Inducible Phosphorylation of NF-κB p65 at Serine 468 by T Cell Costimulation Is Mediated by IKKϵ*

Here we identify IKKϵ as a novel NF-κB p65 kinase that mediates inducible phosphorylation of Ser468 and Ser536 in response to T cell costimulation. In addition, the kinase activity of IKKϵ contributes to the control of p65 nuclear uptake. Serines 468 and 536 are evolutionarily conserved, and the surrounding amino acids display sequence homology. Down-regulation of IKKϵ levels by small interfering RNA does not affect inducible phosphorylation of Ser536 but largely prevents Ser468 phosphorylation induced by T cell costimulation. Ser536-phosphorylated p65 is found predominantly in the cytosol. In contrast, the Ser468 phosphorylated form of this transcription factor occurs mainly in the nucleus, suggesting a function for transactivation. Reconstitution of p65–/– cells with either wild type p65 or point-mutated p65 variants showed that inducible phosphorylation of Ser468 serves to enhance p65-dependent transactivation. These results also provide a mechanistic link that helps to explain the relevance of IKKϵ for the expression of a subset of NF-κB target genes without affecting cytosolic IκBα degradation.

The NF-B transcription factor system serves to control the expression of an extraordinarily wide array of genes in response to infections, inflammation, and other harmful situations (1,2). NF-B target genes (such as immunoreceptors, cytokines, and chemokines) contribute to the innate immune response but also serve to control cell survival and proliferation (3). NF-B is a collective name for homo-or heterodimers composed of five different DNA-binding subunits, with the most frequently detected form being a heterodimer of p50 and p65 (RelA). The p65 subunit contains two strong, acidic transactivation domains called TAD1 3 and TAD2 in its C-terminal portion (4). A large variety of different inducers leads to NF-B activation by activation of numerous cellular and membrane receptors, including toll-like receptors and the T cell receptor. Thus far, three major pathways mediating NF-B activation have been identified, the so-called canonical and noncanonical pathways and the DNA damage-induced NF-B pathway. All NF-B activating events have in common that they lead to the proteasome-dependent generation of DNA-binding dimers (5). NF-B signals activating the canonical pathway funnel into the IKK complex, which is composed of the enzymatically active subunits IKK␣ and IKK␤ and the regulatory subunits IKK␥/NEMO (6,7) and ELKS (8). IKK␤-mediated I␤ phosphorylation allows subsequent ubiquitination and proteolytic destruction of this inhibitory protein. This leads to an unmasking of the p65 nuclear localization sequence and results in NF-B nuclear immigration, DNA binding, and gene expression.
Once activated, inducible post-translational modifications, including phosphorylation, acetylation, ubiquitination, or prolyl isomerization, allow the regulation of NF-B transcriptional activity (9,10). Thus far, eight different phosphorylation sites have been mapped for the strongly activating NF-B p65 subunit. Three sites are contained in the N-terminal Rel homology domain, whereas five sites (Ser 468 , Thr 505 , Ser 529 , Ser 535 , and Ser 536 ) are contained within both C-terminal TADs. Inducible phosphorylation of Ser 276 and threonine 311 promotes the interaction of p65 with the coactivating acetylase CREB-binding protein/p300, thus leading to p65 acetylation and stimulating NF-B-driven transcription. Various experimental approaches revealed that phosphorylation of serines 529, 535, and 536 serve to stimulate NF-B-dependent transcription. Expression of a p65 protein mutated in Ser 529 in p65 Ϫ/Ϫ cells revealed only a minor role of Ser 529 in transactivation, since it only contributes to achieve the Tax-induced maximal transcriptional response (11). Phosphorylation of Ser 535 is mediated by the calmodulindependent protein kinase IV, which results in an increase of NF-B-dependent transcription, as revealed by a phosphomimetic mutation where Ser 535 was replaced by glutamic acid (12). Basal phosphorylation of Ser 468 , a recently discovered phosphorylation site within TAD2 (13), is exerted by GSK3␤ (14). The same site can also be phosphorylated by IKK␤ in response to TNF␣ or IL-1 stimulation (15). We have previously shown that phosphorylation of p65 Ser 468 can be induced by T cell costimulation (13), but the responsible kinase(s) is not yet known. Phosphorylation of p65 NF-B at Ser 536 couples p65 to TAFII31-mediated transcription and is mediated, dependent on the stimulus, by various kinases, including IKK␣/␤, RSK1, TBK1 (TANK-binding kinase-1)/ NAK (NF-B-activating kinase)/T2K (TRAF2-associated kinase), and IKK⑀ (also called IKKi) (16,17). TBK1 and IKK⑀ show sequence homology to IKK␣/␤ but are not components of the IKK complex (18,19). Both kinases have been recently mainly recognized for their ability to phosphorylate interferon regulatory factor proteins in response to viral infection (20,21). IKK⑀ overexpression promotes dimerization and nuclear translocation of interferon regulatory factor-3 but also enhances the DNA binding activity of C/EBP␦ (22). IKK⑀ is mainly regulated via inducible, NF-B-dependent expression but is prominently expressed in T cells (19,23). Its kinase activity is triggered in response to T cell costimulation or the phorbol ester phorbol-12-myristate-13-acetate (PMA) but not by the cytokines TNF␣ or IL-1 (19). Whereas NF-B regulates IKK⑀, the role of IKK⑀ for NF-B activation remains elusive, since IKK⑀ Ϫ/Ϫ cells show unchanged inducible IB␣ phosphorylation and DNA binding. On the other hand, lipopolysaccharide-induced expression of some late NF-B target genes, including COX-2, regulated on activation, normal T cell expressed and secreted (RANTES), and interferon-␥-inducible protein 10 (IP-10) is lost in IKK⑀-deficient cells (22), suggesting a role of IKK⑀ for the regulation of NF-B at a later step.
Here we reveal an additional function for IKK⑀ and show that it serves as a novel p65 kinase that mediates inducible phosphorylation at serines 468 and 536 in vitro and in vivo, thus stimulating gene expression.

MATERIALS AND METHODS
Cell Culture, Transfections, and Stimulations-Mouse embryonic fibroblasts (MEFs) lacking p65 (24) and human 293T and HeLa cells were cultivated in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal calf serum, 2 mM L-glutamine, and 1% (v/v) penicillin/streptomycin. Adherent cells were transfected using Rotifect (Roth) according to the manufacturer's instructions. Jurkat T leukemia cells and Jurkat Tet-on cells were grown in supplemented RPMI 1640 medium. Jurkat cells (1.5 ϫ 10 7 ) were transfected by electroporation using a gene pulser (Bio-Rad) at 250 V/950 microfarads with constant amounts of DNA. Stable Jurkat cell lines were generated by cotransfection of IKK⑀-encoding plasmids or pSUPER-IKK⑀ and a plasmid carrying a selection marker (pEF-Puro). After further growth in the presence of puromycin (1 g/ml), either pools of stably selected cells were used for further experiments (small interfering RNA), or single cells were isolated by limiting dilution (Jurkat Tet-on cells). Costimulation of Jurkat cells was performed in a final volume of 500 l (2 ϫ 10 7 cells) by adding PMA (40 ng/ml) together with ionomycin (1 M). Proteasome activity was inhibited upon preincubation of cells for 1 h with 50 M MG132.
Cell Extracts, Phosphatase Treatment, and Western Blotting-After washing with ice-cold phosphate-buffered saline, cells were collected by centrifugation. The pellet was resuspended in Nonidet P-40 lysis buffer (20 mM Tris/HCl, pH 7.5, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 10 mM NaF, 0.5 mM sodium vanadate, leupeptin (10 g/ml), aprotinin (10 g/ml), 1% (v/v) Nonidet P-40) and incubated on ice for 20 min. Cell debris was removed by centrifugation at 13,000 rpm at 4°C for 10 min. Phosphatase treatment was performed by incubating the cell extract lacking phosphatase inhibitors with 400 units of -phosphatase for 1 h at 30°C according to the instructions of the manufacturer (Biolabs). Equal amounts of protein contained in the supernatant were further analyzed by reducing SDS-PAGE and Western blotting onto a polyvinylidene difluoride membrane (Millipore Corp., Bedford, MA). After blocking, membranes were incubated with the appropriate primary and horseradish peroxidase-coupled secondary antibodies, followed by protein detection using the Amersham Biosciences enhanced chemiluminescence system.
Purification of GST-p65 and Kinase Assays-The GST-p65-(354 -551) fusion protein was expressed in Escherichia coli BL21 cells and purified by affinity chromatography on glutathione-Sepharose 4B according to standard protocols. The immune complex kinase assays were done by immunoprecipitation of FLAG-tagged IKK⑀ using ␣-FLAG antibodies or by immunoprecipitation of endogenous IKK⑀. The precipitate was washed three times in Nonidet P-40 lysis buffer and two times in kinase buffer (20 mM Hepes/KOH, pH 7.4, 25 mM ␤-glycerophosphate, 2 mM dithiothreitol, 20 mM MgCl 2 ). The kinase assay was performed in a final volume of 20 l of kinase buffer containing 40 M ATP and 2 g of the purified GST-p65 substrate protein. After incubation for 20 min at 30°C, the reaction was stopped, separated by SDS-PAGE, and analyzed by immunoblotting with phosphospecific antibodies.
Co-immunoprecipitation Experiments-Human 293 T cells were transiently transfected to express the hemagglutinin-tagged p65 variants. 36 h later, cells were lysed in Nonidet P-40 lysis buffer, and one aliquot was used to confirm correct expression of the proteins. Equal amounts of protein contained in the remaining supernatants were immunoprecipitated either with ␣-hemagglutinin or with control antibodies and 25 l of protein A/G-Sepharose and rotated for 4 h on a spinning wheel at 4°C. The immunoprecipitates were washed five times in Nonidet P-40 buffer and then boiled in 1ϫ SDS sample buffer prior to SDS-PAGE and further analysis by Western blotting.
Subcellular Fractionation-Nuclear and cytosolic proteins were separated upon resuspending pelleted cells in 400 l of cold buffer A (10 mM Hepes/KOH, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride) by gentle pipetting. After incubation for 20 min on ice, 10 ml of 10% Nonidet P-40 was added, and cells were lysed by vortexing. The homogenate was centrifuged for 30 s in a microcentrifuge. The supernatant representing the cytosolic fraction was collected, and the pellet containing the cell nuclei was dissolved in 100 l of buffer C (20 mM Hepes/KOH, pH 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride). The Eppendorf tubes were incubated for 15 min on ice and centrifuged for 10 min with 13,000 rpm at 4°C. The supernatant representing the nuclear fraction was collected.

IKK⑀ Directly
Phosphorylates NF-B p65 at Ser 468 -Basal phosphorylation at Ser 468 is mediated by GSK3␤, whereas TNF␣-and IL-1induced phosphorylation of this site is maximal already 7.5 min after stimulation and is mediated by IKK␤ (15). We have previously shown that Ser 468 is inducibly phosphorylated by T cell costimulation (13), but the kinase mediating inducible phosphorylation in response to this stimulus is not known. Ser 468 is contained in a sequence motif called TAD1Ј that shows homology to TAD1 (4). The relative positions of Ser 536 and Ser 468 within the sequence motif are conserved (Fig. 1A), raising the possibility that both sites may employ the same kinase. From all of the kinases known to mediate Ser 536 phosphorylation, only IKK␣/ IKK␤ and IKK⑀ are known to be induced by T cell costimulation or by PMA and ionomycin, which mimic T cell costimulation upon protein kinase C activation and calcium release, respectively (32). To compare the relative roles of IKK␣, IKK␤, and IKK⑀ for p65 phosphorylation, Jurkat T cells were transfected with expression vectors encoding these three IKKs together with very low amounts of a vector directing expression of a GFP-tagged p65 protein, allowing expression of this fusion protein at physiological levels. This experimental approach was taken, because the low transfection efficiency of Jurkat T leukemia cells hampers the analysis of the endogenous p65 protein, thus enabling the detection of the slower migrating GFP-p65 fusion protein, which is fully regulated and functional (25). Whereas expression of IKK␣ only slightly induced p65 Ser 536 phosphorylation, IKK␤ strongly triggered Ser 536 phosphorylation of GFP-p65 and also of the endogenous p65 protein (Fig. 1B). IKK␤ also caused a slight induction of Ser 468 phosphorylation.
In contrast, even faint amounts of IKK⑀ potently stimulated p65 phosphorylation at serines 536 and 468 and even caused the appearance of a slower migrating p65 form (Fig. 1B). In contrast to IKK␤, the overexpression of IKK⑀ failed to cause the phosphorylation of the endogenous p65, which is in complex with IB proteins. A similar experimental approach revealed that this strong IKK⑀-induced phosphorylation could not be further enhanced by PMA/ionomycin stimulation (Fig.  1C). Is the slower migrating form of p65 due to IKK⑀-mediated phosphorylation, or is it also caused by other modifications? To address this question, extracts from cells coexpressing GFP-p65 and IKK⑀ were incubated with -phosphatase. This treatment completely converted the slower migrating form of p65 into the faster migrating version (Fig.  1D), indicating that the upper band represents a phosphorylated form of p65. Of note, phosphorylated Ser 536 occurred in the lower and in the upper band, whereas modified Ser 468 was found only in the upshifted p65 form. These results raise the possibility that both phosphorylations depend on each other. To test this hypothesis experimentally, IKK⑀mediated phosphorylation of each individual site was determined with a substrate GFP-p65 protein, where the respective other phosphorylation site was mutated to alanine. These experiments revealed full Ser 468 phosphorylation in the presence of a mutated Ser 536 and vice versa ( Fig.  2A), showing that both sites can be phosphorylated independently from each other. Mutation of the phosphorylatable serines to alanine precluded binding of the phosphospecific antibodies, thus confirming their specificity. Next, we wanted to test whether IKK⑀ phosphorylates p65 directly or causes phosphorylation upon activation of a downstream kinase. Cells were transfected to express epitope-tagged wild type or kinase-inactive forms of IKK⑀ or the control protein IKK␤, followed by purification via immunoprecipitation and in vitro kinase assays. The GST-p65-(354 -551) protein was efficiently phosphorylated at Ser 468 and Ser 536 by IKK⑀ wild type but not by the kinase-inactive IKK⑀ point mutant, as revealed by immunoblotting with phosphospecific antibodies (Fig. 2B). These in vitro experiments also revealed that IKK⑀ can cause the induction of a slower migrating p65 form. We also found a direct phosphorylation of p65 at both sites when IKK␤ was used as a kinase source. In summary, these results identify IKK⑀ as a kinase directly mediating p65 phosphorylation at Ser 468 and Ser 536 .
NF-B p65 Phosphorylated at Ser 468 Is Found Predominantly in the Nucleus-Following T cell costimulation, p65 phosphorylated at Ser 536 is predominantly found in the cytosol (33). We thus asked whether the same holds true for p65 phosphorylated at Ser 468 and stimulated Jurkat cells for various time periods with PMA/ionomycin, followed by subcellular fractionation into cytosolic and nuclear extracts (Fig. 3A). The Ser 468 -phosphorylated p65 protein was found predominantly in the nucleus, which is in contrast to the Ser 536 -phosphorylated p65 occurring mainly in the cytosol. Also, the kinetics revealed differences, since Ser 536 phosphorylation started to vanish already 30 min after stimulation, whereas Ser 468 phosphorylation displayed a delayed kinetics and was unchanged even 45 min after PMA/ionomycin treatment.
To test whether the same intracellular distribution occurs when p65 phosphorylation is triggered by IKK⑀, Jurkat cells were transfected to express GFP-p65 in the absence or presence of cotransfected IKK⑀. Two days after transfection, nuclear and cytosolic extracts were prepared and p65 phosphorylation was analyzed by immunoblotting (Fig. 3B). Also in this setting, Ser 468 -phosphorylated p65 was found predominantly (but not exclusively) in the nucleus. Overexpressed IKK⑀ was found in the nucleus and in the cytosol, which reflects the distribution of the endogenous kinase that is also found in both fractions (Fig. 3A).
To address the question of whether IKK⑀-mediated phosphorylation of p65 Ser 468 can also occur in the nucleus, we tested the effects of IKK⑀ on phosphorylation of the Gal4-p65 protein, which is constitutively nuclear (26). IKK⑀-triggered Ser 468 phosphorylation was further augmented by treatment with PMA/ionomycin (Fig. 3C). These results show that IKK⑀ phosphorylates nuclear p65 that is not in complex with IB but do not exclude the possibility that IKK⑀-mediated p65 phosphorylation can also happen in the cytoplasm. In contrast, coexpression of a kinase-inactive IKK⑀ point mutant completely inhibited this inducible phosphorylation, pointing to the relevance of IKK⑀ for this pathway.
The Kinase Activity of IKK⑀ Is Activated in Response to T Cell Costimulation-To address the question of whether the kinetics of IKK⑀ activation parallels that of p65 Ser 468 phosphorylation, Jurkat cells either containing or lacking the IKK␥/NEMO protein were stimulated with PMA/ionomycin for various periods. Immune complex kinase assays using the immunoprecipitated endogenous IKK⑀ protein as a kinase source revealed phosphorylation of the p65 substrate protein 20 and 45 min after stimulation (Fig. 4A). These data show that induction of IKK⑀ kinase activity parallels p65 Ser 468 phosphorylation and that IKK␥/NEMO, which is essential for the function of the IKK complex, is not important for primary IKK⑀ activation at the time points analyzed. The amount of IKK⑀ protein is unchanged in response to T cell costimulation and not influenced by the presence of IKK␥/NEMO (Fig. 4B).
Phosphorylation of p65 at Ser 468 Requires Its Release from IB␣-We next asked whether free or IB-bound p65 is phosphorylated at Ser 468 . To address this question, Jurkat cells lacking the IKK␥/NEMO protein and thus being unable to phosphorylate IB␣ and to release the p65 protein from the cytosol (6) were stimulated for various periods with PMA/ionomycin. IKK␥/NEMO-deficient cells failed to induce p65 phosphorylation at Ser 468 and Ser 536 , whereas IKK␥/NEMO-retransfected control cells showed full phosphorylation at both sites (Fig. 5A), suggesting that only free and untrapped p65 can be phosphorylated.
To substantiate this finding by an independent experimental approach, Ser 468 phosphorylation was tested in cells where IB␣ phosphorylation was blocked by the IKK␤ inhibitor AS602868 (34). Also, this experimental approach confirmed that the prevention of IKK activation precluded nuclear translocation of p65 and its phosphorylation at Ser 468 (Fig. 5B). Similarly, IKK⑀-mediated Ser 468 phosphorylation of GFP-p65 was strongly diminished upon coexpression of IB␣ (Fig. 5C). This experiment also showed strongly diminished IB␣ levels in the presence of overexpressed IKK⑀, which is presumably due to the function of IKK⑀ as an IB␣ Ser 36 kinase (19).
Given the ability of IKK⑀ to trigger IB␣ decay, we tested whether IKK⑀ expression can cause p65 Ser 468 phosphorylation by compensating the inactivity of the classical IKK complex in IKK␥/NEMO-deficient Jurkat cells. Expression of IKK⑀ caused Ser 468 and Ser 536 phosphorylation of coexpressed GFP-p65 in the absence and presence of IKK␥/ NEMO (Fig. 5D). This suggests that this kinase can liberate and phosphorylate the p65 protein independent from the IKK complex, a process that may be facilitated by the possibility that not all of the expressed GFP-p65 protein is bound by IBs.
In summary, all of these experiments show that IKK⑀-mediated Ser 468 phosphorylation does not occur when the release of IB-bound p65 is inhibited. Accordingly, stabilization of the IB⅐NF-B complex by the proteasome inhibitor MG132 enhanced PMA/ionomycin-induced and IKK␤-mediated phosphorylation of IB and p65 Ser 536 within this complex (Fig. 5E). In contrast, T cell costimulation-induced p65 phospho-FIGURE 2. IKK⑀ directly phosphorylates p65 in vitro and induces the occurrence of a slower migrating p65 form. A, Jurkat cells were transfected to express GFP-p65 or various GFP-p65 mutants, where Ser 468 and/or Ser 536 were changed to alanine together with FLAG-tagged IKK⑀ at the indicated combinations. Phosphorylation of p65 and expression of IKK⑀ was analyzed by immunoblotting as shown. B, Jurkat cells were transfected with expression vectors for epitope-tagged forms of IKK␤, IKK⑀, and kinase-inactive point mutants of these kinases, respectively. After 36 h, cells were lysed, and the IKK proteins were immunoprecipitated (IP) and purified from cell lysates with polyclonal FLAG or ␣-Myc antibodies, respectively. Subsequently, kinase activity was monitored by immune complex kinase assays (KA) using recombinant GST-p65-(354 -551) as substrate. Phosphorylation of p65 was determined by immunoblotting using phosphospecific antibodies as shown. Note that IKK⑀ causes the upshift also in these in vitro assays. In the control immunoblot, the ␣-FLAG antibody was detected first, followed by visualization of IKK␤ with ␣Myc antibodies. rylation at Ser 468 was diminished in the presence of MG132, corroborating the concept that this phosphorylation does not take place within the IB⅐NF-B complex. Accordingly, co-immunoprecipitation experiments showed no influence of the p65 Ser 468 phosphorylation status on its ability to associate with IB␣ (supplementary Fig. 1).
IKK⑀ Serves to Modulate p65 Nuclear Import-To investigate the effects of IKK⑀ on p65, we generated stable Jurkat cell lines allowing doxycycline-dependent expression of the wild type or kinase-inactive IKK⑀. To measure the impact of IKK⑀ on nuclear import of NF-B, cells were induced with doxycycline to trigger IKK⑀ expression and stimulated for various periods with PMA/ionomycin. Nuclear and cytosolic extracts were tested for the occurrence of NF-B by immunoblotting. Control cells and cells induced to express IKK⑀ behaved similarly and showed p65 nuclear immigration starting from 7.5 min and reaching its maximum 30 min post PMA/ionomycin treatment (Fig. 6). In contrast, cells expressing IKK⑀ kinase-active showed a strongly impaired accumulation of nuclear p65 (Fig. 6), suggesting that IKK⑀ contributes to the control of p65 nuclear import. Decreased nuclear import was not only seen for p65 but also for its dimerization partner p50.
T Cell Costimulation-induced p65 Ser 468 Phosphorylation Depends on IKK⑀-The relative contribution of IKK⑀ for p65 phosphorylation in response to T cell costimulation was investigated by testing the effect of small interfering RNAs specific for IKK⑀. Jurkat cells were transfected either with the control vector pSUPER or with pSUPER-IKK⑀, a vector that directs the synthesis of small interfering RNAs specific for IKK⑀ together with a plasmid conferring resistance to puromycin. After selection of stably transfected Jurkat cell pools, p65 phosphorylation was investigated in untreated and PMA/ionomycin-stimulated cells. Whereas Ser 536 phosphorylation was not significantly affected by impaired IKK⑀ levels, phosphorylation at Ser 468 was significantly reduced (Fig. 7A), thus showing an important contribution of IKK⑀ for the phosphorylation at this site. Since GSK3␤ mediates basal p65 Ser 468 phosphorylation in unstimulated cells (14) and AS602868 displays some inhibitory effects on GSK3␤, a possible contribution of this kinase for  PMA/ionomycin-induced phosphorylation was tested. Jurkat cells were left untreated or stimulated in the absence or presence of the GSK3␤ inhibitor LiCl (35). Detection of Ser 468 phosphorylation by immunoblotting revealed no significant changes in the presence of LiCl (Fig. 7B), thus excluding a role of GSK3␤ for Ser 468 phosphorylation induced by T cell costimulation.
Phosphorylation at Ser 468 and Ser 536 Enhances the Transcriptional Activity of p65-The impact of IKK⑀-mediated Ser 468 and Ser 536 phosphorylation on gene expression was tested in transient reporter gene assays. p65 Ϫ/Ϫ MEFs were cotransfected with a NF-B-dependent luciferase gene and increasing amounts of either wild type p65 or variants thereof that are point-mutated in the phosphoacceptor sites. The phosphomimetic p65 S468E mutant, where the serine was changed to glutamic acid, boosted NF-B-dependent gene expression more strongly than the wild type protein (Fig. 8A), indicating that phosphorylation of Ser 468 serves to enhance transcription. An increased transcriptional activity was also observed for p65 S536E and the double mutant, where both serines were changed to glutamic acids. Whereas mutation of Ser 536 to Ala did not significantly impair the transcriptional activity of the mutant protein, the Ser 468 to ala mutant showed a markedly decreased ability to trigger gene expression. Accordingly, a p65 mutant where both serines were changed to alanine showed a strongly reduced transcriptional activity. To test whether mutation of Ser 468 has an impact on DNA binding (e.g. by changes in the intramolecular conformation), cells were transfected with GFP-p65 or with a GFP-p65 mutant S468A mutant. Following TNF␣ stimulation to liberate p65 from IB and to induce p65 phosphorylation, DNA binding activity was measured by EMSAs (Fig. 8B). The DNA-bound form of GFP-tagged p65 can FIGURE 5. Induced p65 Ser 468 phosphorylation does not occur for cytoplasmically trapped p65. A, Jurkat cells lacking IKK␥/NEMO and control cells retransfected to express IKK␥/NEMO were stimulated for the indicated periods as shown. Cytoplasmic and nuclear extracts were analyzed by immunoblotting (IB) for the phosphorylation of p65 and the occurrence of control proteins as shown. B, Jurkat cells were preincubated for 60 min with AS602868 (1.2 g/ml) and stimulated for 20 min with PMA/ionomycin. Equal amounts of protein contained in nuclear extracts were analyzed by Western blotting for phosphorylation of p65 Ser 468 and the occurrence of p65 and HDAC-1. Note that the IKK␤ inhibitor prevents nuclear entry of p65. C, Jurkat cells were transiently transfected with expression vectors encoding GFP-p65 and FLAG-tagged IKK⑀ and IB␣ at the indicated combinations. Total cell extracts were analyzed for p65 phosphorylation and the occurrence of the indicated proteins. D, IKK␥/NEMO-deficient and control cells were transfected to express the indicated combinations of GFP-p65 and IKK⑀ as shown, followed by the analysis of p65 phosphorylation by Western blotting. E, Jurkat T cells were left untreated or stimulated for 20 min with PMA/ionomycin in the absence or presence of MG132. Total cell extracts were analyzed by immunoblotting for p65 phosphorylation at Ser 468 or Ser 536 and also for the occurrence or phosphorylation and degradation of IB␣ as shown.
be conveniently distinguished from endogenous NF-B, since it migrates more slowly. Both p65 forms displayed similar DNA binding, suggesting that the changed activity of p65 mutated in Ser 468 is attributable to the transactivation potential rather than to altered DNA binding.

DISCUSSION
IKK⑀ is important for the lipopolysaccharide-induced expression of a specific subset of NF-B target genes, including TNF␣, IL-1, IP-10, and RANTES, but not the early activated gene IB␣. On the other hand, IKK⑀ Ϫ/Ϫ cells show normal lipopolysaccharide-induced phosphorylation and degradation of IB␣ and no changes in NF-B DNA binding (22). These data point to the importance of IKK⑀ for NF-B-mediated transactivation, and accordingly an earlier study provided evidence that IKK⑀ contributes to DNA binding activity of C/EBP␦, thus affecting the expression of target genes, which often depend on the coordinate binding of NF-B and C/EBP (22). However, since CCAAT/enhancer-binding protein ␦ DNA binding is only partially compromised but target gene transcription is completely abolished in IKK⑀ Ϫ/Ϫ cells, it can be assumed that further mechanisms contribute to the stimulatory effect of IKK⑀ on NF-B-dependent transcription. Here we provide evidence that IKK⑀ serves to mediate p65 phosphorylation at Ser 468 and Ser 536 , as revealed by in vitro and in vivo experiments. Whereas the p65 protein can enhance transcription of the IKK⑀ gene (36), this study shows that IKK⑀ in turn can control the phosphorylation of p65, thus establishing an autoregulatory loop.
Also, the IKK⑀ homologue TBK1 was identified as a p65 Ser 536 kinase that exerts its function together with the NAP1 (for NAK-associated protein 1) adapter protein (37). Accordingly, recombinant IKK⑀ and TBK1 enzymes are enzymatically similar to each other, and both require phosphorylation of a critical Ser in their activation loops for kinase activity (38). Deletion of the TBK1 gene leads to embryonic lethality at approximately embryonic day 15 from massive hepatocyte apoptosis (39), which resembles the phenotype of p65 Ϫ/Ϫ mice (40). Similar to IKK⑀, TBK1 Ϫ/Ϫ MEFs show normal IB degradation and DNA binding but defects in the transactivation of specific NF-B target genes (39). Of note, these transcriptional effects are not seen when NF-B-dependent transcription is measured with synthetic reporter genes (20), raising the possibility that the effects of IKK⑀ and TBK1 are specific for individual genes. It will be interesting to learn in future studies whether these promoter specific effects are due to p65 phosphorylation or to other events such as NF-B subunit exchange or chromatin remodeling (41). Thus, we suggest that IKK⑀ and TBK1 not only serve to mediate interferon-regulatory factor-3 phosphorylation following engagement of Toll-like receptors 3 and 4 or viral infection (20,21) but also fulfill a function for NF-B activation upon phosphorylation of p65 and controlling its nuclear import. The inhibitory effect of kinase-inactive IKK⑀ on p65 nuclear import is not linked to p65 Ser 468 phosphorylation, since this site is not relevant for its subcellular distribution (data not shown) (15). Therefore, it is plausible that these effects are due to phosphorylation of p65 at additional sites or by modification of further substrate proteins.
IKK⑀ fails to induce an upshift for p65 that is mutated in serines 536 and 468, but since p65 phosphorylation must not necessarily result in the occurrence of an upshifted form (9), we cannot exclude the possibility that IKK⑀ is also modifying p65 at additional sites. Here we show that IKK⑀ mediates inducible p65 Ser 468 phosphorylation in response to T cell costimulation, thus identifying the third kinase for this site. Whereas GSK3␤ mediates basal phosphorylation in unstimulated cells (14), IKK␤ was shown to mediate Ser 468 phosphorylation in response to TNF␣ or IL-1 (15). However, these cytokines trigger Ser 468 phosphorylation with a very fast kinetics, and this p65 modification cannot be detected even 15 min after stimulation. Since the kinase activity of IKK⑀ cannot be augmented by these two cytokines (19), it can be assumed that FIGURE 6. IKK⑀ contributes to the control of NF-B nuclear uptake. Jurkat Tet-on cells stably transfected either with empty expression vector, IKK⑀, or kinase-inactive IKK⑀ were treated for 36 h with doxocycline to induce IKK⑀ expression, followed by PMA/ionomycin stimulation for the indicated periods. Nuclear and cytosolic extracts were prepared and analyzed for the occurrence of p50 and p65. Immunoblotting (IB) for the control proteins p105 and HDAC-1 was performed to ensure the purity of the fractions. IKK⑀ will not play a relevant role for these stimulatory pathways. In costimulated T cells, Ser 468 phosphorylation occurs with a significantly delayed kinetics, and a contribution of IKK␤ for this process is unlikely, since the activity of this kinase already drops at these late time points (33), and various experimental approaches including small interfering RNA and a dominant negative IKK⑀ mutant revealed the importance of this kinase for Ser 468 phosphorylation triggered by costimulation. As several phosphorylation sites within p65 are modified by various kinases (9,10), it remains to be seen whether other stimuli affecting Ser 468 phosphorylation employ further kinases. The concept that various stimuli employ distinct kinases for a given p65 phosphorylation site also holds true for the Ser 536 phosphorylation site. In this case, Ser 536 phosphorylation triggered by TNF␣ or T-cell costimulation is mainly mediated by IKK␤ (27,33), whereas DNA damage employs RSK1 to modify the same site (42). The stimulus preferentially employing IKK⑀ for phosphorylation of Ser 536 remains to be identified in future studies.
Both modified serines are evolutionarily conserved between mammals and amphibians (Fig. 9), arguing for their functional importance. The role of Ser 536 phosphorylation does not yield a coherent picture, since expression of a p65 Ser 536 3 Ala mutant in a p65 Ϫ/Ϫ background failed to reveal a role of Ser 536 for TNF-mediated IL-6 gene induction (24), but on the other hand, this phosphorylation was required for TNFor lipopolysaccharide-induced activation of an NF-B reporter gene (11,43). Ser 536 phosphorylation has been linked to further effects, including association with TAFII31 (16), the promotion of stimulus-induced p65 turnover (17), and the control of p65 nuclear uptake kinetics (33). Along this line, a recent study showed that oscillatory phosphorylations of p65 Ser 536 appear to be a consequence of its shuttling between the cytoplasm and the nucleus (44). It will be interesting in future studies to investigate a potential contribution of IKK⑀ on these oscillatory p65 phosphorylations.
Our results indicate that p65 Ser 468 phosphorylation, which occurs within the short TAD1Ј sequence motif shared with TAD1 (4), primarily affects p65-dependent transactivation. The TAD1Ј sequence motif was shown to activate transcription when fused as a multimer to the DNA binding domain of Gal4 (45), supporting a primary role in transactivation. The TAD1 region can form an inducible ␣-helix upon binding to other proteins (45), and accordingly secondary structure predictions suggest that also the homologous TAD1Ј region containing Ser 468 could fold in such an ␣-helix (46). In such a conformation, phosphorylation of Ser 468 would increase the negative charge on one helix surface. This phosphorylation could then affect binding to corepressors or coactivators, such as CREB-binding protein/p300, which contact p65 in the adjacent region between amino acids 477 and 503 (47). In support of a possible role in the regulation of gene expression, Ser 468 -phosphorylated p65 is preferentially found in the nucleus. Phosphorylation of Ser 468 induced by T cell costimulation does not occur alone and is accompanied by further p65 modifications. We suggest that distinct p65 modification patterns control the subset of induced target genes as well as the duration and amplitude of the NF-B response.