Association of p59 fyn with the T Lymphocyte Costimulatory Receptor CD2

Human CD2 is a 50–55-kDa cell surface receptor specifically expressed on the surface of T lymphocytes and NK cells. Stimulation of human peripheral blood T cells with mitogenic pairs of anti-CD2 monoclonal antibodies (mAbs) is sufficient to induce interleukin-2 production and T cell proliferation in the absence of an antigen-specific signal through the T cell receptor. CD2 has been shown previously to associate physically with the Src family protein-tyrosine kinases p56 lck and p59 fyn . We now report that stimulation of T cells with mitogenic pairs of anti-CD2 mAbs enhanced the association of the Fyn polypeptide with the CD2 complex, whereas stimulation with single anti-CD2 mAb had minimal effect. Using glutathione S-transferase (GST) fusion proteins, we found that CD2 bound to the Src homology (SH) 3 domain of Fyn. Interestingly, the CD2-Fyn association was negatively regulated by the Fyn SH2 domain; CD2 bound poorly to GST fusion proteins expressing both the SH2 and SH3 domains of Fyn. However, the inhibitory effect of the Fyn SH2 domain on binding of the Fyn SH3 domain to CD2 was relieved by peptides containing a phosphorylated YEEI sequence that bound directly to the Fyn SH2 domain. In addition, we found that the ability of the Fyn SH2 domain to precipitate tyrosine-phosphorylated proteins, including the CD3ζ chain, was enhanced after T cell stimulation with mitogenic pairs of CD2 mAbs. Finally, overexpression of a mutated Fyn molecule, in which the ability of the Fyn SH2 domain to bind phosphotyrosine-containing proteins was abrogated, inhibited CD2-induced transcriptional activation of the nuclear factor of activated T cells (NFAT), suggesting a functional involvement of the Fyn SH2 domain in CD2-induced T cell signaling. We thus propose that stimulation through the CD2 receptor leads to the tyrosine phosphorylation of intracellular proteins, including CD3ζ itself, which in turn bind to the Fyn-SH2 domain, allowing the direct association of the Fyn SH3 domain with CD2 and the initiation of downstream signaling events.

The activation of T lymphocytes is initiated by engagement of the T cell receptor (TCR) 1 by peptide embedded in MHC molecules expressed on the surface of antigen-presenting cells (APCs) (1,2). In addition to the TCR-CD3 complex, T cells also express cell surface coreceptors such as CD4, CD8, CD28, LFA-1, and CD2 that bind to their cognate ligands on APCs. The binding of the co-receptors with their ligands appears to modulate the avidity of T cell/APC interactions. Moreover, coreceptor ligation may also initiate intracellular signaling pathways that regulate T cell activation (3)(4)(5)(6).
CD2 is a 50 -55-kDa glycoprotein expressed on a majority of thymocytes, T cells, and NK cells. A number of ligands of human CD2 have now been identified including CD58, CD48, CD59, and a novel carbohydrate structure associated with CD15 (7)(8)(9)(10)(11)(12). CD2 engagement has been shown either to synergize with or to inhibit antigen-induced T cell responses depending, in part, upon the specific experimental system (13)(14)(15)(16). Unlike many other coreceptors, stimulation of T cells with certain pairs of anti-CD2 mAbs can lead to IL-2 production and T cell proliferation in the absence of direct ligation of the TCR-CD3 complex (17). However, it appears that optimal CD2induced T cell activation requires the expression of components of the CD3 complex such as CD3, even though direct engagement of the TCR-CD3 complex is not necessary (18). Recent reports have shown that CD2 ligation is capable of reversing T cell anergy (19) and regulating T cell responsiveness to IL-12 (20), further implicating CD2 as a critical regulator of the immune response.
The molecular determinants of CD2-mediated T cell activation are currently under investigation. A number of proteins, including CD3⑀ and CD3 chains, phosphoprotein p29/30, Lck, Fyn, CD4, CD5, CD8, p85 subunit of phosphatidylinositol 3-kinase, CD45, and ␣ and ␤ tubulins have been shown to coimmunoprecipitate with CD2 (21)(22)(23)(24)(25)(26)(27)(28). The functional outcome and biological significance of these protein associations remain to be determined. The 116-amino acid cytoplasmic tail of CD2 contains no tyrosine that upon phosphorylation could mediate interaction with SH2 domain-containing signaling elements. Mutational and deletional analyses of human CD2 have shown that repeated proline-rich domains, potentially able to mediate * This work was supported by the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ ‡ To whom correspondence should be addressed. Present address: NHLBI, National Institutes of Health, Bldg. 10 binding to SH3-containing proteins, are important for CD2 signaling (29 -31). In this report, we have examined the association of CD2 with Fyn, a Src family protein-tyrosine kinase (PTK). We find that stimulation with mitogenic anti-CD2 mAbs increased the association of Fyn with CD2 and that it was the SH3 domain of Fyn that bound to the CD2 cytoplasmic domain. The Fyn SH2 domain was unable to bind to CD2 directly; nevertheless, it inhibited the binding of the Fyn SH3 domain to CD2. The inhibitory effect of the Fyn SH2 domain on Fyn SH3 binding to CD2 was able to be reversed by specific phosphotyrosine-containing peptides. In addition, the binding of the Fyn SH2 domain to tyrosine-phosphorylated proteins, including CD3 chains, was enhanced after stimulation with mitogenic combinations of anti-CD2 mAbs. Finally, we demonstrated that overexpression of a Fyn loss-of-function SH2 domain mutant, but not wild-type Fyn or a Fyn loss-of-function SH3 domain mutant, inhibited CD2-induced NFAT transcriptional activity. Taken together, our data suggest that the SH2 domain of Fyn regulates the CD2-Fyn association in an activation-dependent manner; multimolecular complexes of CD2-CD3-Fyn may thus quantitatively regulate the amount of Fyn polypeptide in the CD2 complex.

EXPERIMENTAL PROCEDURES
Cell Culture-The T cell leukemia cell line Jurkat clone J77 was a gift of K. Smith (Cornell University, New York, NY). Jurkat cells transformed with SV40 large tumor (T) antigen were also used in transient transfection experiments. Lymphocytes were grown in RPMI 1640 (Mediatech, Herndon, VA) supplemented with 10% heat-inactivated fetal bovine serum (Sigma), 100 units/ml penicillin (Life Technologies, Inc.), 100 g/ml streptomycin (Life Technologies, Inc.), 2 mM glutamine (Life Technologies, Inc.), 10 mM HEPES, pH 7.3, and 50 M 2-mercaptoethanol (Sigma), termed 10% fetal calf serum medium. Resting human peripheral blood T cells were isolated from normal volunteers by centrifugation through Ficoll (Organon Teknika, Durham, NC), plastic adherence, and nylon wool filtration. Contaminating red cells were lysed with Tris-buffered ammonium chloride. The purified human T cells were cultured in 10% fetal calf serum medium and used within 24 h of isolation.
cDNA Constructs and Fusion Proteins-The wild-type Fyn construct, a loss-of-function mutated Fyn construct in which the Arg was replaced with Lys at amino acid 176 in the Fyn SH2 domain (denoted Fyn SH2*), and a loss-of-function mutated Fyn construct in which the Pro was replaced with Val in at amino acid 134 in the Fyn SH3 domain (denoted Fyn SH3*) were the kind gift of R. Perlmutter (Merck-Research Institute, Rahway, NJ). Wild-type or mutated Fyn constructs were subcloned into pALTER vector (Promega, WI) and used for transfection into Jurkat cells. The GST fusion constructs containing different domains of Fyn have been described previously (33). Escherichia coli DH5␣ cells were transformed with PGEX2T.K containing the Fyn SH2 domain (termed GST-FynSH2), the Fyn SH3 domain (termed GST-FynSH3), the Fyn SH3 domain mutant (termed GST-FynSH3W119K) in which a tryptophan (W) at amino acid 119 was replaced by a lysine (K), and the Fyn SH3-SH2 domains (termed GST-FynSH3SH2).
Transformed E. coli were induced with 0.4 mM of isopropyl-1-thio-␤-D-galactopyranoside (Sigma). The cells were resuspended in Tris-buffered saline (150 mM NaCl, 10 mM Tris, pH 8.0) containing 1% Triton X-100 and lysed by sonication. The GST fusion proteins were affinitypurified using glutathione-Sepharose beads (Amersham Pharmacia Biotech). Immobilized proteins were aliquoted and stored as a 50% slurry (v/v) in Tris-buffered saline with 0.1% Triton X-100 at Ϫ80°C. In some experiments, fusion proteins were also eluted from the beads according to the manufacturer's instruction (Amersham Pharmacia Biotech). Cell Stimulation, Immunoprecipitations, and in Vitro Kinase Assays-Jurkat cells or human purified T cells (1-2 ϫ 10 7 as indicated) were resuspended in 0.5 ml of buffer A (RPMI 1640 supplemented with 100 units/ml penicillin, 100 g/ml streptomycin, 2 mM glutamine, 10 mM HEPES, pH 7.3), incubated with 0.5 l of T11 2 and 0.5 l of T11 3 ascites on ice for 15 min, and then transferred to 37°C for 5 min, unless otherwise indicated. The cells were lysed by addition of 0.5 ml of cold lysis buffer B (2% Brij 97, 150 mM NaCl, 25 mM Tris, pH 7.5, 2 mM EDTA, 2 mM Na 3 VO 4 , 20 g/ml leupeptin, 20 g/ml aprotinin, and 2 mM phenylmethylsulfonyl fluoride) on ice for 20 min. Lysates were clarified by centrifugation at 15,000 ϫ g for 10 min at 4°C. Equivalent amounts of mAbs were added to unstimulated cells after lysis to normalize the final antibody concentrations.
In CD2 immunoprecipitations, 20 l of 1:1 ratio of protein A:protein G slurry and 0.5 l of T11 2 and 0.5 l of T11 3 ascites were added together to the clarified cell lysates and incubated at 4°C for 2 h. The beads were then washed three times with washing buffer C (0.1% Brij 97, 150 mM NaCl, 25 mM Tris, pH 7.5, 1 mM Na 3 VO 4 ). In the in vitro kinase reaction, the washed beads were incubated in 50 l of kinase buffer D (10 mM MnCl 2 , 5 mM p-nitrophenyl phosphate, 25 mM Hepes, pH 7.5, and 10 Ci of [␥-32 P]ATP) for 15 min at room temperature. Kinase reactions were stopped by addition of 2ϫ SDS sample buffer. Phosphoproteins were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to polyvinylidene difluoride (PVDF), and visualized by autoradiography. In reimmunoprecipitation experiments, co-immunoprecipitated proteins were dissociated by boiling in 50 l of 2% SDS for 10 min, after which the beads were separated and removed by filtration through microspin tubes (Costar, MA). Dissociated proteins were renatured by the addition of 900 l of Tris-buffered saline with 1% Nonidet P-40, reimmunoprecipitated with the indicated antibodies, separated by SDS-PAGE, transferred to PVDF membranes, and then visualized by autoradiography or immunoblotted as indicated.
In the experiments using immobilized GST fusion proteins, fusion proteins previously bound to glutathione-Sepharose beads were added to the clarified cell lysates and incubated at 4°C for 2 h. The beads were then washed three times with washing buffer C and subjected to deglycosylation and Western blot analysis, as detailed below. In the peptide binding experiments, both GST fusion proteins and cell lysates were incubated separately with 50 M -YEEI-or -pYEEI-peptides for 30 min at 4°C, prior to GST precipitation.
Protein Deglycosylation and Western Blotting-CD2 immunoprecipitates or GST fusion protein precipitates were boiled in 30 l of denaturing buffer E (100 mM Tris, pH 8.0, 10 mM EDTA, 0.5% SDS, 1% ␤-mercaptoethanol), renatured in 1% Nonidet P-40 and 50 M Na 3 PO 4 and then incubated with 500 units of PNGase F (New England Biolabs) at 37°C for 2 h. Deglycosylated proteins were boiled in SDS sample buffer, separated by SDS-PAGE, transferred to PVDF, and probed with primary antibodies followed by horseradish peroxidase-conjugated secondary antibodies. Polypeptides recognized in the Western blot were detected using the ECL method according to manufacturer's instructions (Amersham Pharmacia Biotech).
Transient Transfection and Luciferase Assays-SV40 large tumor (T) antigen-transformed Jurkat cells (1 ϫ 10 7 ) were incubated with 20 g of vector, wild-type or mutated Fyn constructs as indicated, with 5 g of a reporter plasmid p3xNFAT-luc (34), carrying the luciferase gene driven by three tandem repeats of the distal NFAT sequences derived from the IL-2 promoter, for 15 min at room temperature. Cells were then electroporated at 250 V, 800 microfarads (Life Technologies, Inc.). After electroporation, the cells were transferred to 10% fetal calf serum medium and incubated at 37°C for 12 h. Transfected cells were stimulated for 6 h with 1:1,000 dilution of T11 2 and T11 3 , a mitogenic pair of anti-CD2 mAbs, or with PMA (10 ng/ml) plus ionomycin (2 M). Cells were washed with PBS, and samples were prepared using the Enhanced Luciferase Kit (Analytic Luminescent Laboratory, San Diego, CA), according to the manufacturer's instructions. The relative luciferase units are presented as the percentage of the maximal stimulation for each transfection condition induced by PMA plus ionomycin.

Stimulation with Mitogenic Pairs of Anti-CD2 mAbs Increased the Association of CD2 with Phosphorylated Lck and
Fyn-The Src family tyrosine kinases Lck and Fyn have been shown to associate with CD2 (21), yet how these key proteins are regulated in CD2-dependent signaling events remains to be established. We examined whether extracellular ligation of CD2 altered the in vitro kinase activity of CD2-associated intracellular signaling proteins including Lck and Fyn. Different cell surface receptor complexes were immunoprecipitated from Jurkat T cells. The immune complexes were then incubated in kinase buffer containing [␥-32 P]ATP, and the resulting in vitro phosphorylated proteins were separated by SDS-PAGE, transferred to PVDF membrane, and visualized by autoradiography ( Fig. 1). Several phosphorylated proteins were present in CD2 immunoprecipitates prepared from resting Jurkat cells including prominent phosphoproteins in the 50 -60-kDa size range (Fig. 1, lane 4). A number of phosphoproteins were also found in CD3 immune complexes; no apparent kinase activity was detected in association with MHC class I molecules or CD43 (Fig. 1, lanes 1 and 3). Stimulation with a mitogenic combination of CD2 mAbs (T11 2 ϩ T11 3 ) increased the total amount of CD2-associated kinase activity (Fig. 1, lane 5). There was enhanced phosphorylation of a number of proteins, including proteins migrating at 28, 40, 50 -60, 120, and 150 kDa.
When comparable studies were performed using purified human T cells, the phosphoprotein profiles from resting and anti-CD2 stimulated cells were somewhat different than those of Jurkat cells. Nevertheless, CD2 stimulation increased the phosphorylation of CD2-associated proteins in the range of 50 -60 kDa (Fig. 2, lanes 1 and 2). We performed reimmunoprecipitation experiments to determine if these phosphopro-teins included Lck and Fyn, which migrate at a similar size range. Equivalent amounts of CD2 protein were immunoprecipitated from human purified T cells without or with prestimulation at 37°C with the mitogenic combination of anti-CD2 mAbs T11 2 ϩ T11 3 . After in vitro phosphorylation (lanes 1 and 2), the CD2 complexes were dissociated by boiling and then subjected to reimmunoprecipitation using antisera directed against either Fyn or Lck. Ligation of CD2 led to a modest increase in the amount of phosphorylated Lck and to a pronounced increase in the amount of phosphorylated Fyn associated with CD2 (Fig. 2, lanes 5-8). We further characterized the latter interaction.
CD2-dependent Stimulation Quantitatively Increased the Amount of Fyn Associated with CD2-We determined whether the increase in phosphorylated Fyn associated with CD2 was secondary to quantitative changes in the amount of Fyn polypeptide recruited to the CD2 complex. Human T cells were stimulated with mitogenic combination of anti-CD2 mAbs for varying times at 37°C, after which CD2 was immunoprecipitated from cell lysates. The Fyn associated with CD2 was detected by Western blot (Fig. 3A, upper panel). Although minimal Fyn was associated with CD2 in the resting cells (Fig. 3A, lane 1), the amount of Fyn associated with CD2 increased after stimulation of the cells with mitogenic pairs of anti-CD2 mAbs (T11 2 ϩ T11 3 ) in a time-dependent fashion (Fig. 3A, lanes 2-5). The amount of Fyn associated with CD2 was maximal at about 5 min after stimulation (Fig. 3A, lane 4) and decreased by about 15 min (Fig. 3A, lane 5). Similar results were observed in experiments performed in Jurkat cells (data not shown). In contrast to mitogenic pairs of anti-CD2 mAbs stimulation, using either of the single anti-CD2 mAb T11 2 or T11 3 alone only marginally altered the amount of CD2-Fyn association (Fig. 3B).
A trivial explanation for the above result could be that different amounts of CD2 were precipitated from resting versus stimulated cells, or at different time points, but that the stoichiometry of the Fyn association to CD2 was unchanged. We therefore developed a procedure to establish the amount of precipitated CD2. CD2 is a highly glycosylated protein that migrates as a diffuse band of approximate 50 -60 kDa in SDS- PAGE and is indistinguishable from Ig heavy chains following Western blot (Fig. 3C, lane 1). However, after treatment of the precipitated material with PNGase F to remove N-linked carbohydrates residues, CD2 migrated as a distinct band of 40 kDa in SDS-PAGE. The deglycosylated CD2 was then detected by the anti-CD2 mAb 3G12, which recognizes the cytoplasmic tail of CD2 (Fig. 3C, lane 2). Quantitatively similar amounts of CD2 were precipitated at different time points following stimulation (Fig. 3, A and B, lower panel). Taken together, these results suggest that stimulation with mitogenic pairs of anti-CD2 mAbs led to a time-dependent increase in the amount of Fyn polypeptide associated with CD2.
The SH3 Domain, but Not the SH2 or the Amino-terminal Unique Domain, of Fyn Was Able to Bind to CD2-The Fyn molecule contains an amino-terminal unique domain (termed NT), an SH2 domain, an SH3 domain, and a catalytic kinase domain. The NT, SH2. and SH3 domains of Fyn mediate protein-protein interactions and it is believed that these interactions regulate the subcellular localization and substrate-binding characteristics of Fyn (35). GST fusion proteins containing the NT (termed GST-FynNT), SH2 (termed GST-FynSH2) or SH3 (termed GST-FynSH3) domains of Fyn were expressed in bacteria and purified on glutathione-Sepharose (33); their ability to bind to CD2 was tested (Fig. 4A). Neither GST-FynNT nor GST-FynSH2 was able to precipitate CD2 from either resting or CD2 stimulated cells (Fig. 4A, lanes 2-5). In contrast, GST-FynSH3 was able to precipitate CD2 from T cell lysates in an activation-independent manner (Fig. 4A, lanes 6 and 7). The binding of GST-FynSH3 to CD2 was demonstrated in unstimu-   6 and 7). CD2 was precipitated directly using mAb 9.1 and served as a positive control (lane 8). The precipitates were treated with PNGase F and analyzed by Western blot using the anti-CD2 mAb 3G12. B, unstimulated Jurkat T cells (lanes 2 and 4) or cells stimulated with anti-CD2 mAbs T11 2 and T11 3 (lanes 1, 3, and 5) were lysed and precipitated with glutathione beads bound GST, GST-FynSH3, or the Fyn SH3 domain mutant GST-FynSH3W119K, as described under "Experimental Procedures." The precipitates were treated with PNGase F, separated by SDS-PAGE, and analyzed by Western blot using the anti-CD2 mAb 3G12. lated cells; binding was not increased after stimulating with mitogenic pairs of anti-CD2 mAbs. Furthermore, the specific binding of GST-FynSH3 and CD2 was abrogated by a point mutation in the ligand-binding pocket of the SH3 domain in which a tryptophan is replaced by a lysine at position 119 (GST-FynSH3W119K) (Fig. 4B, lanes 4 and 5). These results suggest that the binding of CD2 to Fyn was mediated through the Fyn SH3 domain, and that binding required Fyn SH3 sequences essential for binding to proline-rich sequence motifs (36,37).
The SH2 Domain of Fyn Negatively Regulated the Binding of the SH3 Domain to CD2-The association of CD2 with Fyn was increased in cells stimulated in vivo with mitogenic combinations of anti-CD2 mAbs (Fig. 3B). However, the association of the SH3 domain of Fyn with CD2 appeared unchanged after stimulation (Fig. 4A, lanes 6 and 7). In vitro binding of Fyn SH2 domain to the Fyn SH3 domain has been demonstrated (33); it has been suggested that interaction of the SH2 and SH3 domains might interfere with binding to other ligands (33). We therefore tested whether the binding of the Fyn SH3 domain to CD2 was regulated by other domains of Fyn, and, specifically, whether the Fyn SH2 domain inhibited the binding of the Fyn SH3 domain to CD2. GST fusion proteins containing both the SH2 and SH3 domains of Fyn (termed GST-FynSH3SH2) were used (Fig. 5A). The Fyn SH3SH2 domain bound poorly to CD2 compared with the Fyn SH3 domain alone (Fig. 5A, lower  panel, lane 2 and 3). In contrast, GST-FynSH3SH2 and GST-FynSH3 bound to the phosphoprotein Cbl (Fig. 5A, upper panel,  lanes 2 and 3), demonstrating the ability of the GST-FynSH3SH2 protein to interact with target proteins. These data suggest that the presence of the Fyn SH2 domain was able to inhibit the binding of the Fyn SH3 domain to CD2. In addition, the inhibitory effect of the Fyn SH2 domain on Fyn SH3 was not universal because the Fyn SH3SH2 domain was able to precipitate other polypeptides such as Cbl.
It has been suggested that the interaction of the Fyn SH2 and SH3 domains functionally interferes with binding to other ligands; an 11-mer phosphorylated peptide EPQpYEEIPIYL (termed -pYEEI-), a high affinity ligand of Fyn SH2 domain, has been shown to be able to abolish Fyn SH2-SH3 interactions (33). We therefore tested whether the same peptide was able to reverse the inhibitory effect of the Fyn SH2 domain on the binding of the Fyn SH3 domain to CD2 (Fig. 5B). As expected, the GST-FynSH3SH2 bound to CD2 poorly (Fig. 5B, lane 2); however, preincubation with the -pYEEI-peptide restored the ability of the GST-FynSH3SH2 domain to bind to CD2 (Fig. 5B,  lane 4). Phosphorylation of the tyrosine in the peptide was required to reverse the inhibition, because the unphosphorylated -YEEI-peptide had no effect (Fig. 5B, lane 3). Coomassie Blue staining demonstrated that similar amounts of the GST-FynSH3SH2 fusion proteins were used in all lanes (data not shown).
Stimulation via CD2 Enhanced the Association of the FynSH3SH2 Domain to Tyrosyl-phosphorylated Proteins, One of Which was the CD3 chain-We sought to demonstrate one practical example of a phosphorylated polypeptide able to bind to the Fyn SH2 domain and thereby potentially able to modify the ability of the Fyn SH3SH2 domain to bind to CD2. In unstimulated Jurkat cells, the GST-FynSH3SH2 domain was able to bind to a number of tyrosyl-phosphorylated proteins as detected by blotting with the anti-phosphotyrosine mAb 4G10 (Fig. 6A, lane 3); stimulation of the cells with anti-CD2 mAbs quantitatively increased the tyrosyl-phosphorylated proteins detected. GST alone was unable to bind to any detectable phosphorylated proteins (Fig. 6A, lanes 1 and 2). Again, Coomassie Blue staining of the membrane demonstrated that com-parable amounts of GST fusion proteins were used in the different lanes (data not shown). A tyrosyl-phosphorylated protein of approximately 20 kDa was able to bind to the SH2 domain but not to the SH3 domain of Fyn (Fig. 6B, lanes 3-6); the molecular weight approximated that of CD3. Because CD3 chains have been shown to associate with CD2 (24), we tested whether the 20-kDa protein that bound to Fyn SH3SH2 was CD3. Proteins bound to the FynSH3SH2 domain were dissociated from the complex by boiling in SDS buffer and subjected to re-immunoprecipitation using anti-CD3 antibodies (Fig. 6C, lanes 1 and 2) or control (Fig. 6C, lanes 3 and 4). CD3 was present in the Fyn SH3SH2 complex in both unstimulated and stimulated cells. However, the amount of CD3 that bound to FynSH3SH2 was significantly increased upon anti-CD2 stimulation (Fig. 6C, lanes 1 and 2).
Overexpression of Fyn Mutant with Loss-of-function Mutation in the SH2 Domain Inhibited CD2-induced NFAT Transcriptional Activity-We have shown that CD2 is able to asso- ciate with the Fyn SH3 domain and that Fyn SH2 domain negatively regulates CD2-Fyn SH3 association. We reasoned that a loss-of-function Fyn SH2 mutant containing an intact SH3 domain would compete with endogenous Fyn for binding to CD2 (and other ligands), and may therefore interfere with CD2-induced signal transduction. To test the functional involvement of Fyn in CD2-mediated signaling, we transfected Jurkat cells transiently with wild-type Fyn or a mutated Fyn construct containing point mutations in either the SH2 (Fyn SH2*) or SH3 domain (Fyn SH3*) that abolished the ability to bind to phosphotyrosine-containing or proline-rich proteins, respectively. Jurkat cells were cotransfected with the reporter plasmid p3xNFATluc, carrying the luciferase gene driven by three tandem repeats of the NFAT sequences derived from the distal IL-2 promoter, to monitor NFAT transcriptional activation (Fig. 7). The expression of wild-type Fyn and mutant Fyn molecules were comparable as detected by Western blot (data not shown). In the absence of stimulation, the basal level of NFAT activity was low in transfectants expressing vector alone. Overexpression of wild-type Fyn, Fyn SH3* (and, to a lesser extent, Fyn SH2*) enhanced the basal level NFAT activity, albeit minimally. Mitogenic pairs of anti-CD2 mAbs greatly enhanced NFAT-driven transcription. At optimal concentrations of anti-CD2 stimulation, overexpression of wild-type Fyn or Fyn SH3* induced comparable NFAT transcription as vector-alone transfection. Overexpression of Fyn SH2*, however, inhibited CD2-induced NFAT transcriptional activity, suggesting that the Fyn SH2 domain was functionally important in CD2-induced signal transduction pathways. Because the kinase domains were intact in wild-type and mutated Fyn molecules, the inhibition by Fyn SH2* of CD2-mediated signaling FIG. 6. FynSH3SH2 bound to increased amount of tyrosylphosphorylated proteins after stimulation, one of which was CD3 chain. A, unstimulated Jurkat cells (lanes 1 and 3) or cells stimulated with anti-CD2 mAbs T11 2 and T11 3 (lanes 2 and 4) for 5 min were lysed and then precipitated with glutathione bead-bound GST (lanes 1 and 2), or GST-FynSH3SH2 fusion proteins (lane 3 and 4). The associated proteins were separated by SDS-PAGE and analyzed by Western blot using the anti-phosphotyrosine mAb 4G10. B, unstimulated Jurkat cells (lanes 1, 3, 5, 7, and 9) or cells stimulated with anti-CD2 mAbs T11 2 and T11 3 for 5 min (lanes 2, 4, 6, 8, and 10) were lysed and then precipitated with glutathione bead-bound GST (lanes 1  and 2), GST-FynSH2 (lanes 3 and 4), GST-FynSH3 (lanes 5 and 6), or GST-FynSH3SH2 fusion proteins (lanes 7 and 8). CD3 were immunoprecipitated using anti-CD3 mAb and served as a positive control (lanes 9 and 10). The precipitates were separated by SDS-PAGE and analyzed with Western blot using anti-phosphotyrosine mAb 4G10. C, unstimulated Jurkat cells (lanes 1 and 3) or cells stimulated with anti-CD2 mAb T11 2 and T11 3 (lanes 2 and 4) were lysed and then precipitated with glutathione bead-bound GST-FynSH3SH2 fusion proteins as described in A. After precipitation, the beads were washed and boiled in 2% SDS buffer. Dissociated proteins were re-immunoprecipitated with protein A plus anti-CD3 antibody (lanes 1 and 2)  A reporter plasmid p3xNFATluc, carrying the luciferase gene driven by three tandem repeats of the distal NFAT sequences derived from the IL-2 promoter, was co-transfected into Jurkat cells to monitor the transcriptional activity of NFAT, as described under "Experimental Procedures." Twelve hours after transfection, 10 6 cells were left unstimulated, stimulated with mitogenic pairs of anti-CD2 mAbs, or stimulated with PMA (10 ng/ml) plus ionomycin (2 M) for 6 h. Cells were then washed and then lysed, and the soluble extract was assayed for luciferase activity. The relative luciferase units were presented for each transfection condition as the percentage of maximal stimulation induced by PMA plus ionomycin.
A alone (lanes 3 and 4). Re-immunoprecipitated proteins were separated by SDS-PAGE and analyzed by Western blot using mAb 4G10 to detect tyrosyl-phosphorylated proteins.
was mediated, at least in part, through the non-catalytic domain of Fyn. DISCUSSION In this report, we have investigated the association of CD2 with Src family PTK, and, specifically, the molecular regulation of Fyn binding to CD2. All Src family members share a similar NH 2 -terminal myristoylated domain that permits membrane localization, an SH3 domain that binds to proline rich sequences, an SH2 domain that interacts with tyrosylphosphorylated proteins, and a COOH-terminal kinase domain (35). Although the exact substrates of Src PTK are unknown, a number of intracellular proteins such as rasGAP, the p85 subunit of phosphatidylinositol 3-kinase, Syk, ZAP-70, and Cbl have been shown to associate with Src PTK in T cells (35,38,39). T cells lacking the expression of Lck (40) or expressing a dominant negative form of Fyn (41) are defective in response to TCR stimulation. Furthermore, the involvement of Src family PTK in CD2-mediated signaling has been suggested in many studies. Lck and Fyn have both been reported to be associated with CD2 in immunoprecipitation studies (21,25). CD2 signaling is defective in the Lck-negative Jurkat variant cell line JCam.1 (42); CD2-dependent signaling is also defective in cell lines deficient in expression of CD45 (43), a transmembrane protein-tyrosine phosphatase believed to regulate the kinase activities of Src family members (44). Finally, Fyn has been shown to colocalize with CD2 in immunofluorescence studies (45).
We and others have shown that Fyn is able to associate physically with CD2 (Fig. 2, and Ref. 24). We have demonstrated here that the binding of CD2 to Fyn was mediated by the SH3 domain of Fyn and required the specific amino acids within the Fyn SH3 domain that have been shown to mediate binding to proline-rich ligands (Fig. 4, A and B). The human CD2 cytoplasmic tail has five different proline-rich sequences; these regions are largely conserved among human, rat, and mouse species (46). Rat CD2 has been reported to bind to Lck; binding of Lck to CD2 was shown to be mediated by the SH3 domain of Lck (31), but the regulation of Lck SH3 binding was not further explored. It will be important to determine the specific sequences of CD2 involved in SH3 binding of Fyn in comparison to Lck and perhaps to other signaling proteins that cooperatively participate in CD2 signaling and to correlate these structural studies with function. These studies are currently under way.
The CD2-Fyn association was dramatically increased in vivo in cells stimulated with mitogenic combinations of anti-CD2 mAbs compared with unstimulated cells or cells stimulated with a single anti-CD2 antibody alone (Fig. 3B). However, by in vitro analysis using GST fusion proteins, the association of the SH3 domain of Fyn with CD2 appeared robust but unchanged after stimulation. We demonstrated further that the Fyn SH2 domain negatively regulated the Fyn SH3 domain binding to CD2 (Fig. 5A). Regulation of SH3 domain interaction to ligand has been suggested previously (33,47,48). The initial crystallographic structure of the minimal Lck SH3-SH2 domain revealed that the SH2 and SH3 domain of Lck could form intermolecular dimers (49). However, structural analysis of the entire Src molecule revealed that dimerization occurs not by intermolecular associations, but by intramolecular associations between the SH3 domain of Src and the linker region located between the SH2 and kinase domains (50). Intramolecular folding of Src renders the kinase inactive; this analysis left open the possibility that intermolecular SH2-SH3 interactions were possible following stimulation of kinase activity. The structural analysis of Hck, another member of the Src kinase family, also supports an intramolecular association (47). In addition, the multidimensional NMR study of Itk, a T cellspecific tyrosine kinase belonging to the Tec tyrosine kinase family, has demonstrated an intramolecular interaction between the Itk SH3 domain and the proline-rich sequence of Itk (51). Finally, direct interactions between isolated SH2 and SH3 domains of Fyn in vitro have also been reported (33), suggesting that the SH2-SH3 interaction may decrease the binding of the Fyn SH3 domain to a subset of unidentified ligands (33). Our observations support a role for CD2 as one of the "regulated" ligands of the Fyn SH3 domain. Occupancy of the SH2 domain with phosphorylated ligands following T cell activation may compete for Fyn SH2 binding and allow the binding of Fyn SH3 domain to CD2 and other proline-rich ligands (33). Our observation that the binding of Fyn SH3-SH2 domain to CD2 was restored upon treatment with tyrosyl-phosphorylated peptide (Fig. 5B) is consistent with the model. It is also worth noting that the inhibitory effect of the Fyn SH2 domain on the Fyn SH3 domain to bind to ligands is not universal; the binding of Fyn SH3 domain to Cbl was largely unchanged even in the presence of the Fyn SH2 domain (Fig. 5A). The affinity of Cbl toward the Fyn SH3 domain may be higher than that of CD2, allowing it to overcome the inhibitory effect of Fyn SH2 domain. Alternatively, the Fyn SH2 domain binding to tyrosylphosphorylated sequences on Cbl may help to relieve the Fyn SH2-dependent inhibition of Fyn SH3 domain binding (39).
It has been shown that CD2-mediated activation of T cells requires the CD3 chains or other CD3 complex components and that CD3 physically associates with CD2 (18,24). We have demonstrated increased binding of the Fyn SH2 domain to CD3 after anti-CD2 stimulation (Fig. 6, B and C), suggesting that the trimolecular complex may serve to amplify the early CD2-mediated activation signals. CD3 contains multiple EX 2 YX 2 L/IX 7 YX 2 L/I motifs, termed immune receptor tyrosinebased activation motifs (ITAMs). ITAMs are phosphorylated upon activation and may serve as potential ligands of the Fyn SH2 domain (18,52,53). Lck has been shown to be able to phosphorylate CD3 (54); whether Lck catalyzed CD3 phosphorylation in vivo is not clear. In unstimulated cells, minimal amount of Fyn were associated with CD2 (Fig. 3A). Anti-CD2 mAbs stimulation resulted in the tyrosyl phosphorylation of a number of intracellular proteins, including CD3, mediated by a tyrosine kinase, potentially Lck. Phosphorylation of CD3 rendered the CD3 able to bind to the Fyn SH2 domain (Fig.  6B); occupancy of the Fyn SH2 domain by ligand appeared to be able to reverse the inhibitory effect of the Fyn SH2 domain on the binding of the Fyn SH3 domain to CD2. We propose that as a result of occupancy of the Fyn SH2 domain, possibly by CD3 itself, quantitatively more Fyn is able to be recruited to the CD2 complex, and, cooperatively, recruitment of Fyn (and other PTKs) may potentially catalyze further tyrosyl phosphorylation of CD3 and other downstream effectors. Thus, formation of the multimolecular CD2 complex, including CD3 and Fyn, appears to result in amplification of intracellular signaling. Although our data suggest the functional involvement of CD3 in CD2-mediated signaling, it does not exclude the involvement of other CD3 components. Other CD3 components containing the ITAM motif(s), including CD3⑀, have been shown to play a role in CD2-mediated signaling (55).
The association of CD2 with Src family PTK, including Lck and Fyn, has been demonstrated previously (21,23,25). CD2induced signal was abolished in Jurkat cells lacking Lck expression (42). However, the functional significance of Fyn in CD2-induced signal transduction has not been addressed. We show here that a loss-of-function Fyn SH2* mutant inhibited CD2-mediated NFAT-transcriptional activity, suggesting a critical role of Fyn in the CD2-mediated signal pathway. Fur-thermore, the function of Fyn in CD2-induced pathway was, at least in part, mediated by the non-catalytic domain, as the kinase domain was intact in Fyn SH2* mutant. This result is consistent with our model in that the Fyn SH2* molecule (with an intact SH3 domain) will compete with endogenous Fyn for binding to CD2 but, unable to bind to or recruit phosphotyrosine-containing signaling molecules, will therefore inhibit CD2-induced signaling.
CD2 is able to associate with both Fyn and Lck. Our data suggest that each kinase plays independent roles in CD2-induced transduction pathway. Lck and Fyn are not functionally interchangeable; although thymocyte development was arrested at an early stage in mice with a homozygous deletion of Lck expression (56), thymocyte development was grossly normal in mice lacking Fyn expression (57,58) or bearing a dominant negative form of Fyn (59). Lck but not Fyn has been reported to associate with CD4 and CD8 (60, 61), whereas Fyn but not Lck has been shown to bind the phosphoprotein p120/ p130 FYB (62). The SH3 domain of Fyn has been shown to precipitate quantitatively higher amounts of Cbl and phosphatidylinositol 3-kinase than the SH3 domain of Lck from cell lysates (39,63); the functional significance of these differences are not known. The characterization of Pyk2, a downstream molecule of Fyn but not Lck (64), further suggested that Fyn and Lck may lead to divergent signal transduction pathways. In certain immunofluorescence studies, CD2 co-localized with Fyn but not with Lck after anti-CD2 mAbs treatment (45). Finally, in some T cell lines rendered anergic, the kinase activity of Fyn was elevated, whereas that of Lck was unchanged or diminished (65,66). It is of note that CD2 stimulation has been shown to be able to reverse anergy and T cell unresponsiveness (19); the recruitment of Fyn by CD2 may play a role in reversing the anergic state.