INTRODUCTION

Activation of T cells through the T cell receptor (TCR)¹ induces the tyrosine phosphorylation of critical signaling intermediates. This is regulated by protein-tyrosine kinases and the CD45 family of transmembrane protein-tyrosine phosphatases (1, 2). In humans, five CD45 isoforms are generated by regulated alternative splicing of three exons (3, 4, 5). These exons, commonly known as exons A, B, and C, are located near the 5-end of the gene and give rise to isoforms that differ only in the length and glycosylation of their extracellular domains. Although individual lymphocytes simultaneously express more than one CD45 isoform (6, 7, 8), they are differentially expressed by subsets of T cells with distinct functions and activation requirements (9, 10, 11, 12). Furthermore, CD45 isoform expression is dynamic. Individual T cells alter their isoform expression in a highly regulated manner during thymic selection and upon antigen exposure in the periphery (7, 8, 13, 14, 15).

The importance of CD45 in T cell activation is demonstrated by the failure of CD45 mutants to respond to anti-CD3 or antigen (16, 17). CD45 has been shown to regulate the protein-tyrosine kinase activity of Lck and Fyn by dephosphorylation at one or more sites (18, 19, 20, 21, 22, 23). However, additional substrates, including the TCR -chain and LPAP, have been implicated (24, 25), and the complete spectrum of CD45 substrates is not yet known. Although TCR-mediated activation requires the presence of the cytoplasmic CD45 protein-tyrosine phosphatase domains (26, 27, 28), the role of the tightly regulated extracellular domains is poorly understood. It has been hypothesized that the CD45 extracellular domain superimposes regulatory influences upon the activity or substrate specificity of the protein-tyrosine phosphatase domains (29). In agreement, we recently demonstrated that the CD45(ABC) and CD45(0) isoforms differentially regulate activation-induced IL-2 secretion and tyrosine phosphorylation of several cellular proteins including Vav (30).

Vav is the 95-kDa product of the vav proto-oncogene. Its importance in lymphocyte activation and proliferation has been confirmed in Vav-deficient mice; however, the signaling pathways involved have yet to be clarified (31, 32). Vav is rapidly phosphorylated after ligation of the TCR·CD3 complex, CD28, or the IL-2 receptor, and Vav overexpression up-regulates IL-2 promoter activity (33, 34, 35, 36, 37). Vav contains an array of signaling and protein-protein interaction motifs, including Dbl and pleckstrin homology domains and a carboxyl-terminal SH2 domain flanked by two SH3 domains (33, 38). Point mutations within the Vav SH2 domain inhibit its transforming potential (38), implicating this domain in Vav-mediated signaling pathways. In this regard, Vav has been shown to interact with other signaling molecules, including Shc, Grb2, ZAP-70, VAP-1, CD19, and phosphatidylinositol 3-kinase (p85), through SH2- and/or SH3-mediated interactions after T or B cell activation (33, 39, 40, 41, 42).

We therefore wished to determine whether or not CD45 isoform-specific differences in the tyrosine phosphorylation of Vav were associated with the differential interaction of Vav with downstream signaling molecules. Utilizing our model system, whereby different single CD45 isoforms are uniquely expressed in Jurkat cells whose endogenous CD45 expression has been blocked by an antisense gene, we noted that a 76-kDa protein associates with Vav and also undergoes activation-induced isoform-dependent phosphorylation on tyrosine. We have now identified this Vav-associated molecule as SLP-76, a recently cloned protein that binds Grb2 in T cells (43). We demonstrate that SLP-76 undergoes CD45 isoform-dependent differential phosphorylation and recruitment to Vav through the Vav SH2 and SH3 domains. These results further delineate both Vav-mediated and CD45 isoform-specific signaling pathways resulting from T cell activation.

MATERIALS AND METHODS

As described previously (30), endogenous CD45 expression was blocked in the Jurkat human leukemic CD4⁺ T cell line by stable transfection with an antisense plasmid construct targeting a region just upstream of the CD45 initiation codon. One G418-resistant clone (J-AS-1), completely lacking surface or cytoplasmic CD45, was stably cotransfected with cDNAs encoding the smallest (CD45(0)) or the largest (CD45(ABC)) isoform plus pPGKhyg (encoding hygromycin B resistance) (30). In resistant clones, CD45 expression of only the transfected isoform was documented by immunofluorescence and immunoblotting (30). Cells were grown in RPMI 1640 medium supplemented with 10% iron-fortified calf serum, L-glutamine, and gentamycin. Transfectants were maintained in G418 (0.5 mg/ml) with or without hygromycin (0.3 mg/ml), as appropriate.

mAbs against CD2, CD3, CD4, CD28, CD45, and CD45RA were from Dr. C. Morimoto (Dana-Farber Cancer Institute, Boston, MA), Coulter Corp. (Hialeah, FL), and Dako Corp. (Carpinteria, CA). Anti-CD45RO was from Dr. P. Beverly (University College Hospital, London). Cell phenotype was routinely monitored for these markers using a BD FACSTAR IV as described (30). Anti-Vav (rabbit polyclonal antibody) was generated against residues 575-594 (from Dr. A. Altman, La Jolla Institute for Allergy and Immunology, La Jolla, CA) (34) or the Vav Dbl region (from Drs. X. Bustelo and M. Barbacid, Bristol-Meyer Squibb Pharmaceutical Research Institute, Princeton, NJ) (33). Anti-Vav (mouse mAb) was from Dr. J. Griffin (Dana-Farber Cancer Institute) (44). Anti-SLP-76 sheep antiserum was generated against amino acids 136-235 of human SLP-76 as described (45).

Cells (15-20 × 10⁶/sample) were stimulated at 37 °C either with anti-CD3 (14 µg/ml) or with pervanadate (3 mM H₂O₂, 100 µM Na₃VO₄) plus 10 µM phenylarsine oxide, which mimics the effects of TCR ligation (46, 47). At the times indicated, ice-cold stop solution (phosphate-buffered saline with phosphatase inhibitors) was added, followed immediately by brief centrifugation in a microcentrifuge, supernatant removal, and resuspension of the pellets in ice-cold 1% Nonidet P-40 lysis buffer containing 50 mM Tris-HCl (pH 8.0) with 150 mM NaCl, 2 mM aminoethylbenzenesulfonyl fluoride, 10 µg/ml aprotinin, 10 µg/ml leupeptin, 10 mM iodoacetamide, 1 mM sodium vanadate, 10 mM NaF, and 10 mM sodium pyrophosphate for 20 min at 4 °C, followed by centrifugation at 14,000 rpm for 15 min as described (30).

Immunoprecipitations were performed as described (30, 48). Briefly, after preclearing, equivalent amounts of protein from each lysate (Bio-Rad DC protein assay) were incubated with anti-Vav or anti-SLP-76, followed by immunoprecipitation with protein A-Sepharose or GammaBind Plus Sepharose (Pharmacia Biotech Inc.), respectively. After extensive washing, immunoprecipitates were subjected to 8-12% SDS-PAGE (reducing conditions), transferred to nitrocellulose, and blocked with 5% nonfat milk in phosphate-buffered saline. Membranes were immunoblotted with anti-phosphotyrosine (anti-Tyr(P)) mAb 4G10 (provided by Dr. B. Drucker, University of Oregon Health Sciences Center, Portland, OR), anti-Vav mAb, or anti-SLP-76 sheep serum, followed by the appropriate horseradish peroxidase-conjugated secondary Ab, and developed with enhanced chemiluminescence (DuPont NEN).

Constructs encoding GST fusion proteins contained the following portions of Vav: SH2+SH3_(COOH) (from Dr. B. Margolis, University of Michigan, Ann Arbor, MI), SH2 (from Dr. A. Altman, with permission of Dr. S. Katzav, Hebrew University, Jerusalem, Israel), and SH3_(COOH) (from Dr. X. Bustelo). Fusion proteins were induced in transformed Escherichia coli DH5 cells with 0.1 mM isopropyl--D-thiogalactopyranoside and affinity-purified using glutathione-agarose (Sigma). Purified proteins were quantitated by comparison with protein standards on Coomassie Blue-stained SDS-polyacrylamide gels. For precipitation, agarose-associated GST fusion proteins (2.5-10 µg) were incubated with cell lysates (0.25 ml) for 3 h at 4 °C, extensively washed in high salt (500 mM LiCl) and plain lysis buffers, and then immunoblotted as described above.

RESULTS AND DISCUSSION

The Jurkat human leukemic T cell line normally expresses CD45 at high levels. Like other T lymphocytes, individual cells express multiple isoforms simultaneously (5, 30). To examine the role of CD45 and its individual isoforms free from potential confounding influences of unknown mutations, we directly targeted endogenous CD45 expression in Jurkat cells by stable transfection of a CD45 antisense construct (30). One resulting CD45 clone, J-AS-1, was then stably transfected with cDNA encoding either the smallest isoform, denoted CD45(0), which lacks alternative exons, or the largest isoform, denoted CD45(ABC), which includes all three alternative exons. Two transfectants uniquely expressing CD45(ABC) (clones J[ABC]-1 and -2) and three transfectants expressing CD45(0) (clones J[0]-1, -2, and -3) (30) were used in this study. Although total CD45 expression was higher in wild-type Jurkat cells than in the transfectants, the expression of individual CD45(0) and CD45(ABC) isoforms by J[0] and J[ABC] cells, respectively, was similar to their expression in wild-type cells (30). Overall, CD45 expression in J[0] and J[ABC] cells was similar. The expression of other cell-surface molecules, including CD2, CD3, CD4, and CD28, was comparable between cell lines (data not shown).

Using these cell lines, we previously reported CD45 isoform-associated differences in the tyrosine phosphorylation of Vav (30). Activation-induced tyrosine phosphorylation directs SH2-mediated interactions between Vav and a number of other signaling molecules that are likely to be important in Vav-mediated signaling (33, 39, 40, 41). We therefore wished to determine whether the differential tyrosine phosphorylation of Vav resulting from the expression of distinct CD45 isoforms was associated with differential phosphorylation and/or interaction with other signaling molecules.

As before, anti-CD3 stimulation consistently induced significantly greater tyrosine phosphorylation of Vav in Jurkat cells and particularly in J[ABC] transfectants compared with J[0] transfectants or the CD45 J-AS-1 cells (Fig. 1A). Reprobing the same membrane with anti-Vav mAb confirmed similar precipitation of Vav protein in each lane (Fig. 1B). In addition, we noted that a 76-kDa phosphoprotein (pp76) is consistently coprecipitated with Vav from Nonidet P-40 lysates and follows exactly the same pattern of tyrosine phosphorylation as Vav (Fig. 1A). Thus, Vav-associated pp76 is phosphorylated on tyrosine after T cell activation and undergoes relative hyperphosphorylation in Jurkat and CD45(ABC)-expressing transfectants compared with transfectants expressing CD45(0) or lacking CD45 altogether. This same pattern was also observed 5 min after stimulation of each cell line, at which time diminished tyrosine phosphorylation of both proteins is usually observed (data not shown). We found that pp76 can be coprecipitated by antibodies directed against several different regions of Vav (data not shown). Whereas each of these antisera recognizes Vav by immunoblotting, none recognizes Vav-associated pp76, indicating that this molecule is not a proteolytic cleavage product of Vav. Thus, pp76 is specifically coprecipitated with Vav and, like Vav, exhibits CD45 isoform-specific phosphorylation after cellular activation. Additional proteins of 65 and 70 kDa were often coprecipitated with Vav, but these bands were variable and did not routinely exhibit CD45 isoform-dependent tyrosine phosphorylation. This emphasizes the selective nature of isoform-specific signals. The band at 70 kDa may correspond to ZAP-70, reported by others to associate with Vav (40).

We next examined the time course of Vav and Vav-associated pp76 tyrosine phosphorylation after anti-CD3 stimulation of Jurkat cells (Fig. 2). Both molecules are rapidly phosphorylated, reaching a maximum by 30 s to 1 min. The dephosphorylation of both bands also follow similar kinetics. Although the rate of dephosphorylation showed some interassay variability, a gradual decrease in phosphorylation of both bands was usually noticeable by 5-10 min. Thus, as determined by anti-Tyr(P) immunoblotting of Vav immunoprecipitates, the phosphorylation of Vav and Vav-associated pp76 parallel one another. Taken together with the CD45 isoform-specific differences in phosphorylation, these data suggest that pp76 and Vav phosphorylation could be linked to the same protein-tyrosine kinase pathway whose activity depends upon CD45, but is differentially regulated by distinct CD45 isoforms. Of course, these studies do not distinguish between increased pp76 phosphorylation and increased recruitment to Vav. Therefore, it is also possible that CD45 isoform-dependent phosphorylation of Vav may result in increased recruitment of pp76, whose phosphorylation is regulated in a manner unrelated to the phosphorylation of Vav.

To map the sites of interaction between Vav and pp76, GST-Vav fusion proteins were used. A GST fusion protein containing the Vav SH2 domain plus the carboxyl-terminal SH3 domain (GST-Vav-SH2+SH3_(COOH)) does precipitate some tyrosine-phosphorylated pp76 from unstimulated cell lysates (Fig. 3). However, the precipitation of tyrosine-phosphorylated pp76 by GST-Vav-SH2+SH3_(COOH) or by a fusion protein containing the Vav SH2 domain alone (GST-Vav-SH2) is greatly augmented following cellular activation. Neither control GST-Sepharose alone nor GST-Vav-SH3_(COOH) (up to 100 µg/ml) precipitates phosphorylated pp76. Thus, interaction between the carboxyl-terminal tail of Vav and phosphorylated pp76 is dependent on Vav SH2-mediated binding. However, as can be seen in Fig. 3, the Vav SH2 plus SH3_(COOH) domains bind tyrosine-phosphorylated pp76 more efficiently than equivalent amounts of the Vav SH2 domain alone. These data suggest that although the Vav SH2 domain is necessary and sufficient for association to occur, phosphorylated pp76 may bind cooperatively to the Vav SH2 and SH3 domains. Cooperative interactions between SH2 and SH3 domains have been described previously (49). As a measure of specificity, precipitations were carried out using decreasing amounts of each fusion protein (range of 12.5 to 2.5 µg/sample). Although the intensity of each band decreased, the overall pattern of bands precipitated by each fusion protein remained unchanged (data not shown). Compared with native Vav, the GST-Vav fusion proteins precipitated a similar spectrum of tyrosine phosphoproteins but relatively larger amounts of the proteins at 65 and 70 kDa.

Given an apparent molecular mass of 76 kDa and evidence of interaction with the Vav SH2 and probably SH3 domains, we examined whether pp76 was identical to SLP-76, a recently cloned 76-kDa molecule that binds Grb2 (43). SLP-76 undergoes activation-related tyrosine phosphorylation (50, 51) following kinetics similar to those we observed for pp76 (Fig. 2). Furthermore, SLP-76 contains three tandem Tyr(P)-Glu-(Ser/Pro)-Pro sequences that resemble the predicted optimal sequence for binding by the Vav SH2 domain (Tyr(P)⁺¹-Glu and Tyr(P)⁺³-Pro) (52). As shown in Fig. 4 (left panel), anti-SLP-76 precipitates a tyrosine-phosphorylated band that comigrates with the pp76 band precipitated by anti-Vav or GST-Vav-SH2+SH3_(COOH). Anti-SLP-76 immunoblotting of a parallel membrane (Fig. 4, center panel) indicates that pp76 precipitated by both anti-Vav and GST-Vav-SH2+SH3_(COOH) contains SLP-76. Furthermore, the amount of SLP-76 coprecipitated by Vav or GST-Vav increases after cellular activation. Whether SLP-76 binding to Vav or to this Vav fusion protein prior to cellular activation reflects a phosphotyrosine-dependent association mediated by low basal levels of SLP-76 phosphorylation versus phosphotyrosine-independent binding is not yet certain. Finally, anti-SLP also coprecipitates Vav in an activation-related fashion (Fig. 4, right panel). These results identify pp76 as SLP-76.

Given the activation-related increase in SLP-76 association with Vav (Fig. 4) and the involvement of the Vav SH2 domain (Fig. 3), it was important to determine whether the CD45 isoform-dependent differential tyrosine phosphorylation of Vav and SLP-76 observed in our single isoform transfectants was associated with parallel differences in the physical interaction of these two molecules. As shown above, anti-CD3 stimulation results in relative hyperphosphorylation of Vav and Vav-associated SLP-76 in wild-type and CD45(ABC)-expressing transfectants compared with CD45 (J-AS-1) and transfectants expressing CD45(0) (Figs. 1A and 5A). Direct comparison with a parallel membrane immunoblotted with anti-SLP-76 clearly demonstrates that the differences in SLP-76 tyrosine phosphorylation observed in these cell lines reflect the differential association of SLP-76 with Vav (Fig. 5, compare A and B).

When SLP-76 is directly immunoprecipitated from single isoform transfectants with anti-SLP, anti-Tyr(P) immunoblotting reveals that total SLP-76, and not just that fraction associated with Vav, undergoes activation-induced CD45 isoform-specific tyrosine phosphorylation (Fig. 6). Consistent with the findings above, CD45(ABC) transfectants demonstrate relative hyperphosphorylation of SLP-76 and increased coprecipitation of tyrosine-phosphorylated Vav compared with cells expressing CD45(0). Taken altogether, our results demonstrate that the expression of distinct CD45 isoforms is associated with differential tyrosine phosphorylation of Vav and SLP-76 and differential recruitment of SLP-76 to Vav.

Critical for TCR-mediated signaling, CD45 appears able to either increase or decrease the activity of Src family kinases such as Lck and Fyn by regulating the dephosphorylation of one or more tyrosine residues (18, 19, 20, 21, 22, 23). Reconstitution of CD45 mutants with chimeric protein-tyrosine phosphatase molecules that lack the CD45 transmembrane or extracellular domains restores near-normal patterns of activation-induced tyrosine phosphorylation (26, 27, 28). Although these studies demonstrate the requirement for the CD45 protein-tyrosine phosphatase domains in TCR-mediated signaling, they do not exclude a potentially important superimposed regulatory role for the physiological extracellular domains. This role has recently been documented by the direct comparison of cells that differ only in their CD45 isoform expression (30, 53). We now extend our previous observations and establish a novel link between CD45 isoform expression, Vav, and SLP-76. It is interesting to note that each of the three molecules identified in this pathway is expressed only in hematopoietic cells. CD45 isoform-specific pathways appear to involve only a subset of those proteins undergoing activation-induced tyrosine phosphorylation (30). Regardless, the overall functional significance of isoform-specific pathways is demonstrated by the significant differences in IL-2 secretion exhibited by cells expressing distinct CD45 isoforms both in our model (30) and in the mouse thymoma model of Novak et al. (53).

It has been proposed that differential interactions between the various CD45 extracellular domains and unknown ligand(s) might direct the cytoplasmic protein-tyrosine phosphatase domains toward distinct intracellular substrates (29, 54). The actual mechanism by which distinct CD45 isoforms differentially regulate Vav and SLP-76 phosphorylation is unknown. Although both molecules might be preferentially dephosphorylated by the CD45(0) isoform, this would not easily explain their decreased phosphorylation in CD45 cells or increased phosphorylation in wild-type cells (which express both CD45(0) and CD45(ABC)). A more consistent hypothesis is that SLP-76 and Vav are phosphorylated by one or more closely related protein-tyrosine kinases whose activity is differentially regulated by distinct CD45 isoforms. Our data further suggest that the tyrosine phosphorylation of SLP-76 augments its recruitment to Vav, through the Vav SH2 and SH3 domains. The phosphorylation of Vav does not appear to contribute to SLP-76 binding since fusion proteins containing the SLP-76 SH2 domain do not precipitate Vav (45), and SLP-76 does not contain alternative phosphotyrosine recognition (PTB) domains. This is the first demonstration that the expression of distinct CD45 isoforms differentially regulates physical interactions between downstream signaling molecules.

Which protein-tyrosine kinase(s) are responsible for the phosphorylation of these molecules in vivo is unknown. Vav can be phosphorylated by Lck in vitro, although IL-2-dependent phosphorylation of Vav occurs in Lck-deficient cells (34, 36). CD28 ligation results in a temporal association between Itk and Vav phosphorylation, suggesting a possible link in T cells (35). This finding may be bolstered by a study showing that IL-3 induces the specific association of Tec kinase with Vav through its Tec homology domain in hematopoietic progenitor cells (55). Finally, it has been reported that ZAP-70 specifically associates with the Vav SH2 domain after T cell activation (40).

Although the consequences of TCR-induced tyrosine phosphorylation and physical association of SLP-76 and Vav are not yet known, they are likely to be of physiological importance. Using transient cotransfection assays, the overexpression of Vav and of SLP-76 have independently been shown to increase TCR-stimulated activity of IL-2 promoter reporter constructs (37, 45). More recent data show that simultaneous overexpression of SLP-76 and Vav results in synergistic augmentation of IL-2 promoter activity, suggesting possible functional interaction (56). However, the regulation of these signals may be complex. Our experiments indicate that Vav phosphorylation and SLP-76 phosphorylation are inversely related to IL-2 secretion in our single CD45 isoform transfectants (30). This raises the possibility that the tyrosine phosphorylation of particular sites might be inhibitory, perhaps through the induction of down-regulatory interactions. Alternatively, additional signaling defects that affect IL-2 secretion may be present in cells singly expressing the CD45(ABC) isoform. Clearly, identification of the effector molecules lying both up- and down- stream of Vav and SLP-76 will be required for a more detailed understanding of this important signaling pathway.

The tyrosine phosphorylation of Vav may direct its association with other signaling molecules (39), and we have now shown that the phosphorylation of SLP-76 up-regulates its association with Vav. Both Vav and SLP-76 constitutively associate with Grb2 through interactions requiring the carboxyl-terminal SH3 domain of Grb2 (42). The ability of Grb2 to interact with Sos suggests a mechanism by which Vav might be linked to the Ras pathway. However, whether Sos binding in vivo requires only the amino-terminal SH3 domain or both SH3 domains of Grb2 is controversial (51, 57, 58, 59, 60). Thus, it is unclear whether or not a Vav·SLP-76·Grb2 complex would also contain Sos. Regardless, both SLP-76 and Grb2 contain SH2 domains capable of recruiting additional molecules into a complex containing Vav. For example, the SH2 domain of Grb2 has been shown to associate with Shc, a 116-kDa phosphoprotein recently identified as p120^cbl, and an uncharacterized 36-kDa phosphoprotein (50, 51, 61). The SH2 domain of SLP-76 associates with proteins of 62 and 130 kDa (45). It will now be crucial to determine which of these molecules can simultaneously associate into a signaling complex. These molecules will become prime candidates for differential regulation through this CD45 isoform-dependent pathway. Given the ability of individual T cells to alter their expression of CD45 isoforms during thymic development and after antigen exposure in the periphery, an understanding of the signaling pathways preferentially utilized by particular CD45 isoforms should ultimately provide important insight into the signals involved in maturation and development of functional repertoire.