IκB Kinases Serve as a Target of CD28 Signaling*

Optimal T cell activation and interleukin-2 production requires a second signal in addition to antigen-mediated T cell receptor (TCR) signaling. The CD28 molecule has been demonstrated to act as an effective costimulatory molecule upon binding by B7.1 or B7.2 present on antigen-presenting cells. The CD28 signal acts in concert with the TCR signal to significantly augment activation of the NF-κB family of transcription factors. The interleukin-2 gene is regulated by NF-κB among other transcription factors, in part, via a CD28 responsive element (CD28RE) present in the IL-2 promoter. Enhanced activation of NF-κB by CD28 is mediated by rapid phosphorylation and proteasome-mediated degradation of the NF-κB inhibitory proteins IκBα and IκBβ, which allows for accelerated nuclear expression of the liberated NF-κB. Herein, we provide evidence that the catalytic activities of two recently identified IκB kinases, IKKα and IKKβ, are significantly elevated when T cells are stimulated through CD28 in addition to mitogen treatment. Catalytically inactive forms of IKKs are able to block the in vivo phosphorylation of IκBα induced by mitogen and CD28. Furthermore, CD28-mediated reporter gene transactivation of the CD28RE/AP-1 composite element is consistently attenuated by the IKK mutants. These findings suggest that cellular signaling pathways initiated at the TCR and CD28 converge at or upstream of IKK, resulting in more robust kinase activity and enhanced and prolonged NF-κB activation.

T cell activation and IL-2 1 production is critically dependent on the transmission of signals derived from the cell surface to the nucleus in order to modulate changes in gene expression (1). It is now well established that activation and signaling through the T cell receptor (TCR) alone is not sufficient for IL-2 production or proliferation (2). Antigen-presenting cells achieve maximal activation of the antigen-reactive T cells by binding accessory molecules present on the T cell surface in addition to antigen presentation to the TCR via the context of major histocompatibility complex class II molecules. One of the most intensively studied accessory molecules, CD28, binds to the B7.1 and B7.2 molecules present on the surface of macrophages and dendritic cells (3). CD28 has been demonstrated to act as a costimulatory signal for T cells since IL-2 production and proliferation are enhanced when CD28 is engaged in addition to the TCR (4). CD28 also confers post-transcriptional mechanisms to enhance T cell activation by prolonging the half-life of IL-2 mRNA (5). However, engagement of CD28 alone has no measurable effect on T cell activation. The IL-2 gene promoter contains an enhancer known as the CD28 responsive element (CD28RE) which functions as an integrator of transcription factors activated through the TCR and CD28 and is essential for IL-2 transcription mediated through CD28 (6). The CD28RE enhancer also forms a composite element with a juxtaposed AP-1 binding site, and it has been shown that this composite element mediates CD28 responsiveness (7,8). Studies by several laboratories suggest that members of the NF-B, AP-1, and ATF-CREB transcription factor families bind to the CD28RE/AP-1 composite element (7,9).
The NF-B/Rel family of transcription factors is composed of a set of structurally related, evolutionarily conserved DNAbinding proteins consisting of p50, p52, p65, c-Rel, and RelB (10). The NF-B complexes are sequestered in the cytosolic compartment as latent complexes by members of the IB family, all of which have characteristic ankyrin repeat domains required for interactions with NF-B proteins (reviewed in Refs. 10 and 11). The two major IB proteins, IB␣ and IB␤, both have two regulatory N-terminal serine residues that are phosphorylated in response to a wide array of signals (12)(13)(14). The phosphorylated IBs are then ubiquitinated and targeted to the proteasome for proteolytic degradation (15). Signals such as TNF-␣ or mitogens such as PMA, which selectively induce the degradation of only IB␣, are associated with the transient activation of NF-B since the IB␣ gene is positively regulated by NF-B factors (16 -19). However, signals such as lipopolysaccharide, ␣CD3 ϩ ␣CD28, IL-1, or the Tax protein of type I human T-cell leukemia virus-I induce a persistent NF-B activation, which appears to be due to both prolonged IB␣ degradation (20,21) and degradation of IB␤ (22)(23)(24). The IB␤ gene is presumably not under the control of NF-B since the protein is not rapidly replenished as is seen with IB␣ (22).
Recently two serine kinases termed IKK␣ and IKK␤, which are part of a large multiprotein complex known as the IKK signalsome have been cloned and demonstrated to phosphorylate both IB␣ and IB␤ in response to cytokines and other signals known to activate NF-B (25)(26)(27)(28)(29). An upstream kinase, NIK, has also been identified and shown to stimulate NF-B in response to distinct stimuli such as TNF-␣ and IL-1 (30). The mechanism may be direct since NIK has been demonstrated to phosphorylate IKK␣ on Ser 176 (31).
Although the CD28 signaling pathway has been shown to accelerate TCR-induced nuclear expression of various NF-B/Rel transcription factors, the underlying molecular mechanism remains elusive. We and others have previously demonstrated that ligation of CD28 initiates a potent costimulatory signal leading to the rapid and persistent degradation of IB␤ and enhanced degradation of IB␣ (20,23). However, it is not known if CD28 is mediating enhanced IB kinase activity. We report here that CD28 potentiates the kinase activity of IKK␣ and IKK␤, which are only weakly activated by mitogen or TCR signals alone.

MATERIALS AND METHODS
Cell Culture and Reagents-Jurkat T cells (ATCC) and Jurkat cells expressing the SV40 large T antigen (Jurkat Tag) (32) were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, and antibiotics. Human peripheral blood T cells were prepared from lymphocyte-enriched human blood (Biological Specialty Corporation, Colmar, PA) with a Ficoll-Hypaque gradient (Amersham Pharmacia Biotech) followed by negative selection with human T cell enrichment immunocolumns (Biotex Laboratories Inc., Edmonton, Alberta, Canada). C305 (anti-clonotypic Jurkat TCR) was provided by Dr. Arthur Weiss (University of California, San Francisco) and used at a 1:1000 dilution. The monoclonal antibody for human CD28 (clone 9.3) was provided by Bristol-Myers Squibb Pharmaceutical Research Institute and used at a 1:10,000 dilution (0.3 g/ml). The antibody against the influenza hemagglutinin (HA) epitope tag (anti-HA) and protein Aagarose was obtained from Boehringer Mannheim. Anti-IKK␣ (H744) and anti-IKK␤ (H470) were purchased from Santa Cruz Biotechnology, Inc. Anti-IB␣ antiserum was provided by Dr. Warner Greene.
In Vitro Kinase Assays-In vitro kinase assays were done essentially as described previously (25). Briefly, cell lysates were incubated with specific antisera (IKK␣ or IKK␤) for 1 h, and then 20 l of protein A-agarose (Boehringer Mannheim) was added and incubated for an additional 3 h. The immunoprecipitates were washed three times with cell lysis buffer containing 1% Nonidet P-40, 20 mM Hepes, 250 mM NaCl, 20 mM ␤-glycerophosphate, 1 mM EDTA, 0.1 mM sodium vanadate, 1 mM dithiothreitol, 20 mM p-nitrophenyl phosphate, 1 mM phenylmethylsulfonyl fluoride, and 1:100 of a protease inhibitor mixture. The immunoprecipitates were then washed once with cell lysis buffer ϩ 8 M urea, and twice with kinase buffer (20 mM Hepes and 20 mM magnesium chloride). Kinase reactions were then performed at 30°C for 30 min in the presence of [␥-32 P]ATP and either GST-IB␣ 1-55 or GST-IB␤ 1-82. The reactions were terminated upon addition of 5ϫ sample buffer followed by SDS-polyacrylamide electrophoresis and autoradiography.
Immunoblotting-Jurkat cells, or transiently transfected Jurkat-Tag cells were stimulated with the indicated inducers and then collected by centrifugation at 800 ϫ g for 5 min. Whole cell and subcellular extracts were prepared as described previously (35,36). For immunoblotting analyses, whole cell extracts (ϳ15 g) were fractionated by reducing 8.75% SDS-polyacrylamide gel electrophoresis, electrophoretically transferred to nitrocellulose membranes, and then analyzed for immunoreactivity with the indicated primary antibodies using an enhanced chemiluminescence detection system (ECL; Amersham Pharmacia Biotech).
Luciferase Reporter Gene Assays-Jurkat cells were transiently transfected with DEAE-dextran with 2 g of the CD28RE/AP-1 reporter and either empty vector or the indicated amounts of IKK␣ K44M or IKK␤ K44A. Transfectants were split into two and either left untreated or treated with PMA (10 ng/ml) or PMA and anti-CD28 (1:10,000 dilution) for 8 h. The extracts were harvested with reporter lysis buffer (Promega) and then measured for luciferase activity as described previously (23).

The CD28 Signal Potentiates IKK␣ Activation in Both Jurkat and Primary Human T Cells-We and others have previously demonstrated that IB␣ degradation is enhanced by CD28
costimulation (20,23), although the underlying mechanism has remained unclear. We examined whether CD28 potentiates mitogen-mediated IKK␣ activation. We first performed in vitro kinase assays with Jurkat T cells utilizing GST-IB␣ 1-55 as a substrate. The kinase activity of IKK␣ was slightly induced by PMA treatment as described previously (26) (Fig. 1A, lane 2, upper middle panel). However, when cells were treated with CD28 antibody in addition to mitogen treatment, the kinase activity of IKK␣ was significantly elevated (lane 3), although CD28 alone had no effect on the kinase activity (lane 4). Autophosphorylation of IKK␣ was also strongly induced by PMA  3, upper panel). It should be noted that the time of each stimulation for this and subsequent experiments was 7 min, which represented maximal kinase activity as exerted by each stimulus. In a time course experiment, the kinase activity was highest at 7 min, was sustained for at least 15 min, and finally subsided by 30 min (data not shown). Pretreatment of cells with TPCK, a chymotrypsin-like protease inhibitor, known to block IB␣ phosphorylation by unknown mechanisms (37), abolished all kinase activity associated with PMA and anti-CD28 treatment (Fig. 1A, lane 5,  upper middle panel). To ensure that IKK␣ kinase activity was directed to the two N-terminal serine residues of IB␣ (serine 32 and 36), we used a GST-IB protein with alanine residues substituted for the two serines as substrate. Importantly, this substrate was not phosphorylated by IKK␣ in Jurkat cells treated with mitogen and anti-CD28 (lane 6). Immunoblotting of the immunoprecipitated IKK␣ revealed equal amounts of IKK␣ for all samples (Fig. 1A, lower middle panel). In addition to IKK␣ kinase activity, we examined the fate of endogenous IB␣ from the same extracts used for kinase assays. Immunoblots of IB␣ revealed that PMA only slightly induced IB␣ phosphorylation as assessed by a slower migrating band on SDS-polyacrylamide gel electrophoresis gels (Fig. 1A, lane 2, lower panel). When CD28 antibody was added in addition to PMA, a strong phosphorylated band was readily observed (lane 3), and as expected, CD28 alone did not induce IB␣ phosphorylation (lane 4). TPCK also inhibited the inducible phosphorylation of IB␣ (lane 5). Taken together, IKK␣ kinase activity is enhanced by PMA and CD28 treatment and is well correlated with in vivo IB␣ phosphorylation.
To confirm the physiological relevance and generality of these findings, we recapitulated our experiments in primary T cells. To this end, we purified T cells from human peripheral blood, treated with either PMA alone or PMA together with anti-CD28, and performed IKK␣ kinase assays with GST-IB␣ as a substrate. As observed in Jurkat cells, the untreated human primary T cells exhibited no detectable IKK kinase activity (Fig. 1B, lane 1). Treatment with PMA alone induced moderate IKK␣ kinase activity (lane 2), which was markedly enhanced in the presence of CD28 antibody (lane 3). Autophosphorylation of IKK␣ was also evident when the cells were treated with both PMA and anti-CD28 (lane 3) Therefore, CD28-mediated IKK kinase activation occurs in both Jurkat and primary human T cells. To confirm that these findings were not specific to the mitogen PMA but rather reflected signaling through the TCR in a more physiological manner, we also used C305 ascites (anti-clonotypic Jurkat TCR) (38). As seen with PMA, treatment of Jurkat cells with C305 resulted in a small degree of IKK␣ kinase activity (Fig. 1C, lane 2, upper  panel). This was significantly enhanced when CD28 was present (lane 3). The levels of IKK␣ protein were similar in all samples as detected by immunoblotting (Fig. 1C, lower panel). We conclude that CD28 mediates a costimulatory signal resulting in enhanced IKK␣ mediated IB␣ phosphorylation.
Given that IB␤ is also rapidly degraded when T cells receive a costimulatory signal (23), we next examined the effect of CD28 ligation on IKK␣-mediated phosphorylation of IB␤. For these purposes, we used GST-IB␤ 1-82, which contains the two IKK phosphorylation sites at serines 19 and 23. As expected, IKK␣ from untreated Jurkat cells displayed no kinase activity toward IB␤ (Fig. 2, lane 1). When treated with PMA, there was slight kinase activity toward IB␤ (lane 2). Importantly, when Jurkat cells were treated with both PMA and CD28, there was a significant up-regulation of kinase activity directed toward IB␤ (lane 3). Once again, CD28 alone had no effect on IKK␣ kinase activity (lane 4), emphasizing the re-quirement for two signals to achieve maximal IKK␣ kinase activity. Together, it appears that CD28 potentiates IKK␣mediated phosphorylation of both IB␣ and IB␤.
CD28 Also Mediates Enhanced IKK␤ Phosphorylation of IB␣ and IB␤-We next focused on the other cytokine-responsive IB kinase IKK␤ to determine if it had a similar activation response to CD28 costimulation. First, we examined the effects of endogenous IKK␤ on GST-IB␣. In untreated Jurkat cells, there was no IKK␤ activity toward IB␣ (Fig. 3A, lane 1). Upon treatment with PMA, there was slight IKK␤ activation (lane 2), which was subsequently elevated with CD28 treatment (lane 3). As expected, CD28 alone had no effect on IKK␤ kinase activity (lane 4). These results suggest that IKK␤ activation is reminiscent of IKK␣ as shown in Fig. 1. We also tested the kinase activity of IKK␤ toward IB␤. Similarly, PMA alone induces a degree of phosphorylation (Fig. 3B, lane 2), which is augmented by CD28 (lane 3). We conclude that IKK␤, in addition to IKK␣, is responsive to the CD28 costimulatory signal.
A Catalytically Inactive IKK␣ Mutant Blocks CD28-mediated in Vivo IB␣ Phosphorylation-CD28 synergizes with TCR signaling to accelerate IB␣ phosphorylation and degradation (20,23). To determine the role played by IKK␣ in the in vivo phosphorylation of IB␣ mediated by CD28 we utilized a catalytically inactive IKK␣ mutant. This mutant, which has been previously demonstrated to inhibit TNF-␣-induced RelA nuclear translocation (28), has a methionine substituted for a lysine at position 44, resulting in defective ATP binding. We transiently transfected Jurkat-Tag cells with an HA-tagged IB␣ construct, and split the transfection into two, leaving one sample untreated and treating the other sample with a combination of PMA and anti-CD28. As expected, we observed an unphosphorylated IB␣ by immunoblotting when the cells were not treated (Fig. 4, lane 2). When the cells were treated with PMA/CD28, two bands were readily detected, with the slower migrating band representing the phosphorylated form (lane 3). Interestingly, when the catalytically inactive form of IKK␣ was cotransfected with IB␣, the phosphorylation induced by PMA and CD28 was completely blocked (lanes 5 and  7). This result suggests that IKK␣ is required for CD28-mediated IB␣ phosphorylation.
Catalytically Inactive Forms of IKK␣ and IKK␤ Inhibit CD28-mediated CD28RE/AP-1 Transactivation-It has been previously reported that CD28 responsiveness within the IL-2 promoter is mediated by a CD28RE/AP-1 composite element (7,8). We transfected a luciferase reporter construct driven by this composite element into Jurkat cells and either treated with PMA alone or PMA together with anti-CD28. PMA alone was unable to substantially activate this reporter gene (Fig. 5A,  lane 2), but when CD28 was added in conjunction with PMA, reporter activity rose to approximately 12-fold above that observed with untreated cells (lane 3 versus lane 1). We next cotransfected the catalytically inactive IKK␣ with the CD28RE/AP-1 reporter and treated it with PMA and CD28. Interestingly, transfection of a moderate amount (150 ng) of IKK␣ K44M cDNA reduced the reporter activity by approximately 50% (Fig. 5B, lane 3). With a higher dose (300 ng) transfection of this IKK mutant, more pronounced inhibition was observed (lane 4). It should be noted that this catalytically inactive mutant was previously demonstrated to block TNF-␣induced IKK activation by about 2-or 3-fold (28), suggesting the role of other compensatory factors to account for the activity observed. The catalytically inactive IKK␤ K44A also inhibited reporter gene activity by about 50% at the first dose tested (100 ng) (Fig. 5C, lane 3). However, a higher dose of IKK␤ K44A (200 ng) acted as an even more potent inhibitor of CD28mediated reporter gene activation, reducing the activity to about 25% of the control (lane 4 versus lane 2). When both IKK␣ K44M and IKK␤ K44A were cotransfected, the inhibition observed was only slightly greater than the inhibition seen with a high dose of IKK␤ K44A alone (data not shown). The lack of complete inhibition of reporter gene activity by the dominant negative IKKs is likely due to the high sensitivity of this assay, but we cannot preclude the involvement of additional IB kinases in CD28-mediated NF-B activation. From these experiments we conclude that both IKK␣ and IKK␤ are required for CD28-mediated transactivation of a CD28RE/AP-1 composite element.

DISCUSSION
The NF-B transcription factors are activated by a diverse array of signals that target the inhibitory proteins IB␣ and IB␤ for phosphorylation and proteasome-mediated degradation (10). The T cell auxiliary molecule, CD28, which provides a costimulatory signal for T cell activation and IL-2 production, is a potent inducer of NF-B (23,39). Although CD28 promotes the rapid degradation of both IB␣ and IB␤, the mechanism has not been elucidated to date. Possibilities include synergy of IB kinase activity, activation of an IB kinase unique to the CD28 pathway, or perhaps enhanced ubiquitination and/or direct degradation. In this study, we report that CD28 specifically enhances the kinase activity of IKK␣ and IKK␤, two recently identified cytokine-responsive IB kinases. Specifically, we found that the kinase activities of IKK␣ and IKK␤ were elevated for both IB␣ and IB␤ when T cells were treated with mitogen and anti-CD28. In addition, a catalytically inactive mutant of IKK␣ effectively inhibited in vivo CD28-mediated IB␣ phosphorylation. Inactive forms of IKK␣ and IKK␤ also attenuated CD28RE/AP-1 luciferase gene reporter activity induced by PMA and CD28. This study provides strong evidence that signaling through the TCR and CD28 converge at or upstream of IKK␣ and IKK␤, resulting in enhanced kinase activity and NF-B activation.
Where could the TCR and CD28 pathways possibly converge? Kinases upstream of IKK include the mitogen-activated protein kinase kinase-related molecule, NIK, and the mitogenactivated protein kinase/ERK kinase kinase 1, MEKK1 (40). A recent report suggests that MEKK1 preferentially activates IKK␤, while NIK activates IKK␣ and IKK␤ equally well (41). Previous studies suggest that MEKK1 is a downstream target of CD28 signaling, and that its kinase activity is up-regulated when stimulated with both anti-CD3 and anti-CD28 (42,43). It is not likely that MEKK1 is solely responsible for the CD28mediated up-regulation of IKK kinase activity since we consistently observed stronger kinase activity associated with IKK␣ rather than IKK␤. NIK is a potential candidate to mediate the signal integration between the TCR and CD28 if the signals do in fact converge upstream of IKK. NIK is already known to integrate signals from pathways initiated by IL-1 and TNF-␣ to activate NF-B (30). We are currently investigating any potential role NIK may play in CD28 mediated NF-B activation.
Besides MEKK1, several other kinases have been identified which have up-regulated kinase activity when T cells are stimulated with anti-CD3 and anti-CD28. Full activation of JNK in T cells is dependent on integration of the two signals (44). As expected, kinases within the JNK pathway such as p21-activated kinase (43), SEK (45), and MKK7 (45) are similarly dependent on T cell costimulation for full activation. The transactivation capacity of the c-Jun protein, one of the AP-1 components, is activated as a result of signaling through the JNK pathway (46). The vital role that AP-1 plays in IL-2 transcrip- FIG. 3. IKK␤ kinase activity is potentiated by the CD28 costimulatory signal. A and B, Jurkat cells were treated exactly as described in Fig. 2, except that anti-IKK␤ was used for the immunoprecipitation. Cell lysates were subjected to in vitro kinase assays with GST-IB␣ 1-55 in A and GST-IB␤ 1-82 as substrate in B.

FIG. 4. A catalytically inactive IKK␣ mutant blocks CD28-mediated in vivo IB␣ phosphorylation.
Jurkat Tag cells (5 ϫ 10 6 ) were transfected with 2 g of an HA-tagged IB␣ cDNA together with either an empty vector (lanes 1-3), or the indicated amounts of HAtagged IKK␣ K44M (lanes 4 -7). Approximately 40 h later, transfectants were split into two and either left untreated (lanes 1, 2, 4, and 6) or treated with PMA and anti-CD28 for 30 min (lanes 3, 5, and 7). Whole cell lysates were collected and subjected to immunoblotting with an anti-HA antibody. tional regulation is underscored by the finding that mice deficient in SEK1, a direct activator of JNK, are impaired in CD28-mediated IL-2 production (47). However, it is also known that NF-B is a critical regulator of the IL-2 gene as demonstrated by gene targeting of the c-rel gene (48). To date, kinases solely within the NF-B pathway have not been identified to be targets of CD28 mediated activation. Our finding that IKK␣ and IKK␤ have enhanced kinase activity when T cells are costimulated is the first such demonstration. It is likely that other kinases within the NF-B pathway are also targeted by CD28.
We have observed an excellent correlation between in vitro kinase activities of IKK␣ and IKK␤ and in vivo phosphorylation of IB␣ mediated by CD28 (see Figs. 1 and 3). IB␤ is also strongly phosphorylated in vitro by both IKK␣ and IKK␤ due to CD28 (Figs. 2 and 3). With regard to the in vivo phosphorylation of IB␤, it is more difficult to address this question, since a band shift is not apparent after cellular stimulations. Rather, IB␤ is degraded partially within 15 min and completely in 30 min in response to mitogen and CD28 treatment (23). It is possible that IB␤ is not phosphorylated as well in vivo due to the folding of the protein or perhaps due to the binding of other proteins that may interfere with the accessibility of IKKs to the two N-terminal serine residues. Further studies with in vivo 32 P-labeling of IB␤ will more directly answer this question.
In conclusion, we have determined the mechanism of CD28mediated NF-B activation to be at the level of enhanced IB kinase activity. It appears that CD28 targets the two cytokineresponsive IB kinases, IKK␣ and IKK␤, which are able to respond to multiple signals. It is therefore not likely that CD28 induces an IB kinase distinct from IKK␣ or IKK␤, which is unique to the CD28 pathway. We also consider it unlikely that CD28 directly enhances the ubiquitination or degradation of IB␣ and IB␤. Studies are in progress to further delineate the TCR and CD28 pathways and to pinpoint the convergence of the two pathways in NF-B activation.