Protein Kinase C θ Cooperates with Vav1 to Induce JNK Activity in T-cells*

Here we show that in human T-cell leukemia cells Vav1 and protein kinase C θ (PKCθ) synergize for the activation of c-Jun N-terminal kinase (JNK) but not p38 MAP kinase. Vav1 and PKCθ also cooperated to induce transcription of reporter genes controlled either by AP-1 binding sites or the CD28RE/AP composite element contained in the IL-2 promoter by stimulating the binding of transcription factors to these two elements. Dominant negative versions of Vav1 and PKCθ inhibited CD3/CD28-induced activation of JNK, revealing their relative importance for this activation pathway. Gel filtration experiments revealed the existence of constitutively associated Vav1/PKCθ heterodimers in extracts from unstimulated T-cells, whereas T-cell costimulation induced the recruitment of Vav1 into high molecular weight complexes. Several experimental approaches showed that Vav1 is located upstream from PKCθ in the control of the pathway leading to synergistic JNK activation. Vav1-derived signals lead to the activation of JNK by at least two different pathways. The major contribution of Vav1 for the activation of JNK relies on the PKCθ-mediated Ca2+-independent synergistic activation pathway, whereas JNK is also activated by a separate Ca2+-dependent signaling route.

target proteins including phospholipase C␥, which controls the phosphatidyl inositol lipid metabolism, thereby producing inositol triphosphate and diacylglycerols (6). Whereas inositol triphosphate results in a rapid and sustained calcium increase, diacylglycerol mediates activation of PKC family members (6,7). Among those, the novel Ca 2ϩ -independent PKC isoform PKC is of special importance for T-cells, because it is rapidly recruited to the site of contact between T-cells and antigenpresenting cells (8). Another protein tyrosine kinase-induced signaling route is mediated by the Vav1 protein family member Vav1, which is exclusively expressed in hematopoietic cells (9). T-cell costimulation induces the membrane recruitment of Vav1 via indirect, adaptor protein SLP-76-mediated binding to the membrane protein LAT (linker for activation of T-cells) (10). Protein tyrosine kinase-induced phosphorylation and phosphatidylinositol-3,4,5-triphosphate binding of Vav1 activate its GDP/GTP exchange factor activity for the Rho family of GTPases such as Rac and Cdc42 (11,12) and results in the stimulation of signaling pathways and alterations in cell shape and motility.
T-cell costimulation also leads to the activation of mitogenactivated protein kinase pathways. However, the synergistic activation appears to be unique for the mitogen-activated protein kinases JNK and p38 (13), because neither extracellular signal-regulated kinase nor transcription factor nuclear factor of activated T-cells require coreceptor-derived signals. JNK phosphorylates various transcription factors including ATF2, ELK-1, and components of the AP-1 heterodimer, namely JunB, JunD, and c-Jun (14). Because AP-1 contributes to the induced expression of numerous target genes including IL-2 and IL-4, the JNK pathway has been implicated in various functions including cell proliferation, effector T-cell function (15), T-cell activation (16), and the regulation of apoptosis (14). However, these functions are dependent on the inducing signal and the cell type (15,16). JNK is activated by the dual specificity JNK kinases (JNKKs) MKK4/JNKK1/SEK1 and JNKK2/ MKK7 (17). A variety of different kinases can activate the JNKKs, but the cell type and the nature of the JNK-inducing stimulus determines which of these kinases is operational (14).
Vav1 and PKC are constitutively associated in unstimulated T-cells (18), and both proteins synergistically activate transcription factor NF-B, JNK activity, and the expression of IL-2, CD69, and IL-4 (4,19,20). In this study we have addressed the question whether Vav1 and PKC synergize for the activation of JNK for two reasons. 1) Gain-of-function approaches have revealed that both proteins contribute to the activation of JNK (10,(21)(22)(23)(24)(25)(26) and cooperate with constitutively active calcineurin or Ca 2ϩ signals to activate JNK. 2) We have previously seen that the Vav1/PKC module synergistically triggers binding of transcription factors to the P1 and PRE-I elements contained within the IL-4 promoter (19). Because both DNA elements are bound by AP-1 family members, we tested the effects of Vav1 and PKC on the activation of JNK. A variety of experimental approaches revealed cooperative activation of JNK and AP-1-dependent gene expression by Vav1 and PKC.

EXPERIMENTAL PROCEDURES
Cell Culture, Transfections, and Stimulations-Jurkat T leukemia cells expressing the large T antigen were grown in RPMI 1640 medium at 37°C containing 10% (v/v) heat-inactivated fetal calf serum, 10 mM HEPES, 1% (v/v) penicillin/streptomycin, 2 mg/ml G418, and 2 mM glutamine (all from Life Technologies, Inc.). Cells were electroporated using a gene pulser (Bio-Rad) at 250 V/950 microfarads. In all transfections, the amount of total DNA (20 g) was kept constant by the addition of empty expression vector. Stimulations were performed in a volume of 400 l by adding agonistic ␣CD3 (final concentration of 10 g/ml) and/or ␣CD28 (final concentration of 10 g/ml) antibodies.
Electrophoretic Mobility Shift Assays (EMSAs) and Luciferase Determination-EMSAs were performed using nuclear extracts essentially as described (19). Equal amounts of nuclear protein were tested for protein binding to oligonucleotides containing either an AP-1 binding site or a CD28RE/AP element. The coding strands of the oligonucleotides used were as follows: AP-1, 5Ј-CGCTTGATGACTCAGCCGGAA-3Ј; CD28RE/AP, 5Ј-TCTGGTTTAAAGAAATTCCAAAGAGTCATCAG-3Ј. The free and the oligonucleotide-bound proteins were separated by electrophoresis on a native 4% polyacrylamide gel. Following electrophoresis the gel was dried and exposed to an x-ray film (Amersham Hyperfilm).
Luciferase activity in cell extracts was measured in a luminometer (Duo Lumat LB 9507, Berthold) by automatically injecting 50 l of assay buffer and measuring light emission for 10 s after injection according to the instructions of the manufacturer (Promega Inc.). To ensure comparable transfection efficiencies, results were normalized to ␤-galactosidase produced by a cotransfected Rous sarcoma virus-␤galactosidase expression vector.
Cell Extracts and Western Blotting-Cells were washed with phosphate-buffered saline, and the pellets were resuspended on ice for 15 min 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 10% (v/v) glycerol). Cell debris was pelleted upon centrifugation, and the supernatant was either directly analyzed by Western blotting or used for the determination of JNK activity as described below. After separation of cell extracts on reducing SDS-polyacrylamide gels, the  ). B, an expression vector for Flag-tagged p38 (1 g) and Vav1 and/or PKC A/E (5 g, respectively) were transfected in Jurkat cells, which were either left untreated or costimulated for 30 min as shown. The tagged p38 protein was immunoprecipitated and analyzed by Western blotting for p38 expression and phosphorylation. Samples of whole cell lysates were immunoblotted for the expression levels of the various transfected proteins (lower panels).

FIG. 2. Synergistic activation of AP-1 by Vav1 and PKC. A,
Jurkat cells were transfected either with empty expression vector or with plasmids encoding Vav1 (10 g) and/or PKC A/E (10 g) at the indicated combinations. The next day, nuclear cell extracts were prepared, and the DNA binding activity of AP-1 was determined by EM-SAs. An autoradiogram is displayed; the arrow indicates the location of the DNA-AP-1 complex, and the circle indicates the position of the unbound oligonucleotide. B, Jurkat cells were transiently transfected with 5 g of an AP-1 luciferase reporter construct together with increasing amounts of PKC A/E and/or Vav1 at the indicated combinations. Luciferase activity was determined 30 h posttransfection. Gene expression is displayed as the average fold activation relative to vector-transfected cells. Mean values from three independent experiments are shown.
proteins were transferred onto a polyvinylidene difluoride membrane (Millipore) using a semi-dry blotting apparatus (Bio-Rad). The membrane was then incubated in a small volume of TBST buffer (25 mM Tris/HCl, pH 7.4, 137 nM NaCl, 5 mM KCl, 0.7 mM CaCl 2 , 0.1 mM MgCl 2 , 0.1% (v/v) Tween 20) containing various dilutions of the primary antibodies. After extensively washing the membrane, the immunoreactive bands were visualized by enhanced chemiluminescence according to the instructions of the manufacturer (PerkinElmer Life Sciences). Western blots were quantitated using the Lumi-Imager TM from Roche Molecular Biochemicals.
JNK Assays-Two days posttransfection, cell lysates were prepared and precleared with protein A/G-Sepharose. The HA-tagged JNK proteins contained in the cell lysate were precipitated by the addition of 1 g of ␣HA antibody and 25 l of protein A/G-Sepharose. The precipitate was washed three times in 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 2 g of glutathione Stransferase (GST)-c-Jun-(5-89), 20 M ATP, and 5 Ci of [␥-32 P]ATP for 20 min at 30°C. The reaction was stopped by the addition of 5ϫ SDS loading buffer, followed by reducing SDS-PAGE, gel fixation, and quantification of the results in a PhosphorImager.
Gel Filtration-The analysis of cellular multi-protein complexes was performed on a Superose 6 column (Amersham Pharmacia Biotech). Total cell extracts from 3 ϫ 10 8 Jurkat cells contained in 75 l of octylglycoside lysis buffer (20 mM Tris/HCl, pH 7.5, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 5 mM dithiothreitol, leupeptin (10 g/ ml), aprotinin (10 g/ml), 1% (v/v) octylglycoside, and 10% (v/v) glycerol) were analyzed at a flow rate of 0.5 ml/min, and fractions of 500 l were collected. Aliquots from the respective fractions were analyzed by Western blotting for the occurrence of Vav1 and PKC. The washed columns were calibrated with pre-made molecular weight standards (Amersham Pharmacia Biotech).  Fig. 2B, with the exception that a construct controlled by four repeats of the CD28RE/AP element fused to the luciferase reporter gene (4xRE/AP-Luc) was used. WT, wild type. B, Vav1 and/or PKC A/E (4 g, respectively) were expressed in Jurkat cells, and their effects on cotransfected reporter constructs with intact or mutated CD28RE/AP elements was measured. Results are expressed as the average fold activation relative to vector-transfected cells, and error bars indicate standard deviations from three independent experiments. C, Jurkat cells were transfected either with empty expression vector or with plasmids encoding Vav1 (10 g) and/or PKC A/E (10 g) as shown. The next day, cells were stimulated for 4 h with ␣CD3/␣CD28 antibodies as indicated, and equal amounts of protein contained in nuclear extracts were assayed for binding to a labeled CD28RE/AP element by EMSAs. The positions of the constitutive complexes II and III and of the inducible complexes Ia and Ib are indicated. The arrowhead indicates the position of the unbound oligonucleotide; a representative experiment is shown. and/or wild type Vav1. The tagged JNK protein was immunoprecipitated, and its kinase activity was determined by immune complex kinase assays (Fig. 1A). JNK activation triggered by expression of Vav1 and PKC alone was synergistically stimulated upon coexpression of both proteins. Costimulation with agonistic ␣CD3/␣CD28 antibodies further triggered JNK activity elicited by Vav1, PKC A/E, or both. To determine whether the synergy between Vav1 and PKC also occurs for the activation of p38, Jurkat T-cells were cotransfected with expression vectors for Flag-tagged p38 together with different combinations of vectors encoding Vav1 and PKC A/E. The tagged p38 protein was immunoprecipitated and analyzed for its activation (as seen by Thr-180/Tyr-182 phosphorylation) in Western blot experiments (Fig. 1B). These experiments showed that Vav1 and PKC activated p38 only in an additive, non-synergistic manner, thereby revealing that the cooperation only occurs for the activation of JNK. To test whether this synergism is also apparent at the level of induced AP-1 binding to its cognate DNA, different combinations of expression vectors for Vav1, PKC A/E, or the empty expression vector as a control were expressed in Jurkat cells. Analysis of AP-1 DNA binding activity by EMSAs showed that either Vav1 or PKC A/E alone induced DNA binding of AP-1, albeit to different extents. Coexpression of both proteins synergistically stimulated the DNA binding activity of AP-1 (Fig. 2B). The impact of Vav1/PKC expression on the activation of AP-1-dependent transcription was tested in reporter gene assays. An AP-1-dependent luciferase gene was transfected into Jurkat cells together with increasing amounts of Vav1 and/or PKC A/E expression vectors (Fig. 2B). The slight induction of AP-1dependent transcription mediated by Vav1 was strongly enhanced even by moderate amounts of coexpressed PKC A/E. These experiments revealed that the synergism also occurs at the level of DNA binding and gene expression.

RESULTS
JNK induces IL-2 promoter activity not only by targeting the AP-1 site but also via the CD28RE/AP composite element, which is bound by various proteins including members of the NF-B/Rel and AP-1 families of transcription factors and so far only partially characterized proteins (27,31). Because the CD28RE/AP element is absolutely required for the transmission of CD28-derived signals on the IL-2 promoter (27), we asked whether Vav1/PKC-mediated JNK activation also targets gene expression directed from this sequence element. A construct controlled by four repeats of the CD28RE/AP element fused to the luciferase reporter gene (4xRE/AP-Luc) was cotransfected with increasing amounts of Vav1 and/or PKC expression vectors (Fig. 3A). Both proteins cooperatively boosted CD28RE/AP-dependent transcription. To test whether the Vav1/PKC-derived JNK activity targets the CD28RE or the AP-1 half-site within the composite element, reporter constructs with mutations in either or both half-sites were tested for their inducibility. Mutation of either half-site completely prevented Vav1/PKC-mediated transcription (Fig. 3B), revealing a strong interdependence of both half-sites. The effects of Vav1 and PKC on protein binding to the CD28RE/AP element were investigated by EMSAs using the labeled CD28RE/AP element (Fig. 3C). Neither Vav1 nor PKC alone were able to induce DNA binding, but coexpression of both proteins caused DNA binding of transcription factors contained in the inducible DNA-protein complexes Ia and Ib. Formation of both complexes, which are known to contain predominantly c-Fos and c-Jun proteins (31), could be further triggered upon CD3/CD28 stimulation. In contrast to complex I, DNA-protein complexes II and III were already present in extracts from the untransfected, nonstimulated cells and showed no inducibility upon Vav1/PKC A/E coexpression and T-cell activation. The relatively moderate effects of Vav1/PKC on induced binding of proteins to the CD28RE/AP element can be attributed to the limited transfection efficiency of Jurkat cells.
To investigate the relative importance of Vav1 and PKC for the CD3/CD28-mediated activation of JNK, the costimulationinduced activation of this mitogen-activated protein kinase was tested in the presence of DN versions of both signaling proteins. Coexpression of the kinase-deficient point mutant PKC K/R prevented JNK activation induced either by CD3/CD28 ligation or by treatment of cells with the pleiotropic PKC activator phorbol 12-myristate 13-acetate together with CD28 or the ionophore ionomycin (Fig. 4A). Similarly, two Vav1 variants with a mutation or a deletion in the Dbl homology domain (Vav1 LLL/QIF and Vav1⌬319 -356), which is responsible for the activation of Rac, efficiently prevented CD3/CD28-triggered 32 P incorporation into GST-c-Jun by JNK (Fig. 4B).
Vav1 and PKC are constitutively associated in unstimulated T-cells (18) but can also interact with a battery of further proteins, as revealed by various experimental approaches (9,32). Whereas T-cell costimulation leads to the recruitment of Vav1 into an inducible multi-protein complex (10), the association partners of Vav1 in unstimulated cells are not well characterized. To directly address this question, unstimulated Jurkat cells were lysed, and proteins contained in the lysates were separated according to size by gel filtration. The analysis of fractions for the distribution of Vav1 and PKC revealed an identical elution profile for both proteins, as revealed by Western blotting (Fig. 5A). The two fractions (22 and 23) containing most of Vav1 and PKC correspond to a molecular mass of ϳ170 kDa, suggesting that both proteins exist predominantly as heterodimers. Because previous studies showed that T-cell costimulation leads to the disruption of the Vav1/PKC heterodimer (4), we investigated the distribution of Vav1 and PKC in extracts from CD3/CD28-stimulated cells. These experiments revealed the majority of Vav1 in high molecular weight complexes of various sizes up to very large aggregates of more than 1 MDa (Fig. 5B). In contrast, most of the PKC protein coeluted with the 158-kDa marker protein, and only a minor fraction was found in larger complexes (Fig. 5B).
The signaling pathways were characterized by monitoring the effect of coexpressed DN forms of protein kinases from the JNK activation pathway on the activation signals derived from Vav1, PKC, or both (Fig. 6A). A kinase-dead form of MEKK1 affected Vav1-mediated AP-1 activity only moderately but significantly inhibited PKC-and PKC/Vav1-generated signals. A comparison between DN forms of the JNKKs MKK4 and MKK7 revealed that Vav1-derived activation signals were only moderately blocked by expression of DN forms of each of these kinases. In contrast, simultaneous expression of kinase-inactive forms of MKK4 and MKK7 completely prevented Vav1mediated AP-1 activation, indicating that both kinases can mutually compensate the functions of each other. PKC-and Vav1/PKC-derived signals were only partially impaired in the presence of DN MKK4 but efficiently blocked by MKK7 K/L, indicating the special importance of MKK7 for this pathway.
We also tested the impact of pathway-specific inhibitory compounds on Vav1-and/or PKC-induced AP-1-dependent transcription (Fig. 6B). Vav1-derived signaling, but not PKCand Vav1/PKC-mediated AP-1 activation, was preferentially inhibited by cyclosporin A, a compound that blocks the Ca 2ϩdependent activation of the phosphatase calcineurin. The PKC inhibitor bisindolylmaleimide blocked Vav1-and PKC-mediated AP-1 activity, raising the possibility that Vav1 acts upstream from PKC. To address this question directly, Jurkat cells were transfected with various combinations of active and inactive variants of Vav1 and PKC prior to stimulation with ␣CD3/␣CD28 antibodies and subsequent analysis of JNK activity. Vav1-induced JNK activation was efficiently prevented by kinase-dead PKC K/R, but PKC A/E-mediated JNK acti-vation was not affected by DN Vav1 LLL/QIF (Fig. 7A). To investigate whether directional signaling also occurs at the level of AP-1-dependent transcription, an analogous experimental approach was taken by monitoring AP-1-dependent luciferase activity. Similarly, Vav1-induced transcription of the AP-1-dependent luciferase gene was inhibited upon coexpression of PKC K/R (Fig. 7B) or the PKC inhibitor bisindolylmaleimide (data not shown). In contrast, gene activation induced by PKC A/E was not affected by the DN Vav1 variant Vav1⌬319 -356. In summary, these data clearly indicate that PKC acts downstream from Vav1 in a pathway leading to the activation of JNK. DISCUSSION Here we show that Vav1 and PKC synergize for the induction of JNK. Because these two proteins also cooperate for the up-regulation of NF-B, the simultaneous activation of several pathways may be important for the efficient transcription of target genes, because many promoters (e.g. the IL-2 and IL-4 promoter) depend on the coordinated activation of several transcription factors. Interestingly, PKC does not only synergize with Vav1 (this study) but also with calcineurin for the activation of JNK (22). The Vav1 protein cooperates with SLP-76 and Syk family kinases for the activation of IL-2 transcription and nuclear factor of activated T-cells activation, respectively (33,34). One could speculate that the synergistic behavior of signaling proteins for the initiation of T-cell signaling pathways may be an important regulatory principle, especially early in infection when only low doses of antigen are present, but an efficient cellular response is required. In that respect it will be interesting to analyze signaling pathways and the immune response in Vav1/PKC double knockout mice in future experiments.
In this study we show that at least the abundant fraction of PKC and Vav1 proteins is present as heterodimers in unstimulated cells under the conditions used here. Given the importance of both proteins for the initiation of various signaling pathways (10,32) and the necessity to keep activation pathways in nonstimulated T-cells silent (35,36), this heterodimerization might serve the purpose of keeping both proteins in an inactive state. T-cell activation leads to the transient dissociation of Vav1 and PKC by an unknown mechanism (4) and the incorporation of Vav1 into temporally regulated and highly dynamic and semi-stable multi-protein signaling complexes (37). Under the conditions used here, only a minor fraction of the PKC protein is incorporated into high molecular weight complexes in stimulated T-cells. It will be interesting to investigate the composition, stability, and spatial and temporal regulation of these multi-protein complexes.
Our results indicate that Vav1 is located upstream from PKC in the JNK activation cascade. This finding is in good agreement with a previous study that demonstrated a Vav1dependent membrane and cytoskeleton translocation of PKC (20). The same study shows that the effects of Vav1 on PKC are mediated by Vav1-induced actin polymerization and cytoskeletal reorganization. Biochemical and genetic evidence suggests that Vav1 exerts its function by at least two mechanisms. One pathway leads to the release of Ca 2ϩ and the subsequent activation of the serine phosphatase calcineurin, which stimulates nuclear entry and transactivation of transcription factor nuclear factor of activated T-cells. This pathway is independent from the GDP/GTP exchange factor function of Vav1 (38) and cannot be inhibited by the regulatory protein Cbl-b (39). The second pathway is Ca 2ϩ -independent, can be inhibited by Cbl-b, and relies on the GDP/GTP exchange factor function of Vav1, thus leading to actin polymerization, cytoskeletal reorganization, and further processes (39). We favor a model where Vav1 and PKC activate JNK via overlapping and distinct pathways. The PKC-derived signals are Ca 2ϩ -independent and cannot be inhibited by calcineurin, because PKC is located downstream from it. Accordingly, a DN form of PKC inhibits JNK activation mediated by ionomycin, which causes the release of intracellular Ca 2ϩ (compare Fig.  4A). Similarly, the pathway shared by both signaling proteins and mediating the synergistic activation of JNK is Ca 2ϩ /calcineurin-independent. In contrast, the Vav1-mediated JNK activation pathway contains a Ca 2ϩ -dependent component that can be inhibited by cyclosporin A. This model would also explain earlier results that showed only moderate JNK activation by phorbol 12-myristate 13-acetate and ionomycin alone, whereas the simultaneous administration of both compounds strongly activated JNK activity (23).
This study suggests that Vav1-derived signals target MKK4 and MKK7. In contrast, PKC-and Vav1/PKC-derived signals are transmitted preferentially via MKK7, which seems to be of special relevance for this pathway. The importance of MKK4 for the CD3/CD28-induced activation of JNK is still not clear. One group describes a defective JNK activation in thymocytes obtained from MKK4-deficient mice but normal JNK activation in peripheral T-cells (40). Another study demonstrates normal activation of JNK in response to CD3/CD28 stimulation in lymph node cell suspensions from MKK4 Ϫ/Ϫ mice (41). Further studies are required to resolve these differences, which may also depend on the developmental stage of the lymphocytes and the cell types studied. In accordance with the predominant role of MKK7 described here, peripheral T-cells from mice lacking MKK7 show only low levels of JNK activity after CD3/CD28 stimulation (15).
The PKC-mediated signaling pathways are only incompletely understood. Furthermore, it is currently not known whether the competence of PKC to deliver activation signals is controlled by its intracellular localization, interaction with binding partners, or both. Because overexpression of the PKCinteracting protein, PKC-interacting cousin of thioredoxin, inhibits PKC-mediated activation of JNK and NF-B (42), this protein seems to be involved in the first signaling step prior to the separation of the two signaling pathways. Given the importance of PKC for the activation of NF-B and JNK, selective inhibition of this kinase might be a useful strategy to interfere with several activation pathways and thereby modulate T-cell costimulatory signals in inflammatory diseases.