A p56 lck -independent Pathway of CD2 Signaling Involves Jun Kinase*

The p56 Src family non-receptor tyrosine kinase has been shown to be critical for T lymphocyte differentiation and activation. Hence in the absence of p56, T cell receptor triggered activation does not occur. We now provide evidence for a CD2-based signaling pathway which, in contrast to that of the T cell receptor, is independent of p56. CD2-mediated interleukin-2 production occurs via activation of Jun kinase in cell lines lacking p56. Jun kinase then facilitates the binding of c-Jun/c-Fos heterodimers to the AP-1 consensus site and the subsequent transcriptional activity of the interleukin-2 promoter. These data elucidate differences between TCR and CD2 signaling pathways in the same T cells.

The p56 lck Src family non-receptor tyrosine kinase has been shown to be critical for T lymphocyte differentiation and activation. Hence in the absence of p56 lck , T cell receptor triggered activation does not occur. We now provide evidence for a CD2-based signaling pathway which, in contrast to that of the T cell receptor, is independent of p56 lck . CD2-mediated interleukin-2 production occurs via activation of Jun kinase in cell lines lacking p56 lck . Jun kinase then facilitates the binding of c-Jun/c-Fos heterodimers to the AP-1 consensus site and the subsequent transcriptional activity of the interleukin-2 promoter. These data elucidate differences between TCR and CD2 signaling pathways in the same T cells.
CD2 is a cell surface glycoprotein which is expressed on immature thymocytes and on mature T cells and NK cells (1)(2)(3). The ligand for human CD2 is CD58 (LFA-3), a ubiquitously expressed cell surface protein found on many cell types including antigen presenting cells (4,5). Unlike integrins such as CD11a/CD18 (LFA-1), the adhesion between CD2 and CD58 is not dependent on TCR 1 triggering (6,7). Rather, the CD2-CD58 interaction between T cells and their cognate partners facilitates the T cell recognition process and subsequent T cell activation (8 -10).
Cross-linking of CD2 molecules by specific pairs of mAbs recognizing distinct epitopes on human CD2 initiates a signaling cascade which leads to T cell cytokine production, proliferation, and cytolytic activity (11)(12)(13). For optimal mAb-mediated activation two antibody specificities are needed, one directed against conventional CD2 epitopes and the second against an activation-associated epitope of CD2, termed CD2R. The CD2R epitope has recently been mapped to the flexible linker region between the two extracellular domains of CD2 (14). Its appearance coincides with reorientation of the adhesion domain (D1) relative to the membrane proximal domain (D2) of the CD2 extracellular segment. Such domain reorientation occurs upon CD58 binding to CD2 and following T cell activation (14). Furthermore, the CD2 interaction with CD58 regulates the responsiveness of activated human T cells to IL-12 (15,16). Hence, both IL-12-stimulated T cell proliferation and interferon-␥ production are markedly augmented by CD2 ligation. Finally, some studies have uncovered a unique role for CD2 in the regulation of anergy (17). Stimulation of anergized alloreactive T cells with a combination of specific alloantigen in conjunction with CD2-CD58 co-receptor ligation reverses the anergic state. In contrast, identical allostimulation but in the absence of CD2-CD58 co-ligation on T cells and allostimulators, respectively, fails to restore responsiveness (17).
Signaling through CD2 is dependent on its cytoplasmic domain (8, 18 -20). Comparison of CD2 from various species shows the highest homology occurring in this segment (21,22). While the cytoplasmic domain has no intrinsic protein-tyrosine kinase activity and no tyrosine residues which might serve as docking sites for SH2 domains upon phosphorylation (reviewed in Refs. 23 and 24), stimulation via CD2 leads to the tyrosine phosphorylation of several intracellular proteins (25)(26)(27). For these activation events CD2 signaling requires the presence of the CD3 chain in T cells (27)(28)(29)(30)(31) or Fc⑀RI␥ in CD16 expressing NK cells (32).
The CD2 tail contains several proline-rich regions (19). Two of these (PPPGHR) are known to be important in inducing IL-2 production (33). Related proline-rich sequences bind to SH3 domains of non-receptor protein-tyrosine kinases of the Src family (23). For example, the SH3 domain of p56 lck has been shown to bind to the rat CD2 cytoplasmic tail at least in vitro (34). Peptides corresponding to residues 269 -270 and 200 -310 (according to the human CD2 numbers) bind to p56 lck -GST fusion proteins (34). In vivo intracellular colocalization experiments, however, suggest, that p59 fyn rather than p56 lck has a predominant association with CD2 (35), perhaps by binding through the p59 fyn SH3 domain.
The activation of protein tyrosine kinases is one of the proximal steps in signaling cascades which ultimately lead to the activation of T cell effector function. For p56 lck it has been shown that this kinase is involved in the activation of the Ras/Raf/MAPK pathway (36). The function of MAPK is manifold. Several studies reported that MAPK can phosphorylate and activate the transcription factors ELK, Jun, and Stat-1 and thereby participates in the regulation of gene expression (37). A different pathway leading to Jun activation is the Rac/JNK pathway (38). Recently, the Pyk2 tyrosine kinase has been described, shown to be involved in the activation of JNK (39), and has been linked to TCR signaling via p59 fyn (40).
In this study, we have begun to dissect the differences in TCR-and CD2-based signaling pathways. Since p56 lck is thought to be most important at a proximal step in CD3 signaling, we investigated the role p56 lck plays in CD2 signaling. To this end, we have employed the Jurkat variant JCaM1.6s which lacks functional p56 lck and is unable to transmit TCR signals leading to IL-2 production. We here report that although the TCR pathway is not functional in JCaM1.6s, these cells can be induced to produce IL-2 by stimulation via CD2. Furthermore, we present evidence for a p56 lck -independent CD2 signal transduction pathway which uses Pyk2 and JNK and leads to the binding of c-Jun/c-Fos heterodimers to the AP-1 consensus site, important for initiation of IL-2 transcription.
Cell Lines-The Jurkat variant, JCaM1.6, that lacks p56 lck (42) was obtained from ATCC (Rockville, MD) and sorted for high CD2 and CD3 expression (JCaM1.6s) and further sorted for those cells which show a fast and high rise in intracellular free Ca 2ϩ levels following CD2 stimulation (JCaM1.6s.S3). Both JCaM1.6s and JCaM1.6s.S3 were analyzed for p56 lck by in vitro kinase activity and p56 lck was found to be not enzymatically active. J77 is a CD2 pos CD3 pos CD8 neg subclone of the Jurkat cell line (29).

Ca 2ϩ Assay
Ca 2ϩ assay was performed as described (43). Increase in intracellular free Ca 2ϩ in Indo-1 (Molecular Probes Inc., Eugene, OR) loaded cells was induced by either a combination of the anti-CD2 mAbs anti-T11 2 and anti-T11 3 (1:100) or 20 g/ml biotinylated RW28C8 alone or by further cross-linking with 100 g/ml avidin (Sigma). Maximal Ca 2ϩ influx was induced by the addition of 10 g/ml 4-bromo-calcium ionophore A23187 (Sigma). The analysis was performed on an EPICS V cell sorter (Coulter, Hialeah, FL).
In Vitro Kinase Assay-Lysis of cells (1 ϫ 10 7 cells/ml) and immunoprecipitation was carried out as described above. After washing with lysis buffer and kinase buffer (100 mM NaCl, 5 mM MnCl 2 , 5 mM MgCl 2 , 20 mM Hepes, pH 7.4) the beads bound immune complexes were resuspended in 50 l of kinase buffer containing 2 M ATP and were incubated with 10 Ci of [␥-32 P]ATP for 15 min at room temperature. Kinase reaction was stopped by addition of ice-cold lysis buffer containing 20 mM EDTA. After washing 3 times with lysis buffer/EDTA and once with Tris-HCl, pH 7.4, the phosphorylated proteins were eluted by boiling in SDS sample buffer. The proteins were analyzed by 9% SDS-PAGE and after drying the gels were subjected to autoradiography. Some in vitro kinase assays were carried out in the presence of 10 g/sample poly(Glu-Tyr) (4:1) (Sigma). The kinase reaction was stopped by boiling in SDS sample buffer and the samples were further processed as described above.
Western Blotting-Western blotting onto nitrocellulose membranes of proteins resolved by SDS-PAGE was performed according to standard procedures. Protein detection was carried out by incubating the membranes with 1:1000 dilutions of specific mAbs (4G10, PLC-␥1) or rabbit heteroantisera (ZAP70, p56 lck , p59 fyn , Pyk2, pMAPK, Erk1/ MAPK, Raf) for 1 h at room temperature. The blots were washed with TBS/Triton X-100 (0.05%) and incubated with 1:2000 dilutions of HRPO-labeled anti-mouse IgG2b, anti-mouse IgG, or anti-rabbit Abs (Caltag Laboratories, San Francisco, CA), respectively, for 1 h at room temperature. After washing with TBS/Triton X-100, the proteins were visualized by enhanced chemiluminescence (ECL Western blotting detection reagents, Amersham International, Little Chalfont, Buckinghamshire, United Kingdom) and exposing the membranes to films for various time intervals. Unspecific binding of Abs was inhibited by preincubating the membranes with blocking buffer (TBS containing either 5% fetal calf serum or 5% non-fat dry milk and 10 mM NaN 3 ) for at least 2 h at room temperature.

Nuclear Extracts and EMSA
5 ϫ 10 7 cells, unstimulated or stimulated for 4 or 6 h at 37°C with either a combination of stimulatory anti-CD2 (anti-T11 2 plus anti-T11 3 , 1:100) or with a 1:100 dilution of anti-CD3⑀ mAb 2AD2A2, were incubated on ice for 15 min with 400 l of ice-cold Buffer A (10 mM HEPES pH 7.9, 10 mM KCl, 1.5 mM MgCl 2 , 0.5 mM DTT, 0.2 mM PMSF) and spun at 13,000 rpm for 10 s. The pellets were resuspended in 100 l of ice-cold buffer C (10 mM HEPES pH 7.9, 420 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 25% (v/v) glycerol, 0.5 mM DTT, 0.2 mM PMSF) and left for 30 min on ice. After spinning for 5 min at full speed, the supernatants were removed, aliquoted, and stored at Ϫ80°C. The protein concentration in the supernatants was determined by BCA Assay (Pierce) and 10 g/sample were used in the EMSA.
10 g of NE were incubated in binding buffer (10 mM HEPES pH 7.5, 30 mM KCl, 5 mM MgCl 2 , 0.1 mM EDTA, 12% (v/v) glycerol, 0.5 mM DTT, 0.2 mM PMSF) for 30 min on ice with 2 g of poly(dI-dC) (Boehringer p56 lck -independent CD2 Signaling Mannheim) and either 2 l of the respective Ab or a 100-fold excess of unlabeled oligonucleotide as competitor. Subsequently, 0.5 ng of 32 Plabeled oligo were added and incubated with the NE at room temperature for an additional 20 min. The proteins were separated on 5.2 (EMSA) or 4% (supershift) gels. The gels were dried and subjected to autoradiography.

RESULTS
The CD2 pos CD3 pos JCaM1.6s Cells do Not Express p56 lck -JCaM1.6 is a Jurkat cell line variant, which lacks functional p56 lck and is unresponsive to TCR stimulation such that it fails to generate IL-2 upon anti-CD3 cross-linking (42). To exclude the possibility that different TCR and/or CD2 surface expression levels between Jurkat and JCaM1.6 might account for a component of the signaling defect, JCaM1.6 cells were sorted on a FACSVantage for high co-expression of both CD2 and CD3, equivalent to the surface expression number of these receptors found on the Jurkat line J77 (Fig. 1A). The sorted JCaM1.6 (referred to as JCaM1.6s) as well as JCaM1.6s cells, further sorted for maximal Ca 2ϩ mobilization upon CD2 crosslinking (termed JCaM1.6s.S3), were analyzed by in vitro kinase assay for p56 lck autophosphorylation and poly(Glu-Tyr) substrate phosphorylation activity. As shown in Fig. 1B, and in contrast to J77, JCaM1.6s does not express detectable amounts of functional p56 lck , as judged by in vitro autophosphorylation of unstimulated, anti-CD2-, or anti-CD3⑀-stimulated JCaM1.6s cells. Similarly, by in vitro kinase assay, anti-p56 lck antibody immunoprecipitates from JCaM1.6s.S3 cells do not reveal autophosphorylation activity or any enzymatic activity using the poly(Glu-Tyr) substrate (data not shown).
CD2 Stimulation in the Absence of p56 lck Leads to a Prolonged and High Amplitude Rise in Intracellular Free Ca 2ϩ -Although TCR cross-linking by anti-CD3⑀ mAb or anticlonotypic mAb does not activate JCaM1.6 cells (42), the status of the CD2 pathway in these cells was not defined. To first address the integrity of CD2 mediated signaling, we analyzed Ca 2ϩ mobilization in both JCaM1.6s, the Ca 2ϩ sorted JCaM1.6 s.S3 and J77 following anti-CD2 or anti-CD3⑀ stimulation. In contrast to J77, in which both anti-CD2 and anti-CD3⑀ mAbs induced a long lasting and high rise in intracellular free Ca 2ϩ , only anti-CD2 stimulation via the combination of anti-T11 2 plus anti-T11 3 mAbs resulted in a substantial Ca 2ϩ influx in JCaM1.6s (data not shown). After sorting, the Ca 2ϩ response of JCaM1.6s.S3 to anti-CD2 stimulation was equal or even higher than in J77 ( Fig. 2A). Addition of biotinylated anti-CD3⑀ mAb alone was not sufficient for Ca 2ϩ mobilization in JCaM1.6s (data not shown) or in JCaM1.6s.S3 ( Fig. 2A). Rather extensive cross-linking of the biotinylated anti-CD3⑀ mAb by avidin was required to lead to a detectable Ca 2ϩ transient. This appears as a low magnitude rise in Ca 2ϩ of short duration and most probably is due to the initial release of Ca 2ϩ from intracellular Ca 2ϩ stores ( Fig. 2A) (41,44,45).
PLC-␥1 Is Tyrosine Phosphorylated and Activated by CD2 Stimulation in the Absence of p56 lck -Given that phosphorylation and activation of PLC-␥1 leads to increased phosphatidylinositol turnover and the release of Ca 2ϩ from intracellular Ca 2ϩ stores (45,46), the phosphorylation status of PLC-␥1 was determined prior to and following stimulation in the J77 and JCaM1.6s cells. In J77, the stimulation via both CD2 and CD3⑀ results in a strong tyrosine phosphorylation of PLC-␥1 as shown by anti-Tyr(P) Western blot analysis of anti-PLC-␥1 immunoprecipitations (Fig. 2B). In contrast, only activation via CD2 leads to a clear tyrosine phosphorylation of PLC-␥1 in JCaM1.6s (Fig. 2B), whereas anti-CD3⑀ stimulation had a minor effect (Fig. 2B). These differences were not a consequence of gel loading of the anti-PLC-␥1 immunoprecipitates as revealed by sequential Western blotting with anti-PLC-␥1 antibody.
CD2 but Not CD3 Stimulation Results in Interleukin 2 Produc- tion in the Absence of p56 lck -To next determine whether the CD2 pathway in JCaM1.6s could elicit IL-2 production, we determined the levels of IL-2 secretion in the supernatant of anti-CD2 stimulated JCaM1.6s cells in a human IL-2 specific enzymelinked immunosorbent assay (Table I) and by intracellular staining of IL-2 in these stimulated JCaM1.6s (data not shown). Parallel analysis was performed in the same cells following anti-CD3⑀ mAb stimulation. As shown in Table I, stimulation via CD2 in JCaM1.6s cells results in substantial IL-2 production, in fact to a level equivalent to 50% of the maximal induction measured upon bypassing receptor triggering with the combination of calcium ionophore plus PMA. In contrast, and as expected, no IL-2 production is induced by TCR cross-linking by anti-CD3⑀. Data from different subclones of JCaM1.6s obtained by sequential sorting for Ca 2ϩ mobilization following CD2 stimulation are shown (JCaM1.6s.S1, JCaM1.6s.S2, and JCaM1.6s.S3). The amount of IL-2 produced increases with enhanced Ca 2ϩ responsiveness to CD2 stimulation.
Neither p59 fyn nor ZAP70 Are Detectably Tyrosine Phosphorylated following CD2 Stimulation in JCaM1.6s-Tyrosine phosphorylation and activation of kinases such as the Src kinases p56 lck or p59 fyn or the related tyrosine kinase ZAP70 are major events in TCR signaling and are also thought to be involved in signal transduction via CD2 (47). The Jurkat variant JCaM1.6s gives us the opportunity to investigate these events in the absence of p56 lck . To this end, we performed a series of Western blots with J77 and JCaM1.6s cells following anti-CD2 or anti-CD3⑀ stimulation. In J77, activation via both CD2 and CD3 results in the tyrosine phosphorylation of p56 lck and the appearance of a second p60 lck band due to the serine/ threonine phosphorylation of p56 lck (48) (Fig. 3A, upper panel). As expected, both lck bands are absent in the phosphotyrosine blot of p56 lck immunoprecipitates from JCaM1.6s. Note that the p58 band appearing in the Tyr(P) blot of both J77 and JCaM1.6s is nonspecific, being unrelated to p56 lck . Reanalysis of the stripped blot with anti-lck antiserum shows that these p56-and p60-phosphorylated proteins in J77 indeed represent p56 lck (Fig. 3A, lower panel).
In J77, stimulation via the CD3 pathway leads to a strong tyrosine phosphorylation of ZAP70, whereas activation via CD2 induces weaker but definite ZAP70 phosphorylation (Fig. 3B,  upper panel). In contrast to this observation, in JCaM1.6s neither CD2 nor CD3 stimulation induce the phosphorylation of ZAP70 (Fig. 3B, upper panel). Nevertheless, similar amounts of ZAP70 were immunoprecipitated in each lane, as shown by the results of reprobing the stripped Tyr(P) blot with an anti-ZAP70 rabbit antiserum (Fig. 3B, lower panel).
In both J77 and JCaM1.6s, the tyrosine phosphorylation of p59 fyn increases minimally if at all after either CD2 or CD3 stimulation (Fig. 3C, upper panel). The corresponding p59 fyn blot probed with anti-p59 fyn antibody is shown in Fig. 3C, lower panel. Note that the band below p59 fyn is related to immunoglobulin heavy chain.
The Ras-MAPK Pathway Is Not Functional in the Absence of p56 lck -TCR stimulation of T cells leads to the activation of Ras resulting in the serine phosphorylation and activation of Raf, then subsequently the activation of the MAPK signaling cascade and finally to the initiation of IL-2 transcription (49). We investigated the activation of Raf and MAPK by Western blotting of the total lysate of either unstimulated or anti-CD2 or anti-CD3 stimulated J77 or JCaM1.6s.S3 cells with rabbit antisera specific for Raf, pMAPK, or MAPK (Fig. 4, A-C). In J77, both CD2 and CD3 stimulation induced the phosphorylation and molecular weight shift of Raf (Fig. 4A) and the subsequent phosphorylation of MAPK as shown by Western blotting with antisera specific for the phospho-MAPK (Fig. 4B). In contrast, in JCaM1.6s.S3 stimulation via CD2 had no effect on either kinase (Fig. 4, A and B). Surprisingly, anti-CD3 triggering in JCaM1.6s.S3 led to the phosphorylation and molecular weight shift of Raf (Fig. 4A) but not to the functional activation of Raf as judged by the very low or absent phosphorylation of MAPK (Fig. 4B). As shown in Fig. 4C by sequential immunoblotting with a MAPK specific antiserum, the described differences were not due to different amounts of MAPK in the total lysates. Similarly, in J77, but not in JCaM1.6s.S3, stimulation via either CD2 or CD3 enhanced the kinase activity of MAPK toward myelin basic protein in an in vitro kinase assay (data not shown). Additionally, CD2 stimulated IL-2 production in JCaM1.6s.S3 was not inhibited by coculture with the MEK1inhibitor PD98059, which interferes with the MAPK pathway upstream of MAPK (data not shown).
CD2 but Not CD3 Stimulation Activates Jun Kinase in the Absence of p56 lck -Initiation of IL-2 transcription requires the binding of certain DNA-binding proteins to specific regions within the IL-2 promoter (50). One of these DNA-binding elements is the AP-1 complex, which consists of homo-or heterodimers of members of the Jun and Fos protein family (51). In order to bind to and activate the IL-2 promoter, these dimers must assemble and the participating proteins become phosphorylated (51). Jun kinase (JNK) is involved in the serine phosphorylation and activation of Jun family members (52). Following CD2 stimulation in JCaM1.6s.S3 cells, activation of JNK is readily detected as shown by in vitro kinase activity using a GST-Jun fusion protein substrate (Fig. 5A). CD3 stimulation  3. Analysis of protein-tyrosine kinase activation in J77 and JCaM1.6s.S3. Cells were either unstimulated (Ϫ) or stimulated with either a combination of the anti-CD2 mAbs anti-T11 2 plus anti-T11 3 (␣CD2) or the anti-CD3⑀ mAb 2AD2A2 (␣CD3) and cell lysates were immunoprecipitated with antisera specific for p56 lck , ZAP70, or p59 fyn , respectively. Immunoprecipitates in panels A-C using the designated Abs were subjected to SDS-PAGE followed by Western blotting with anti-phosphotyrosine mAb 4G10 (upper row). The blots were stripped and subsequently immunoblotted with the respective Abs (lower row). The migration positions of p56 lck , ZAP70, or p59 fyn are indicated by arrows.

Following CD2 Stimulation There Are Predominantly c-Jun/ c-Fos Heterodimers Binding to the AP-1 Consensus Site-The
IL-2 promoter contains several binding sites for transcription factors which control IL-2 gene expression including AP-1, NF-B, NF-AT, and the CD28RE (50,53). In the case of the AP-1 complex, the major kinase which serine phosphorylates Jun is JNK (52). Since JNK is activated in JCaM1.6s.S3 following CD2 stimulation (Fig. 5A), it was important to determine how this activation would affect AP-1 binding. As shown by EMSA (electromobility shift assay) in Fig. 6A, compared with CD3 stimulation, the CD2 induced Jun kinase activity in JCaM1.6s.S3 is accompanied by enhanced expression and binding of AP-1 complexes to the AP-1 consensus site. This interaction of AP-1 proteins with the AP-1 consensus site double-stranded oligonucleotide is specific for AP-1, as shown by competition experiments using an excess of unlabeled AP-1 oligonucleotide (Fig. 6B). In J77, on the other hand, AP-1 complex binding is similar between CD2 and CD3 stimulation (Fig. 6A).
The AP-1 complex consists of homo-or heterodimers of proteins of the Jun and Fos family (54). It has been suggested that Jun/Fos heterodimers have a higher affinity for the AP-1 site and are more efficient in activating IL-2 gene expression (55). The Jun family contains at least three members, c-Jun, JunB, and JunD, while the Fos family includes c-Fos, Fra1, Fra2, and FosB (51,53). Following CD2 stimulation of JCaM1.6s.S3 both homodimers and heterodimers appear to bind to the AP-1 consensus site as shown by supershift of the AP-1 proteins with the respective Abs specific for JunD, c-Jun, and c-Fos (Fig. 7, A  and B). These AP-1 complexes do not contain JunB (data not shown), Fra1, Fra2, or FosB (Fig. 7B). Given that the supershift with anti-c-Jun and anti-c-Fos Abs results in a mobility shift of their respective complexes to the same position in the gel (Fig. 7A), we infer that c-Jun/c-Fos heterodimers are formed. In contrast, the different mobility of the complex in the supershift with anti-JunD Ab implies that JunD/JunD homodimers are present. CD3 stimulation is less efficient in AP-1 induction in JCaM1.6s.S3 (Fig. 6A). Moreover, with anti-CD2 stimulation the AP-1 complex is supershifted to the greatest extent by anti-c-Fos mAb. In contrast, with anti-CD3 mAb stimulation anti-JunD mAb effects the greatest shift. These results imply that the AP-1 dimer is primarily a complex consisting of JunD/JunD homodimers following anti-CD3 stimulation (Fig. 7A).
Although not shown, analysis of the other transcriptional factors involved in the regulation of IL-2 production including NF-AT, NF-B, and Oct-1 (50,53) showed similar inducible nuclear protein binding profiles to the NF-AT, NF-B, or Oct-1 consensus site in JCaM1.6s.S3 following CD2 or CD3 triggering; moreover, supershift analysis with Abs specific for c-Jun, JunD, or c-Fos revealed no difference in the composition of the complex binding to the NF-AT site upon either anti-CD2 or anti-CD3 stimulation. Similarly, Abs specific for the p50 or p65 subunit of the NF-B complex did not indicate differential effects of these stimulatory pathways as judged by supershift analysis of nuclear proteins binding to the NK-B consensus site (data not shown).
To determine if resting T cells would respond to CD2 stimulation similarly in terms of JNK activation and binding of AP-1 proteins, we performed corresponding experiments in freshly isolated peripheral T cells. JNK is activated via both CD2 and to a lesser extent, via CD3 (Fig. 5B), suggesting, as discussed below, an important role for the CD2 signaling pathway in T cell co-stimulation. In addition, CD2 but not CD3 stimulation also efficiently leads to the activation and binding of AP-1 proteins to the AP-1 consensus site (Fig. 6C).
The Tyrosine Kinase Pyk2 Is Activated following Stimulation via CD2 in the Absence of p56 lck -Recently, Pyk2 a tyrosine kinase homologous to the focal adhesion kinase (FAK) has been identified and linked to the JNK pathway, (39). Pyk2 activation, like JNK activation, requires a strong Ca 2ϩ signal such as that provided by CD2 stimulation in JCaM1.6s.S3. We investigated the activation of Pyk2 in JcaM1.6s.S3 following anti-CD2 or anti-CD3 mAb triggering by Western blot analysis and in vitro kinase assay. Pyk2 is phosphorylated in response to both CD2 and CD3 triggering (Fig. 8A), but activated only after CD2 stimulation as judged by in vitro autophosphorylation (Fig. 8C). Equivalent amounts of Pyk2 were precipitated as shown by sequential immunoblotting of the phosphotyrosine blot (Fig. 8A) with a Pyk2-specific rabbit antiserum (Fig. 8B). DISCUSSION In this study, we identify a p56 lck -independent CD2 signaling pathway capable of inducing IL-2 production. Fig. 9 offers a schematic view of p56 lck -independent as well as -dependent CD2 activation pathways (37,39,56,57). The latter is largely indistinguishable from the p56 lck -dependent TCR pathway except that, unlike the TCR pathway, it is associated with weak CD3 and ZAP70 phosphorylation. In JCaM1.6 cells, only the p56 lck -independent CD2 pathway is operative and, in distinction to the TCR pathway, does not involve p56 lck , ZAP70, or MAPK. Instead, CD2 triggering activates JNK followed by the binding of c-Jun/c-Fos heterodimers to the AP-1 consensus site.
Prior studies of TCR-based signaling showed that, in the absence of p56 lck , TCR triggered signals were abrogated (42). In contrast, as shown here, activation via CD2 in the p56 lckdeficient Jurkat variant JCaM1.6s is not disrupted. CD2 stimulation leads to the tyrosine phosphorylation and activation of PLC-␥1, a strong rise in intracellular free Ca 2ϩ levels, and subsequent production of IL-2. Furthermore, CD2 signaling stimulates JNK, which leads to the activation of transcription factors of the Jun family. Phosphorylated c-Jun/c-Fos heterodimers bind to the AP-1 consensus site, leading to the induction of IL-2 transcription.
Although CD2 signaling in T cells requires the presence of the CD3 chain (27)(28)(29)(30)(31), the precise molecular basis by which cell surface immunoreceptor tyrosine-based activation motif containing TCR subunits permit CD2 signaling remains uncertain. The functional involvement of the CD3 chain in T cells, and the stimulation of phosphatidylinositol turnover (46,58) shared by TCR and CD2 pathways indicates the usage of a common pathway following CD2 and TCR signaling. Indeed, a complex involving CD2, p59 fyn , and CD3 has been shown to exist in T lymphocytes (35). Nevertheless, certain differences among the pathways are also evident. For example, T cell activation via CD2, in contrast to via the TCR, is not associated with Syk phosphorylation and involves minimal if any phosphorylation of ZAP-70 or CD3 (25,27,59). Additionally, the phosphorylation kinetics and cellular substrates are distinct in these two pathways (25,27). Recently, the Tec family kinase ITK has been shown to be activated upon stimulation via CD2 (60). This ITK activation requires the presence of functional p56 lck but is independent of the surface expression of CD3 (60). Moreover, CD2 plays important roles in reversing anergy and augmenting IL-12 responsiveness; neither activities are TCR regulated (15,17). CD2 signaling is mediated by its proline-rich intracytoplasmic domain to which SH3 domains of non-receptor kinases can bind (23,33). Recent studies indicate that p56 lck and p59 fyn are able to bind to CD2 and hence are implicated in CD2 signaling (34,35,61).
p56 lck is present in thymocytes and mature T cells and is involved in signaling via the TCR (62). T cells from mice bearing a disrupted p56 lck gene (63), mutant Jurkat cells lacking functional p56 lck (JCaM1.6) (42), or CTLL-2 cells that lack p56 lck expression show defects in TCR-mediated responses (62). Upon TCR stimulation p56 lck associates with and phosphorylates ZAP70 and it has been speculated that ZAP70 may, therefore be, at least in part, responsible for bringing p56 lck into the signaling complex (64,65). Notwithstanding, other experiments employing mutant CD3 immunoreceptor tyrosinebased activation motif sequences demonstrate that p56 lck can associate with the TCR in the absence of ZAP70 (66). Nevertheless, p56 lck is thought to be responsible for the activation of ZAP70 enzymatic function (48,65).
The tyrosine kinase ZAP70 is a key molecule in TCR signaling, but can in some systems be replaced by p72 Syk (67,68). We and others (25,59) have shown that following CD2 stimulation ZAP70 is only weakly if at all tyrosine phosphorylated or activated in J77. In JCaM1.6s no tyrosine phosphorylation or activation of ZAP70 after either CD2 or CD3 triggering was observed ( Fig. 3B and data not shown). Furthermore, JCaM1.6 and its parental Jurkat line E6 are deficient in Syk expression (69) and there is no evidence for the presence or phosphorylation of Syk in JCaM1.6s (data not shown). We, therefore, conclude that in JCaM1.6s, CD2-triggered IL-2 production does not involve tyrosine kinases of the Syk family.
Events downstream of protein-tyrosine kinase activation include the tyrosine phosphorylation and activation of PLC-␥1, which then leads to an increase in intracellular free Ca 2ϩ concentrations via the generation of inositol trisphosphate (70). In p56 lck expressing T cells, TCR stimulation is followed by an initial high transient Ca 2ϩ peak and a lower amplitude but sustained plateau phase (41,44,46). The initial rise of Ca 2ϩ is caused by the inositol trisphosphate-mediated release of Ca 2ϩ from the endoplasmatic reticulum (45) but is not sufficient for proliferation or IL-2 gene expression (41,71). Rather the prolonged elevation of intracellular Ca 2ϩ due to the Ca 2ϩ influx from extracellular sources is the critical component of the Ca 2ϩ signal (41,71). The mechanism by which Ca 2ϩ enters T cells from extracellular stores in not well understood (72). However, in JCaM1.6s, anti-CD3 stimulation leads only to a short duration Ca 2ϩ increase in the absence of a sustained high Ca 2ϩ rise, suggesting that p56 lck is involved, either directly or indirectly, in TCR-mediated Ca 2ϩ influx from extracellular stores. Inositol trisphosphate is generated by hydrolysis of phosphatidylinositol bisphosphate by PLC-␥1. As reported previously, JCaM1.6 cells fail to show production of inositol phosphates after TCR triggering (73). Consistent with this finding, PLC-␥1, whose catalytic activity is strongly enhanced by tyrosine phosphorylation (74), is only weakly phosphorylated upon TCR triggering in JCaM1.6s (Fig. 2B). In contrast to CD3 stimulation, CD2 stimulation leads to a clear tyrosine phosphorylation of PLC-␥1 (Fig. 2B) and to a strong and long lasting rise in intracellular free Ca 2ϩ levels in JCaM1.6s ( Fig. 2A). Although Hubert et al. (59) failed to detect anti-CD2 mAb induced PLC␥ phosphorylation in JCaM1 cells, a difference in the surface expression of CD3 and CD2 on JCaM1 versus the sorted JCaM1.6s cells herein may explain this discrepancy.
TCR stimulation events downstream of ZAP70 include Shc phosphorylation, the activation of Sos, and subsequently of Ras leading to the activation of the MAPK pathway and finally to IL-2 production (56). This tyrosine kinase activation following TCR engagement has been shown to be dependent on the presence of functional p56 lck (36). Surprisingly, therefore, we observed that although the MAPK pathway is non-functional after CD2 stimulation as well as CD3 stimulation in JCaM1.6s.S3, the ability of CD2 mAbs to induce IL-2 produc-FIG. 7. Supershift analysis of nuclear proteins of JCaM1.6s.S3 binding to the AP-1 consensus site. JCaM1.6s.S3 cells were unstimulated (Ϫ) or stimulated with either a combination of the anti-CD2 mAbs anti-T11 2 plus anti-T11 3 (␣CD2) or the anti-CD3⑀ mAb 2AD2A2 (␣CD3) and nuclear extracts were subjected to EMSAs with [␥-32 P]ATPlabeled oligonucleotides specific for the AP-1 consensus site in the presence of Abs specific for members of the Jun family (A) or the Fos family (B). The complexes were analyzed on 4% gels followed by autoradiography. The migration position of the AP-1 complex, the supershifted complexes and the free oligo are indicated by arrows.
FIG. 8. Pyk2 is activated following CD2 stimulation. JCaM1.6s.S3 cells, either unstimulated (Ϫ) or stimulated for 3 min with either a combination of the anti-CD2 mAbs anti-T11 2 plus anti-T11 3 (␣CD2) or the anti-CD3⑀ mAb 2AD2A2 (␣CD3). Cell lysates were immunoprecipitaed with Pyk2 Ab and were either subjected to sequential Western blot analysis with anti-phosphotyrosine-specific mAb 4G10 (A) and an antiserum specific for Pyk2 (B) or to in vitro kinase analysis (C). The position of Pyk2 at ϳ120 kDa is indicated by an arrow. tion was not compromised. These observations strongly indicate that CD2 stimulation can activate signaling pathways distinct from CD3 stimulation.
Pyk2 was recently identified as a tyrosine kinase, which is involved in Jun kinase activation and associates with the adapter protein Grb2 (39). By forming a complex containing Pyk2 and the GDP-GTP exchange factor Vav (75), Grb2 can link Pyk2 to the GTP-binding protein Rac, whose activation finally leads to the activation of JNK (57). In our study, we show that, following both CD2 and CD3 stimulation, Pyk2 is tyrosine phosphorylated in JCaM1.6s.S3 cells, but only CD2 triggering activates Pyk2 enzymatically, providing a possible link to a stimulation pathway which leads to JNK activation.
JNK is important in the regulation of the IL-2 promoter since JNK activation correlates with IL-2 production (38,51,52,76). IL-2 promoter regulation involves several transcription factors some of which are JNK sensitive (50,51,(77)(78)(79)(80). The AP-1 transcription factor can bind to the IL-2 promoter either directly at the AP-1 site (81,82) or together with NF-AT or Oct at their respective binding sites (83)(84)(85)(86). The AP-1 complex is composed of proteins of the Jun and Fos family (51). JNK, a Ca 2ϩ -sensitive serine/threonine kinase of the MAP kinase family, is critically involved in the post-transcriptional stimulation of AP-1 activity by phosphorylating the activation domain of c-Jun. Su et al. (38) reported that two signals are necessary to efficiently activate JNK, including different combinations of A23187, TPA, anti-CD3 mAb, or anti-CD28 mAb. This dual activation requirement and the reported Ca 2ϩ sensitivity distinguishes JNK from other members of the MAPK/JNK family.
The involvement of CD2 stimulation in the activation of JNK has not yet been previously investigated. In both J77 and JCaM1.6s.S3, JNK is activated following stimulation via CD2 (Fig. 5A), suggesting that CD2 utilizes a p56 lck independent pathway which leads to the activation of JNK. The strong and long lasting Ca 2ϩ mobilization observed in JCaM1.6s.S3 following CD2 stimulation might sensitize JNK to p56 lck independent signals involved in JNK activation (38). Furthermore, in the absence of p56 lck , anti-CD2 stimulation is accompanied by the enhanced binding of the c-Jun/c-Fos heterodimers to the consensus AP-1 sequence leading to IL-2 production. A previous study performed by stimulating Jurkat cells with superantigen pulsed HLA-DR transfectants revealed differences in the composition of the NF-AT⅐AP-1 complexes following costimulation with either LFA-3 or B7 (87). These differences were due to the dimerization of JunD with different members of the Fos family (Fra1 and Fra2), indicating a selective induction of certain nuclear factors depending on the costimulatory pathway. In our current study, we observed no significant difference in the binding of nuclear proteins to the NF-AT consensus site upon anti-CD2 versus anti-CD3 stimulated JCaM1.6s.S3. Since the MAPK pathway, known to be involved in Fos transcription, is non-functional in JCaM1.6, the lack of Fra1 and Fra2 Fos family members complexed to the AP-1 site in JCaM1.6s.S3 is perhaps not unexpected. In addition, the coordinate action of TCR engagement and the B7 versus LFA-3 costimulatory signal could differentially induce Fos family proteins. In the study by Parra et al. (87), AP-1 and NF-B complexes binding to their respective site in the IL-2 promoter revealed no differences after either type of costimulation. By stimulation via CD2 alone, we found not only JunD but also c-Jun induced and heterodimerizing with c-Fos in JCaM1.6s.S3 (Fig. 7A). Since the AP-1 site in the IL-2 promoter is a relatively low affinity site (83), Fos-Jun heterodimers, which are more effective in DNA binding and transactivation (55), might be required for optimal activity. Additionally, c-Jun and c-Fos in contrast to JunB, Fra1, and Fra2 are more efficient transactivators (55,88,89). The function of JunD in IL-2 gene transcription is not defined yet.
Anergy, a state of T cell unresponsiveness to antigenic challenge, which is induced by TCR stimulation in the absence of the CD28 costimulatory signal (90), is accompanied by preferential induction of the inhibitory p50-p50 NF-B homodimer and a reduced binding of AP-1 to the IL-2 promoter (91, 92). Recently, it was shown in an alloreactive system that alloantigen stimulation induced T cell anergy can be reversed in those cells after culture in IL-2 for 7 days only by costimulation with CD58 (17). This ability of CD2 stimulation to reverse anergy is unique and distinct from costimulatory molecules such as CD28 (17). Our observation that CD2 stimulation can activate JNK and leads to the induction and binding of c-Jun/ c-Fos heterodimers to the AP-1 site might provide a basis for the role CD2 plays in regulating anergy. This possibly remains to be investigated in anergized T cells.
The data presented in this study provide evidence for a CD2 signaling pathway distinct from that of the TCR and capable of inducing IL-2 production in the absence of p56 lck . This CD2 stimulation induced signaling cascade does not involve either FIG. 9. Signal transduction via CD2. p56 lck -dependent and independent activation pathways following CD2 stimulation are based on prior results (37,39,56,57,75) and the current studies. In JCaM1.6, only the p56 lck independent signal transduction pathway is functional leading via JNK and subsequent c-Jun activation to the binding of the AP-1 complex to the IL-2 promoter and the initiation of IL-2 transcription. Both pathways apparently activate PLC-␥1 leading to the generation of a rise in intracellular free Ca 2ϩ , resulting in a calmodulin-dependent activation of calcineurin. The latter dephosphorylates NF-AT resulting in its nuclear translocation.
p56 lck or ZAP70, key molecules in T cell activation via the TCR, and can also be initiated by CD2 triggering in peripheral T cells. CD2 activates Pyk2 and undoubtedly other kinases, and subsequently stimulates JNK independent of p56 lck . The precise definition of the intermediate steps in this activation cascade will now be of interest to determine.