fig. 1B). In contrast, phosphorylation of c-Jun N-terminal kinase (JNK) was intact in these MEFs (data not shown). These results indicate that the phospho-p65 antibody specifically detects Ser-536 phosphorylation in vivo.
TNF-␣-induced Ser-536 Phosphorylation of p65 in Vivo-Previously, we demonstrated that TNF-␣ induced phosphorylation of overexpressed wild-type p65, but not p65-S536A mutant in the metabolic labeling experiments using HeLa cells (36). However, there is no evidence that the phosphorylation occurs under physiological conditions. Here, we investigated the TNF-␣-induced phosphorylation of endogenous p65 on Ser-536 using the anti-phospho-p65 antibody. Immunoblotting of whole cell lysates showed that TNF-␣ stimulated the phosphorylation of endogenous p65 on Ser-536 in HeLa cells ( Fig. 2A). The phosphorylation was detected at 2 min after TNF-␣ stimulation and preceded the phosphorylation and degradation of IB␣, suggesting that the phosphorylation occurred prior to nuclear translocation. Interestingly, the phospho-p65 antibody detected doublet bands at 5 min after the stimulation, indicating another modification of p65. After the peak of phosphorylation at 5 min, Ser-536 was rapidly dephosphorylated. A similar time-dependent phosphorylation was observed in TNF-␣stimulated human embryonic kidney 293 cells (Fig. 2B). In contrast, the Ser-536 phosphorylation was sustained in RAW264.7 murine macrophage cells stimulated with Toll-like receptor ligands, lipopolysaccaride (LPS) (Fig. 2C), and unmethylated CpG oligonucleotides (data not shown), and in Jurkat T cells stimulated with phorbol ester plus Ca 2ϩ ionophore A23187 (Fig. 2D). These results indicate that the phosphorylation of p65 on Ser-536 was widely observed in the activation of NF-B. However, the time course of the phosphorylation was dependent on cell types or stimulants.
Phosphorylation of p65 on Ser-536 in the Cytoplasm-The time course experiment raised the possibility that the phosphorylation of Ser-536 occurred in the cytoplasm (Fig. 2). To clarify this possibility, the effect of a proteasome inhibitor, N-acetylleucyl-leucyl-norleucinal (ALLN), on the TNF-␣-induced phosphorylation on Ser-536 was examined (Fig. 3A). Treatment with ALLN inhibited the degradation of IB␣, resulting in the accumulation of the phosphorylated form of IB␣. However, the phosphorylation was not inhibited, but rather increased, by ALLN. The increase may be due to isolation from the dephosphorylating activity in the nucleus, because ALLN inhibited the nuclear translocation of p65. We next performed an in vitro kinase assay using GST-p65 (354 -551) as a substrate to detect Ser-536 phosphorylating activity in the cytoplasm (Fig. 3B). The activity was detected in cytoplasmic extracts prepared from TNF-␣-treated HeLa cells for 2-10 min. These results demonstrated that the activity to phosphorylate Ser-536 was induced in the cytoplasm.
Involvement of TAK1 and TAB1 in TNF-␣-induced Ser-536 Phosphorylation-We previously reported that the endogenous TAK1-TAB1 complex is rapidly activated by TNF-␣ prior to the activation of IKK and plays a key role in the activation of NF-B (26). Therefore, we next investigated the contribution of TAK1 in TNF-␣-induced Ser-536 phosphorylation (Fig. 5A). Wild type TAK1 or kinase-inactive mutant (TAK1-KW) was overexpressed with or without its activator TAB1 in HEK293 cells and then whole cell lysates were assessed by immunoblotting with the anti-phospho-p65 antibody. When co-expression with TAB1, wild type TAK1, but not TAK1-KW, induced Ser-536 phosphorylation of endogenous p65, indicating that the phosphorylation was dependent on a kinase activity of TAK1. This is correlated with our previous observation that the co-expression of TAK1 and TAB1 induced activation of endogenous IKK complex in a TAK1 kinase activity-dependent manner (26). To further explore the role of TAK1 in the TNF-␣induced Ser-536 phosphorylation, we examined the phosphorylation of endogenous TAB1, an essential step for TAK1 activation by TAB1 (48). As shown by the delayed migration on SDS-PAGE, phosphorylation of TAB1 could be detected in TNF-␣-treated HeLa (Fig. 5B) and HEK293 cells (data not shown), indicating the activation of TAK1-TAB1 complex in TNF-␣-treated cells. In addition, a dominant-negative mutant of TAK1 (TAK1-KW) inhibited the TNF-␣-induced phosphorylation of p65 on Ser-536 (Fig. 5C). Furthermore, siRNA against TAK1 induced specific knockdown of TAK1 expression without affecting p65, IKK␣, and IKK␤ expression and inhibited the TNF-␣-induced phosphorylation of endogenous p65 as well as IB␣ degradation (Fig. 5D). In contrast, while siRNA against luciferase (luc) specifically inhibited firefly luciferase gene expression (data not shown), it did not affect the p65 phosphorylation (Fig. 5D). These results indicate that TAK1-TAB1 is involved in the TNF-␣-induced phosphorylation of p65 on Ser-536.
Critical Roles of IKK␣ and IKK␤ in TNF-␣-induced Ser-536 Phosphorylation-We first examined the effect of Bay 11-7082, a chemical inhibitor of IB␣ phosphorylation (49), on the TNF-␣-induced p65 phosphorylation. The compound inhibited the TNF-␣-induced phosphorylation of p65 as well as IB␣ degradation without affecting p38 MAPK phosphorylation in vivo (Fig. 6), suggesting possible involvement of the IKK signaling pathway in the phosphorylation of endogenous p65.
We next examined the potential of IKKs as p65 kinases in vivo. To define the role of IKKs on the phosphorylation of p65 on Ser-536, wild type IKK␣ and IKK␤ were overexpressed (Fig.  7B). IKK␣ alone induced a slight phosphorylation of p65, while phosphorylation was clearly detected in cells transfected with IKK␤. The co-transfection of IKK␣ and IKK␤ resulted in a marked phosphorylation on Ser-536. To further explore the role of IKKs, we examined the effects of dominant-negative mutants of IKK␣ and IKK␤ on the TNF-␣-induced Ser-536 phos- phorylation (Fig. 7C). Both kinase-inactive IKK␣-K44M and IKK␤-K44M showed an inhibition of TNF-␣-induced phosphorylation. Furthermore, siRNAs against IKK␣ and IKK␤ in-duced specific knockdown of the expression of target proteins without affecting p65 and PCNA expression and inhibited the phosphorylation of endogenous p65 as well as the degradation of IB␣ (Fig. 7D). These results further support the idea that IKKs are one of the p65 kinases in the TNF-␣ signaling pathway.
Dephosphorylation of p65 on Ser-536 in the Nucleus-The time course experiment revealed that the phosphorylation of Ser-536 rapidly decreased at 10 -20 min after TNF-␣ stimulation, suggesting possible dephosphorylation in the nucleus. To assess this possibility, the level of phosphorylation of p65 was examined by immunoblotting using nuclear extracts prepared from TNF-␣-stimulated HeLa cells (Fig. 7). The nuclear translocation of p65 and the phosphorylation on Ser-536 were detected at 5 min after the stimulation. The level of phosphorylation was decreased at 20 min without a change in the protein level in the nuclear extracts, indicating dephosphorylation in the nucleus. We further examined the effect of calyculin A (CalyA), a phosphatase inhibitor for protein phosphatase (PP) 1 and PP2A (50), on the decrease of Ser-536 phosphorylation in HeLa cells (Fig. 8). Pretreatment with CalyA resulted in inhibition of the time-dependent dephosphorylation of Ser-536 at 20 min after TNF-␣ stimulation without affecting IB␣ degradation in the cytoplasm and PCNA expression in the nucleus. Collectively, these results indicate that p65 was phosphorylated on Ser-536 in the cytoplasm and dephosphorylated after the nuclear translocation by, at least in part, CalyA-sensitive phosphatases.

DISCUSSION
The study of NF-B activation has recently focused on the phosphorylation of the p65 subunit. Several lines of evidence have suggested that phosphorylation of p65 is involved in the transcriptional activation (27)(28)(29)(30)(31)(32)(33). However, the physiological function of the phosphorylation remains controversial. In addition, the signaling pathways leading to the phosphorylation under physiological conditions are still unclear. In this study, we first analyzed the TNF-␣-induced signaling pathway leading to the phosphorylation of endogenous p65 on Ser-536 by using a novel anti-phosho-p65 (Ser-536) antibody and obtained insight into the regulation of the phosphorylation.
Recent investigations have clarified the TNF-␣-induced signaling pathways leading to the phosphorylation of IB␣. Signaling proteins have been shown to mediate the extracellular signaling for the phosphorylation of IB in the cytoplasm (7-12, 14 -17). In the present study, we demonstrated that TRAF2 and TRAF5 play critical roles in the TNF-␣-induced phosphorylation of p65 on Ser-536. However, a slight inducible phosphorylation remained in traf2 Ϫ/Ϫ traf5 Ϫ/Ϫ MEFs, suggesting a TRAF2-and TRAF5-independent signaling pathway. This is correlated with previous observation that the slight activation of IKK could be detected in TNF-␣-stimulated traf2 Ϫ/Ϫ traf5 Ϫ/Ϫ MEFs (16), suggesting that the TRAF2-and TRAF5-independent signal leading to the p65 phosphorylation is also mediated through the remaining IKK activity.
We previously reported that both IKK␣ and IKK␤ can efficiently phosphorylate p65 on Ser-536 in vitro and their activi- ties are comparable to the activity phosphorylating IB␣ at Ser-32 and Ser-36, and that expression of IKK␣ and IKK␤ induced phosphorylation of Ser-536 in vivo (36). The time course of the TNF-␣-induced p65 phosphorylation is similar to the activation of IKK and the following IB␣ phosphorylation in HeLa cells (36). In the present study, we showed the inhibitory effects of dominant negative mutants, siRNAs and a chemical inhibitor of IKKs on the TNF-␣-induced Ser-536 phosphorylation of p65. In addition, Sizemore et al. (40) recently reported that the IKK complexes from IKK␣-and IKK␤-null MEFs were both deficient in TNF-␣-and IL-1-induced phosphorylation of the transactivation domain of p65. These results suggest that, however we could not rule out the possibility that p65 is phosphorylated by an undefined IKK-regulated kinase, IKK␣, and IKK␤ are potential candidates as the p65 kinases phosphorylating Ser-536.
TAK1 is a member of the MAP3K family that stimulates the activation of the IKK-NF-B signaling pathway as well as the MAPK-AP-1 pathways (25,26,(51)(52)(53)(54)(55). We previously reported that TAK1 is activated by TNF-␣ stimulation prior to the activation of the IKK complex in HeLa cells (26). In addition, TAK1 directly binds to both IKK␣ and IKK␤ subunits in the IKK complex (26). Recently, Takaesu et al. (45) reported that TAK1 interacted with TRAF2 upon TNF-␣ stimulation, and TNF-␣ and IL-1-induced activation of the IKK-NF-B pathway is impaired in HeLa cells transfected with siRNAs against TAK1. This evidence strongly suggests that TAK1 is a key molecule in the TNF-␣-induced activation of IKK. Consistent with the observations, we demonstrated that the expression of TAK1 with its activator protein TAB1 induced the phosphorylation of p65 on Ser-536. In addition, the dominant negative mutant and TAK1 siRNA blocked TNF-␣-induced Ser-536 phosphorylation, suggesting the critical role of TAK1 in the TNF-␣-induced p65 phosphorylation. However, we could not rule out the possibility that other MAP3Ks regulating the activation of IKK such as MEKK3 are involved in the phosphorylation of p65 upon stimulation with TNF-␣ (23,24). In addition, overexpression of NIK, a MAP3K regulating IKK␣-NF-B activation in lymphotoxin ␤ receptor (LT␤R) signaling (56,57), also induced the phosphorylation of p65 on Ser-536 (data not shown), which is correlated with the recent observation that stimulation of LT␤R by an agonistic antibody induces Ser-536 phosphorylation and the transactivation potential of NF-B in LT␤R signaling depends on the phosphorylation of p65 on Ser-536 (32). Taken together, the phosphorylation of p65 on Ser-536 is widely observed and is mediated through signaling pathways similar to the IB␣ phosphorylation elicited by several stimulants for NF-B activation (Fig. 9).
The p65 phosphorylated on Ser-536 was rapidly dephosphorylated in the nucleus in certain cell types. In HeLa cells, for example, the dephosphorylation was carried out, at least in part, by CalyA-sensitive phosphatases at 10 -20 min after TNF-␣ stimulation. In contrast, the p65 phosphorylation in TLRs-stimulated RAW264.7 cells and TPAϩA23187-stimulated Jurkat cells is sustained. CalyA is a potent inhibitor for serine/threonine protein phosphatase (PP) 1 and PP2A. Yang et al. (58) reported that PP2A directly interacts with p65 in melanocytes in vivo and p65, phosphorylated by immunoprecipitated IKK complex in vitro, is dephosphorylated by purified PP2A core enzyme and the constitutive activation of NF-B in melanoma is dependent on the decrease of PP2A activity. These observations suggest that PP2A is responsible for the dephosphorylation of Ser-536 in the nucleus. In any case, the majority of the nuclear p65 at 20 min after the stimulation is unphosphorylated on Ser-536 in HeLa cells. Future characterization of the rapid dephosphorylation in the nucleus helps our under-standings of the physiological function of the Ser-536 phosphorylation.
The COOH-terminal transactivation domain of p65 is composed of TA1 and TA2 domains. The TA1 domain (the COOHterminal 30 amino acids) plays critical roles in the transcriptional activation of NF-B and the Ser-536 residue is located in this domain (59,60). These findings suggest that the Ser-536 phosphorylation regulates the transcriptional activity of the p65 subunit of NF-B. As mentioned above, Jiang et al. (32) recently reported that NIK-IKK␣ cascade-dependent p65 phosphorylation on Ser-536 plays a crucial role in LT␤R-mediated NF-B activation. In addition, it has been shown that the phosphatidylinositol 3Ј-kinase (PI3K)-Akt serine/threonine kinase signaling pathway stimulates transactivation of the p65 subunit in response to TNF-␣ and IL-1, and the Akt-induced transactivation potential of p65 is impaired when Ser-536 is mutated to Ala (38 -41). In addition, IKK␣ and IKK␤ are essential for the PI3K-Akt-induced transcriptional activation of NF-B (40). Therefore, we examined the effect of a PI3K inhibitor LY294002 on the TNF-␣-induced Ser-536 phosphorylation in HeLa cells, however, no inhibitory effect could be detected (data not shown), indicating that the inducible phosphorylation of Ser-536 in the cytoplasm is independent of the PI3K-Akt-induced signaling pathway. This is consistent with the observation that the PI3K inhibitor does not affect the nuclear translocation or the DNA-binding activity of NF-B (40). In contrast, the reconstitution of Ser-to-Ala substituted p65 mutants in p65 Ϫ/Ϫ MEFs revealed that Ser-536 is not critical for TNF-␣-or IL-1-induced IL-6 expression and protection from TNF-␣-induced cell death (33). Thus, it is necessary to further investigate distinct role of the Ser-536 phosphorylation in the differential regulation of stimulus-or gene promoterdependent transcriptional activation of NF-B.
The catalytic subunit of human telomerase (hTERT) is shown to be present in the cytoplasmic fraction and directly associates with p65 in human multiple myeloma (61). Upon TNF-␣ stimulation, the constitutive association is transiently up-regulated and both proteins translocate into the nucleus. The specific IKK inhibitor PS-1145 blocks the TNF-␣-induced nuclear translocation of p65 and hTERT, in which Ser-536 phosphorylation of p65 is also inhibited (61). In addition, upregulation of the association with hTERT is dependent on the phosphorylation of p65 on Ser-536 (61). These observations raise the possibility that p65 is a shuttling protein for hTERT and the rapid dephosphorylation of p65 in the nucleus serves as a signal triggering the release of hTERT from p65. Furthermore, it has been shown that p65 plays roles in the nuclear translocation of tumor suppressor proteins p53 (62) and menin (63), suggesting a similar role for Ser-536 phosphorylation in the nuclear shuttling of tumor-related proteins.
In summary, we demonstrated the regulation of phosphorylation of p65 on Ser-536. Future characterization is necessary to understand physiological functions of the phosphorylation on Ser-536. The microarray analysis of p65 Ϫ/Ϫ MEFs reconstructed with wild type or S536A-mutated p65 is needed to explore the role of Ser-536 phosphorylation in the p65-dependent transcription from intact promoter structures of the chromosomal DNA. Because p65 Ϫ/Ϫ MEFs reconstructed with the S536A mutant are resistant to TNF-␣-induced apoptosis (33), p65-S536A knock-in mice may be resistant to the liver apoptosis observed in p65-deficient mice, therefore, the development and characterization of the knock-in mice will provide more insight into the physiological roles of the phosphorylation.