PAG-Associated FynT Regulates Calcium Signaling and Promotes Anergy in T Lymphocytes

Abstract

Phosphoprotein associated with glycolipid-enriched membranes (PAG), also named Csk-binding protein (Cbp), is a transmembrane adaptor associated with lipid rafts. It is phosphorylated on multiple tyrosines located in the cytoplasmic domain. One tyrosine, tyrosine 314 (Y314) in the mouse, interacts with Csk, a protein tyrosine kinase that negatively regulates Src kinases. This interaction enables PAG to inhibit T-cell antigen receptor (TCR)-mediated T-cell activation. PAG also associates with the Src-related kinase FynT. Genetic studies indicated that FynT was required for PAG tyrosine phosphorylation and binding of PAG to Csk in T cells. Herein, we investigated the function and regulation of PAG-associated FynT. Our data showed that PAG was constitutively associated with FynT in unstimulated T cells and that this association was rapidly lost in response to TCR stimulation. Dissociation of the PAG-FynT complex preceded PAG dephosphorylation and PAG-Csk dissociation after TCR engagement. Interestingly, in anergic T cells, the association of PAG with FynT, but not Csk, was increased. Analyses of PAG mutants provided evidence that PAG interacted with FynT by way of tyrosines other than Y314. Enforced expression of a PAG variant interacting with FynT, but not Csk, caused a selective enhancement of TCR-triggered calcium fluxes in normal T cells. Furthermore, it promoted T-cell anergy. Both effects were absent in mice lacking FynT, implying that the effects were mediated by PAG-associated FynT. Hence, besides enabling PAG tyrosine phosphorylation and the PAG-Csk interaction, PAG-associated FynT can stimulate calcium signals and favor T-cell anergy. These data improve our comprehension of the function of PAG in T cells. They also further implicate FynT in T-cell anergy.

Phosphoprotein associated with glycolipid-enriched membranes (PAG) or Csk-binding protein (Cbp) is a transmembrane adaptor protein expressed in most cell types, including immune cells (4, 16). It contains a short extracellular domain with no known ligand-binding capacity, a transmembrane segment, and a long cytoplasmic domain with sites of tyrosine phosphorylation, proline-rich sequences, and a carboxyl terminus capable of interacting with PSD-95, Discs-large, and ZO-1 (PDZ) domain-containing proteins like EBP-50 (12). Due to the palmitylation of two cysteines in the proximal portion of the cytoplasmic domain, PAG/Cbp is largely localized to lipid-rich membrane microdomains known as lipid rafts.

So far, the major protein shown to associate with PAG/Cbp (hereafter named PAG) is Csk, a cytoplasmic protein tyrosine kinase (PTK) implicated in the inhibition of Src-related PTKs (4, 16, 34). Peptide binding and site-directed mutagenesis studies indicated that this association is mediated by phosphorylation of PAG at tyrosine 314 (Y314) in mice (Y317 in humans) and by the SH2 domain of Csk (4, 6, 16, 28). This interaction was shown to enhance the catalytic activity of Csk and to favor inactivation of Src family kinases in lipid rafts (4, 16, 28). It was proposed that PAG is implicated in the negative regulation of Src kinases in lipid rafts and, thus, is likely to be a negative regulator of cellular processes mediated by Src kinases.

Direct support for this idea was provided by the finding that enforced expression of PAG in COS-7 cells caused pronounced reductions of the enzymatic activities of Src-related kinases (16). This effect was dependent on Y314 of PAG. Furthermore, it was reported that overexpression of PAG in normal T cells, the T-cell line Jurkat, or the basophil leukemia cell line RBL-2H3 suppressed cellular responses mediated by Src-related kinases (4, 6, 21). The inhibitory impact of PAG in normal T cells also required Y314 (6). Finally, it was shown that diminution of PAG expression in mouse embryo fibroblasts by RNA interference caused changes in cell spreading and adhesion similar to those observed in Csk-deficient fibroblasts (25). While these various findings provided a compelling argument that PAG-Csk plays a role in the negative regulation of Src family kinases, it is noteworthy that two groups recently reported that mice lacking PAG exhibited little or no phenotype (8, 36). Given that a severe phenotype was observed in Csk-deficient mice (11, 19), it was postulated that other, PAG-independent mechanisms of Csk recruitment may exist and that these mechanisms compensate for the absence of PAG expression.

Characterizations of the interaction between PAG and Csk have been mostly performed with T cells. In resting T cells, PAG was shown to be constitutively tyrosine phosphorylated and associated with Csk (4, 6, 29). These modifications are rapidly lost upon engagement of the T-cell antigen receptor (TCR) complex. In this light, it was proposed that the PAG-Csk complex may prevent activation of resting T cells and that relief of this inhibition upon TCR engagement may be instrumental in allowing T-cell activation. In agreement with this, we reported that augmentation of expression of wild-type PAG in T cells by transgenesis caused inhibitions of TCR-triggered protein tyrosine phosphorylation, cellular proliferation, and interleukin-2 (IL-2) secretion (6). Opposite effects were seen in transgenic mice expressing a PAG mutant in which Y314 was replaced by phenylalanine (PAG Y314F), supporting the idea that these effects were mediated by Csk.

In addition to Csk, other proteins interact with PAG. In T cells, PAG is associated with the Src-related protein tyrosine kinase FynT (4). While the structural basis of this association is not known, this complex has been reported not to be modulated by TCR stimulation (4), suggesting that it is not mediated by PAG tyrosine phosphorylation. The interaction between PAG and FynT is probably critical for the function of PAG, as PAG tyrosine phosphorylation and binding of PAG to Csk were noted to be dramatically reduced in FynT-deficient T cells (37). A similar association of PAG with Lyn, another member of the Src family, was described for mast cells (20). Moreover, PAG was observed to interact by way of its carboxyl-terminal tail with EBP-50, a PDZ domain-containing protein associated with the cytoskeleton (12). This interaction seemingly links lipid raft-associated PAG to the cytoskeleton.

In this paper, we examined in greater detail the regulation and function of the interaction between PAG and FynT. Our data showed that, as is the case for the PAG-Csk interaction, the PAG-FynT association was rapidly lost in response to TCR stimulation. Dissociation of PAG-FynT complexes preceded disappearance of PAG-Csk complexes, suggesting that the former may trigger the loss of the PAG-Csk interaction in activated T cells. In other experiments, we noted that the extents of the interactions between PAG and FynT, but not PAG and Csk, were augmented in anergic T cells. Moreover, enforced expression in normal T cells of a PAG mutant capable of interacting with FynT, but not Csk, resulted in a selective enhancement of TCR-triggered calcium fluxes and promoted T-cell anergy. Together, these data provide a better understanding of the role of the PAG-FynT interaction in T cells. Furthermore, they further implicate FynT in the induction of T-cell anergy.

MATERIALS AND METHODS

Antibodies and reagents.

Antibodies directed against PAG, FynT, Csk, and phosphotyrosine (either polyclonal rabbit antiphosphotyrosine antibodies generated in our laboratory or antiphosphotyrosine monoclonal antibody [MAb] 4G10) were described previously (5-7). Antibodies against phosphorylated Y314 of mouse PAG (Y317 of human PAG) were generated by immunizing rabbits with the peptide lysine-glutamate-isoleucine-serine-alanine-methionine-phosphotyrosine-serine-serine, which corresponds to amino acids 311 to 319 of human PAG. In the mouse, these antibodies react with PAG phosphorylated at Y314 but not with nonphosphorylated PAG or with PAG phosphorylated at other sites (our unpublished results). Antibodies against Erk, phospho-Erk, IκBα, or phospho-IκBα were purchased from Cell Signaling Technology Inc. (Beverly, MA), Santa Cruz Biotechnology Inc. (Santa Cruz, CA), or Upstate (Charlottesville, VA). Biotinylated anti-CD3 MAb 145-2C11, biotinylated anti-TCR MAb H57-597, and anti-CD28 MAb 37-51 were purchased from eBiosciences (San Diego, CA) or BD Biosciences (Mississauga, Ontario, Canada). Avidin was purchased from Calbiochem-Novabiochem Corporation, San Diego, CA.

Mice.

Transgenic mice expressing wild-type PAG, a PAG mutant in which tyrosine 314 was mutated to phenylalanine (PAG Y314F), or a PAG mutant in which all nine cytoplasmic tyrosines were mutated to phenylalanines (PAG 9YF), under the control of the human CD2 promoter, were reported elsewhere (6). A similar approach was used to create transgenic mice expressing a PAG mutant in which all cytoplasmic tyrosines except Y314 were mutated to phenylalanines (PAG 8YF). Transgenic mice were produced by the IRCM Transgenic Service, according to established protocols. At least two independent founders of each transgenic type were used. The levels of the exogenous PAG proteins were 6- to 12-fold greater than those of endogenous PAG (data not shown). Mice lacking expression of Fyn (fyn^−/−) (27) were obtained from Jackson Laboratory, Bar Harbor, ME. Transgenic mice expressing the class I major histocompatibility complex-restricted, lymphocytic choriomeningitis virus (LCMV) gp33-specific TCR P14 (22) were obtained from Taconic, Hudson, NY. All animals were backcrossed for at least five generations to the C57BL/6 background.

Cell stimulation.

Thymocytes were obtained by making cell suspensions from thymus tissue. CD4⁺ lymph node or spleen T cells, depleted of natural killer T cells, were purified from mice by using a Stem Cell purification kit (Stem Cell Technology Inc., Vancouver, British Columbia, Canada) or T-cell enrichment columns (R&D Systems, Minneapolis, MN). Cell purity was consistently greater than 90% (data not shown). Effector CD4⁺ T cells were generated by stimulating purified CD4⁺ T cells with coated anti-CD3 (3 μg per ml) and soluble anti-CD28 (1 μg per ml) for 2 to 3 days and then expanding them in IL-2 (50 units per ml)-containing medium for 3 days generally. Cells (at a concentration of 30 × 10⁶ per ml) were stimulated for the indicated periods of time at 37°C or room temperature (21°C) with biotinylated anti-CD3 (10 μg) or anti-TCR (10 μg) and avidin (14 μg) in a volume of 200 μl. Unstimulated controls were incubated with avidin alone. After lysis in buffer containing maltoside supplemented with protease and phosphatase inhibitors (6), lysates were subjected to immunoprecipitation or immunoblot analysis.

Cell fractionation.

Cells were lysed in 1 ml of Brij-58-containing buffer supplemented with protease and phosphatase inhibitors, as detailed elsewhere (6). Lysates were then mixed with 80% sucrose (made in the same buffer without detergent) and overlaid sequentially with 30% sucrose and 5% sucrose. After centrifugation at 200,000 × g, 0.5-ml fractions were collected from the top of the gradient. Fractions 2 and 3 contained most of the lipid rafts, while fractions 8 and 9 contained the soluble proteins (data not shown) (6). Individual fractions were subjected to immunoblot analysis after solubilization using 1% maltoside.

Immunoprecipitations and immunoblot analyses.

To extract PAG and PAG-associated proteins from lipid rafts, cells were lysed in a buffer containing maltoside (1% n-dodecyl-β-d-maltoside, 50 mM Tris [pH 7.6], 150 mM NaCl, and 2 mM EDTA) supplemented with protease and phosphatase inhibitors (6). Immunoprecipitations and immunoblot analyses were performed as previously described (7, 33). Radioactivity was quantitated with a Storm PhosphorImager (Molecular Dynamics, General Electrics Canada, Mississauga, Ontario, Canada).

Intracellular-calcium fluxes.

Thymocytes or splenic and lymph node T cells were loaded with Indo-1 (10 μM; Molecular Probes, Eugene, OR) for 45 min at 37°C and stained with phycoerythrin-coupled anti-CD4 MAb L3T4 and fluorescein isothiocyanate-coupled anti-CD8 MAb Ly-2 (BD Biosciences, Mississauga, Ontario, Canada), as described previously (6). After the cells were washed, they were stimulated at 37°C with biotinylated anti-TCR or anti-CD3 and avidin. Changes in intracellular calcium over time were monitored using the FL4 and FL5 channels of a BD LSR cell analyzer (BD Biosciences, Mississauga, Ontario, Canada). CD4⁺ single-positive cells were selectively analyzed by gating on CD4⁺ CD8⁻ cells. As a control, cells were stimulated with the calcium ionophore ionomycin (100 ng per ml).

Anergy induction.

For induction of anergy in vitro using ionomycin or anti-CD3, CD4⁺ T cells were first activated for 48 h with anti-CD3 (3 μg per ml; on plastic) and anti-CD28 (1 μg per ml; soluble) (14, 15, 17). They were then expanded in IL-2 (50 U per ml) for 3 days. After this, anergy was induced by treating cells with ionomycin (1 μM) or coated anti-CD3 alone (1 μg per ml) for 16 h. Controls were with no addition. For anergy induced by ionomycin, cells were subsequently processed immediately, either for biochemical studies (immunoprecipitation or immunoblot analysis) or for functional studies (proliferation induced by anti-CD3 plus anti-CD28). For anergy triggered with anti-CD3, cells were first washed and rested in growth medium for 30 h. They were then restimulated with anti-CD3 (0.11 μg per ml) and anti-CD28 (1 μg per ml) for 48 h. Thymidine incorporation and IL-2 secretion were measured (6). For induction of anergy in vivo, mice were injected intraperitoneally with the superantigen Staphylococcus aureus enterotoxin B (SEB; 100 μg in 100 μl of phosphate-buffered saline [PBS]; Sigma-Aldrich Canada Ltd., Oakville, Ontario, Canada) or with PBS alone (18). After 10 days, CD4⁺ T cells were purified from spleen and stimulated for 48 h with various concentrations of SEB (0.013 to 10 μg per ml) in the presence of irradiated splenocytes from C57BL/6 mice. Thymidine incorporation and IL-2 secretion were measured (6). The percentages of Vβ8.1/8.2-positive cells in purified CD4⁺ T cells were determined by staining with TCR Vβ8.1/8.2-specific MAb MR5-2 (BD Biosciences, Mississauga, Ontario, Canada). To induce anergy in vivo in transgenic mice expressing the class I major histocompatibility complex-restricted, LCMV gp33-specific TCR P14, mice (12 in each group) were injected intravenously with gp33 peptide (300 μg in PBS) or PBS alone on day 0 and day 3. On day 4, CD8⁺ T cells were purified by negative selection, using a Stem Cell purification kit. In all cases, flow cytometry analyses confirmed that more than 90% of purified CD8⁺ T cells expressed the transgenic anti-gp33 TCR (data not shown). Anergy induction was verified by stimulating purified CD8⁺ T cells with gp33 peptide (lysine-alanine-valine-tyrosine-asparagine-phenylalanine-alanine-threonine-methionine; 0.1 μM) and irradiated syngeneic splenocytes and measuring proliferation and IL-2 secretion. Purified CD8⁺ T cells were then processed for biochemical analyses as detailed above.

RESULTS

The interaction between PAG and FynT is dynamically regulated in response to TCR stimulation.

Previous studies provided evidence that PAG-associated Csk plays an inhibitory role during T-cell activation (4, 6). Herein, we investigated the function of PAG-associated FynT. To this end, we first examined the impact of TCR stimulation on the PAG-FynT interaction (Fig. 1). Normal mouse thymocytes were stimulated for various periods of time at 37°C by using biotinylated anti-TCR MAb H57-597 and avidin. They were then lysed, and the PAG-FynT interaction was monitored by probing anti-PAG immunoprecipitates with an anti-FynT immunoblot (Fig. 1A, first panel). The association of PAG with Csk was also examined in parallel, by immunoblotting PAG immunoprecipitates with anti-Csk (Fig. 1A, second panel). This experiment showed that TCR engagement (Fig. 1A, lanes 2 to 6) caused a pronounced (∼80%) and rapid decrease in the association of PAG with FynT (Fig. 1A, first panel), in comparison to no stimulation (Fig. 1A, first panel, lane 1). This effect was seen within 30 seconds of TCR ligation (Fig. 1A, first panel, lane 2) and was maximal at 1 minute (Fig. 1A, first panel, lane 3). Interestingly, the loss of the PAG-FynT association preceded the disappearance of the interaction between PAG and Csk (Fig. 1A, second panel). In the case of PAG-Csk, little or no change was occurring at 30 seconds (Fig. 1A, second panel, lane 2) and maximal dissociation required 2 minutes of stimulation (Fig. 1A, second panel, lane 4). These results are represented graphically in Fig. 1C. The dissociation of PAG-FynT also preceded the induction of overall protein tyrosine phosphorylation in T cells, as revealed by immunoblotting total cell lysates with antiphosphotyrosine antibodies (Fig. 1B).

FIG. 1.

Impact of TCR stimulation on the interactions of PAG with FynT and Csk and on PAG tyrosine phosphorylation. Thymocytes from normal mice were stimulated for the indicated periods of time at 37°C (A to C) or room temperature (D) with biotinylated anti-TCR and avidin. Lysates were then processed for immunoprecipitation or immunoblotting. (A) PAG-FynT interaction, PAG-Csk interaction, and PAG tyrosine phosphorylation. Cells were stimulated at 37°C. The associations of PAG with FynT and Csk were determined by immunoblotting anti-PAG immunoprecipitates with anti-FynT (first panel) and anti-Csk (second panel), respectively. Overall tyrosine phosphorylation of PAG was determined by immunoblotting PAG immunoprecipitates with antiphosphotyrosine (P.tyr) (third panel), whereas phosphorylation of PAG at Y314 was ascertained by probing total cell lysates with an antibody directed against phospho-Y314 of PAG (pY314) (fourth panel). The abundances of PAG, FynT, and Csk were confirmed by immunoblotting total cell lysates with anti-PAG (fifth panel), anti-FynT (sixth panel), and anti-Csk (seventh panel), respectively. (B) Overall protein tyrosine phosphorylation. Total cell lysates from the experiment whose results are depicted in panel A were immunoblotted with antiphosphotyrosine (P.tyr) antibodies. The migrations of prestained molecular mass markers (in kDa) are shown on the right. (C) Quantitation of the extents of PAG-FynT interaction, PAG-Csk interaction, and phosphorylation of Y314 of PAG (pY314) in the experiment whose results are shown in panel A. (D) Quantitation of the extents of PAG-FynT interaction, PAG-Csk interaction, and phosphorylation of Y314 of PAG (pY314). This experiment was performed as for panels A and B, except that cells were stimulated at room temperature (21°C) (data not shown).

The kinetics of dissociation of the PAG-FynT and PAG-Csk complexes in response to TCR stimulation were also compared to that of overall dephosphorylation of PAG (Fig. 1A, third panel) as well as dephosphorylation of PAG at Y314, the site implicated in Csk binding (Fig. 1A, fourth panel) (4, 6, 16). Probing of anti-PAG immunoprecipitates with antiphosphotyrosine antibodies (Fig. 1A, third panel) or of total cell lysates with an antibody specifically recognizing phosphorylated Y314 of PAG (anti-pY314) (Fig. 1A, fourth panel) indicated that TCR-triggered dephosphorylation of PAG and, in particular, Y314 of PAG occurred after loss of PAG-FynT complexes but simultaneously with dissociation of PAG-Csk (see Fig. 1C for quantitation of Y314 dephosphorylation).

To confirm the differences in the kinetics of dissociation of PAG-FynT, dissociation of PAG-Csk, and dephosphorylation of Y314, this experiment was also conducted by stimulating cells at room temperature rather than 37°C (Fig. 1D). This is known to slow down the onset of TCR-triggered signaling events and to delay dephosphorylation reactions. Under these conditions, we found that dissociation of the PAG-FynT complex was maximal after 1 to 2 min of TCR stimulation. In contrast, loss of the PAG-Csk association and Y314 dephosphorylation were delayed, reaching their maxima after 6 min. These observations firmly supported the notion that loss of the PAG-FynT interaction in response to TCR stimulation preceded dephosphorylation of PAG Y314 and disappearance of the PAG-Csk association.

One interpretation of these results is that dissociation of PAG-FynT may lead to dephosphorylation of Y314 and, consequently, dissociation of PAG-Csk. To address further this possibility, the impact of FynT expression on phosphorylation of Y314 was directly examined by immunoblotting total cell lysates from fyn^+/+ and fyn^−/− T cells with the phospho-specific anti-pY314 antibody (Fig. 2A, first panel). We found that cells lacking FynT, both thymocytes (Fig. 2A, lanes 3 and 4) and lymph node T cells (Fig. 2A, lane 6), showed marked decreases in Y314 phosphorylation, in comparison to cells from wild-type mice (Fig. 2A, lanes 1, 2, and 5). It is noteworthy, however, that some degree of PAG tyrosine phosphorylation remained in unstimulated FynT-deficient T cells (Fig. 2A, lanes 3 and 6). This residual tyrosine phosphorylation was partially affected by TCR stimulation (Fig. 2A, lane 4) and was equivalent to the extent of Y314 tyrosine phosphorylation remaining in TCR-stimulated normal T cells (Fig. 2A, lane 2). It also correlated with a residual binding of PAG to Csk in FynT-deficient T cells (Fig. 2A, fourth panel, compare lane 3 with lane 1). While the precise basis of these findings is not known, they likely reflect the existence of a FynT-independent, TCR-regulated pool of tyrosine-phosphorylated PAG in T cells. Thus, combined with the data shown in Fig. 1 and previously published results (37), those shown in Fig. 2 were consistent with the idea that dissociation of the PAG-FynT complex may be responsible for loss of the PAG-Csk interaction during T-cell activation.

FIG. 2.

Impact of FynT deficiency on phosphorylation of PAG at tyrosine 314. T cells from wild-type (fyn^+/+) or Fyn-deficient (fyn^−/−) mice were stimulated or not for 2 min with biotinylated anti-CD3 and avidin. (A) Phosphorylation of PAG at Y314. The extent of phosphorylation of PAG at Y314 was determined by immunoblotting total cell lysates with anti-pY314 (first panel). The interactions of PAG with FynT and Csk were determined by immunoblotting PAG immunoprecipitates with anti-FynT (third panel) and anti-Csk (fourth panel), respectively. The amount of PAG in these immunoprecipitates was verified by reprobing with anti-PAG (fifth panel). The abundances of PAG, FynT and Csk were verified by immunoblotting total cell lysates with anti-PAG (second panel), anti-FynT (sixth panel), and anti-Csk (seventh panel), respectively. Lanes 1 to 4, thymocytes; lanes 5 and 6, purified lymph node T cells. (B) Overall protein tyrosine phosphorylation. Total cell lysates from the experiment whose results are depicted in panel A (lanes 1 to 4) were immunoblotted with antiphosphotyrosine (P.tyr) antibodies. The positions of prestained molecular mass markers (in kDa) are shown on the right.

The PAG-FynT association requires tyrosines in the cytoplasmic domain of PAG and is needed for Csk binding.

To elucidate the mechanism by which PAG interacts with FynT in T cells, the PAG-FynT interaction was analyzed in T cells from transgenic mice expressing wild-type PAG, PAG Y314F, a PAG mutant in which all cytoplasmic tyrosines were replaced by phenylalanines (PAG 9YF), or a PAG variant in which all cytoplasmic tyrosines other than Y314 were mutated to phenylalanines (PAG 8YF) (Fig. 3) (6). Immunoblotting PAG immunoprecipitates with anti-FynT (Fig. 3A, first panel) revealed that T cells with augmented levels of expression of wild-type PAG (Fig. 3A, first panel, lane 2) exhibited a marked increase in the association of PAG with FynT, compared to T cells from nontransgenic control mice (Fig. 3A, first panel, lane 1). A similar increase was seen in T cells expressing PAG Y314F (Fig. 3A, first panel, lane 3). However, the PAG-FynT association was not augmented in T cells expressing PAG 9YF (Fig. 3A, first panel, lane 4) or PAG 8YF (Fig. 3A, first panel, lane 5), even though the two mutants were clearly expressed in amounts similar to those for wild-type PAG (Fig. 3A, fourth panel, lane 2) and PAG Y314F (Fig. 3A, fourth panel, lane 3). Therefore, one or more tyrosines of PAG, other than Y314, were implicated in the PAG-FynT interaction.

FIG. 3.

Associations of various PAG mutants with FynT and Csk in T cells. (A) Associations of PAG with FynT and Csk. T cells from nontransgenic (control) mice or transgenic mice expressing the indicated PAG polypeptides were lysed and processed for immunoprecipitation. The associations of PAG with FynT and Csk were determined by immunoblotting anti-PAG immunoprecipitates (lanes 1 to 5) with anti-FynT (first panel) and anti-Csk (second panel), respectively. Lane 6 shows an immunoprecipitation of lysates from control mice with normal rabbit serum. Overall phosphorylation of PAG was ascertained by immunoblotting anti-PAG immunoprecipitates with antiphosphotyrosine (P.tyr) antibodies (third panel). The abundance of PAG in the immunoprecipitates was verified by probing these immunoprecipitates with anti-PAG (fourth panel). The abundances of FynT and Csk were confirmed by immunoblotting total cell lysates with anti-FynT (fifth panel) and anti-Csk (sixth panel), respectively. PAG wt, wild-type PAG. (B) Phosphorylation of PAG polypeptides at Y314. The extents of the phosphorylations of the various PAG polypeptides at Y314 were determined by immunoblotting total cell lysates with anti-pY314 (first panel). The abundance of PAG was verified by immunoblotting total cell lysates with anti-PAG (second panel). (C) Cell fractionation. The extents of the associations of the various PAG polypeptides with lipid rafts were determined by cell fractionation studies and anti-PAG immunoblot analyses, as detailed in Materials and Methods. Fractions 2 and 3 correspond to lipid rafts, while fractions 8 and 9 contain soluble, non-lipid raft-associated proteins. The percentages of PAG in the two compartments were determined by a PhosphorImager and are shown at the bottom.

We also tested the impacts of these mutations on the capacity of PAG to bind Csk (Fig. 3A, second panel) and to undergo tyrosine phosphorylation (Fig. 3A, third panel). Immunoblotting anti-PAG immunoprecipitates with anti-Csk (Fig. 3A, second panel) demonstrated that overexpression of wild-type PAG (Fig. 3A, second panel, lane 2) greatly augmented the quantity of PAG-associated Csk. However, this enhancement was abrogated in cells expressing PAG Y314F (Fig. 3A, second panel, lane 3) or PAG 9YF (Fig. 3A, second panel, lane 4). These results were consistent with those reported elsewhere (6). In addition, we found that mutation of all tyrosines other than Y314 (PAG 8YF) (Fig. 3A, second panel, lane 5) also abolished the increase in PAG-associated Csk. This observation indicated that, whereas Y314 was necessary for Csk binding, it was not sufficient. When anti-PAG immunoprecipitates were probed by antiphosphotyrosine immunoblotting (Fig. 3A, third panel), we observed that cells with enforced expression of wild-type PAG (Fig. 3A, third panel, lane 2) exhibited a strong increase in the abundance of tyrosine-phosphorylated PAG, in keeping with our earlier results (6). An appreciable, albeit less pronounced, increase was also seen in cells expressing PAG Y314F (Fig. 3A, third panel, lane 3). In contrast, no augmentation was seen in cells expressing PAG 9YF (Fig. 3A, third panel, lane 4) or PAG 8YF (Fig. 3A, third panel, lane 5).

To assess whether the tyrosines other than Y314 contributed to Csk recruitment by promoting the FynT-dependent phosphorylation of Y314, the impacts of the various mutations on Y314 phosphorylation were ascertained by immunoblotting cell lysates with the phospho-specific anti-pY314 antibody (Fig. 3B, first panel). As expected, mutation of Y314 alone (Fig. 3B, first panel, lane 3) or replacement of all cytoplasmic tyrosines of PAG, including Y314 (Fig. 3B, first panel, lane 4), abolished the increased phosphorylation at Y314. Surprisingly, this reactivity was also eliminated when all cytoplasmic tyrosines except Y314 (Fig. 3B, first panel, lane 5) were mutated. Coupled with the observation that FynT was needed for Y314 phosphorylation, this result provided a compelling indication that binding of PAG to FynT is required for phosphorylation of PAG at Y314 and the subsequent association of PAG with Csk.

To ensure that the differences in the associations of the various PAG polypeptides with FynT and Csk were not due to variations in their associations with lipid rafts, cell fractionation experiments using T cells from the various mice were performed, in order to separate lipid raft and soluble proteins (Fig. 3C). In these studies, fractions 2 and 3 contain most lipid raft-associated proteins, while fractions 8 and 9 bear most non-lipid raft-associated proteins (data not shown). The distribution of PAG in these fractions was examined by immunoblotting with anti-PAG. This analysis showed that, for all the various PAG polypeptides, >80% were located in lipid raft fractions (Fig. 3C, lanes 1 to 4). This is consistent with previous assessments of the extents of the associations of endogenous PAG molecules in T cells with lipid raft fractions (4, 6). Thus, it is improbable that variations in the extents of targeting of the various PAG polypeptides to lipid rafts explained the differences in their associations with FynT and Csk.

Evidence that PAG-associated FynT selectively promotes antigen receptor-triggered calcium fluxes in T cells.

These findings showed that one of the roles of PAG-associated FynT is to regulate the recruitment of Csk to PAG. To evaluate whether this pool of FynT molecules may have additional functions, we studied TCR-mediated responses in T cells from transgenic mice expressing wild-type PAG (which binds FynT and Csk), PAG Y314F (which binds FynT but not Csk), or PAG 9YF (which binds neither FynT nor Csk). Along these lines, it is noteworthy that we had already reported that TCR-triggered T-cell proliferation and IL-2 secretion were reduced in T cells expressing wild-type PAG (6). In contrast, these responses were enhanced in T cells containing PAG Y314F or PAG 9YF. These data had implied that wild-type PAG had an inhibitory effect on TCR signaling and that this effect was largely dependent on the capacity of PAG to recruit Csk.

We extended these findings by studying the impacts of the various PAG polypeptides on TCR-triggered calcium fluxes (Fig. 4). Purified splenic and lymph node T cells were loaded with the calcium indicator dye Indo-1, and T cells were stimulated at 37°C with biotinylated anti-TCR and avidin. Changes in intracellular calcium over time were monitored using flow cytometry, by gating on single-positive CD4⁺ T cells. In agreement with our earlier results (6), cells with augmented levels of expression of wild-type PAG (Fig. 4B) exhibited a striking inhibition of TCR-triggered calcium fluxes, in comparison to control nontransgenic cells (Fig. 4A). In contrast, cells containing PAG Y314F (Fig. 4C) had an enhancement of TCR-initiated calcium fluxes, compared to control cells (Fig. 4A). Unexpectedly, however, we found that cells bearing PAG 9YF (Fig. 4D) had normal TCR-mediated calcium changes. These differences were not due to variations between pools of intracellular calcium in the various T cells, since all cells responded equally well to ionomycin, as reported elsewhere (6) (data not shown). Similar results were obtained when calcium responses in thymocytes or CD8⁺ T cells were studied (data not shown).

FIG. 4.

Regulation of TCR-induced calcium fluxes by various PAG polypeptides. Purified splenic and lymph node T cells from nontransgenic (control) mice or transgenic mice expressing the indicated PAG polypeptides were loaded with Indo-1 and were stimulated at 37°C with biotinylated anti-TCR and avidin. Changes in intracellular calcium were monitored using a BD LSR cell analyzer, by gating on CD4⁺ T cells. The ratio of bound Indo-1 (read in the FL4 channel)/free Indo-1 (read in the FL5 channel) is shown on the ordinate. The arrow corresponds to the moment at which the biotinylated anti-TCR antibody and avidin were added. Cells were observed for a total of 10 min. Similar results were obtained when thymocytes or CD8⁺ T cells were analyzed (data not shown). All cells exhibited equivalent calcium fluxes when they were stimulated with ionomycin, as reported elsewhere (6) (data not shown). The various transgenic mice used expressed equivalent amounts of PAG protein (data not shown).

To demonstrate that the effect of PAG Y314F on calcium fluxes was caused by FynT, transgenic mice expressing PAG Y314F were crossed with FynT-deficient mice and calcium responses in these animals were studied (Fig. 5). In agreement with results published by others (1, 27), lymph node T cells lacking FynT expression on their own (Fig. 5C) had no appreciable defect in TCR-initiated calcium fluxes, compared to cells from normal mice (Fig. 5A). Nonetheless, lack of FynT completely abolished the augmented calcium response seen in PAG Y314F-expressing T cells (compare Fig. 5D with Fig. 5B). This result firmly established that the increase in TCR-mediated calcium fluxes induced by PAG Y314F was caused by the enhanced binding of PAG to FynT.

FIG. 5.

Role of FynT in the impact of PAG Y314F on TCR-triggered calcium fluxes. The experiment was performed as explained for Fig. 4, except that mice expressing PAG Y314F in the presence or in the absence of endogenous FynT were analyzed, and biotinylated anti-CD3 was used. All cells responded equally well to ionomycin (data not shown). T cells from transgenic mice expressed equivalent amounts of PAG Y314F and contained FynT or not in agreement with their genotypes (data not shown).

To ascertain whether the effect of PAG Y314F was limited to an increase in TCR-mediated calcium fluxes, other TCR-triggered biochemical signals were also studied (Fig. 6). To this end, peripheral T cells were stimulated for various periods of time at 37°C with biotinylated anti-CD3 and anti-CD28 plus avidin. After lysis, total cell lysates were immunoblotted with various antibodies. An antiphosphotyrosine immunoblot (Fig. 6, first panel) showed that, in comparison to control nontransgenic T cells (Fig. 6, first panel, lanes 1 to 4), T cells expressing PAG Y314F (Fig. 6, first panel, lanes 5 to 8) exhibited a moderate and global increase in protein tyrosine phosphorylation, as reported elsewhere (6). In contrast, immunoblotting with an antibody recognizing activated Erk (Fig. 6, second panel) indicated that cells expressing PAG Y314F had a reduction, rather than an augmentation, of Erk activation. Conversely, immunoblotting with antibodies against phospho-IκB (Fig. 6, fourth panel) and total IκB (Fig. 6, fifth panel) showed no convincing difference in IκB activation and IκB degradation. Since IκB degradation predicates the activation of NF-κB during T-cell activation (35), this suggested that PAG Y314F did not affect NF-κB activation. Thus, together with the results shown in Fig. 4 and 5, those shown in Fig. 6 supported the idea that augmented levels of PAG-associated FynT, in the absence of elevated PAG-Csk, caused an increase in TCR-triggered protein tyrosine phosphorylation and calcium fluxes, coupled with a decrease in Erk activation and no effect on NF-κB.

FIG. 6.

Impact of PAG Y314F on other TCR-triggered signals. T cells were purified from spleen and lymph nodes of the indicated mice and stimulated for various periods of time at 37°C with biotinylated anti-CD3 and anti-CD28, followed by treatment with avidin. Global protein tyrosine phosphorylation was ascertained by immunoblotting total cell lysates with antiphosphotyrosine (P.tyr) antibodies (first panel). Activation of Erk (p42 and p44) was determined by probing total cell lysates with anti-phospho-Erk (second panel), whereas the abundance of Erk was verified by probing lysates with an antibody that recognized all forms, phosphorylated or not, of Erk (third panel). Phosphorylation of IκBα was examined by immunoblotting total cell lysates with an antibody directed against phospho-IκBα, while the abundance of IκBα was ascertained by probing lysates with an antibody recognizing all forms of IκBα. The migrations of prestained molecular mass markers (in kDa) are indicated on the right.

Changes in the abundance of PAG-associated FynT are linked to T-cell anergy.

Strong calcium signaling in the presence of weaker activation of other TCR-driven signals, such as NF-κB and AP-1, can lead to a state of T-cell unresponsiveness known as anergy (or adaptive tolerance) (15, 17). In a variety of systems, it was observed that T-cell anergy is accompanied by an increase in the expression levels of FynT, raising the possibility that FynT may be an effector of anergy (3, 9, 23, 30, 31). To ascertain whether a selective increase in the association of PAG with FynT may favor anergy, we first examined whether anergy was accompanied by a change in the stoichiometry of this association, using the ionomycin-induced T-cell anergy model (Fig. 7A and B) (15, 17). Ionomycin is a very efficient inducer of T-cell anergy, seemingly because it causes strong calcium-dependent signals in the absence of parallel activation of other signaling pathways, such as NF-κB (10, 17).

FIG. 7.

Influence of anergy on the associations of PAG with FynT and Csk. (A) Ionomycin-induced anergy. CD4⁺ T cells from normal mice were activated with anti-CD3 plus anti-CD28 and then rested in IL-2, as detailed in Materials and Methods. They were subsequently treated or not for 16 h with ionomycin (1 μM). The associations of PAG with FynT and Csk were determined by immunoblotting anti-PAG immunoprecipitates with anti-FynT (first panel) and anti-Csk (second panel), respectively. The abundance of PAG in these immunoprecipitates was verified by probing with anti-PAG (third panel). The abundances of FynT and Csk were confirmed by immunoblotting total cell lysates with anti-FynT (fourth panel) and anti-Csk (fifth panel), respectively. (B) Ionomycin-induced anergy. Cells from the experiment whose results are depicted in panel A were restimulated or not for 42 h with anti-CD3 and anti-CD28. Thymidine incorporation and IL-2 release were monitored as detailed in Materials and Methods. Standard deviations of the triplicate values in the assays are shown. (C) Soluble-peptide-induced anergy. Transgenic mice expressing the LCMV gp33-specific TCR (12 in each group) were injected intravenously with soluble gp33 peptide (300 μg in PBS) or PBS alone (control) on day 0 and day 3. On day 4, CD8⁺ T cells were purified by negative selection. Flow cytometry analyses confirmed that equivalent proportions (>90%) of purified CD8⁺ T cells from the two groups expressed the gp33-specific transgenic TCR (data not shown). The association of PAG with FynT and Csk was determined by immunoblotting anti-PAG immunoprecipitates with anti-FynT (first panel) and anti-Csk (second panel), respectively. The abundance of PAG in these immunoprecipitates was verified by probing with anti-PAG (third panel). The abundances of FynT and Csk were confirmed by immunoblotting total cell lysates with anti-FynT (fourth panel) and anti-Csk (fifth panel), respectively. (D) Soluble peptide-induced anergy. Cells from the experiment whose results are depicted in panel C were restimulated or not for 31 h with gp33 peptide (0.1 μM) and irradiated syngeneic splenocytes. Thymidine incorporation and IL-2 secretion were monitored as detailed in Materials and Methods. Standard deviations of the triplicate values in the assays are shown.

Thus, normal mouse T cells were treated or not for 16 h with ionomycin, and the extents of the PAG-FynT and PAG-Csk interactions were subsequently studied (Fig. 7A). This analysis demonstrated that, compared to untreated cells (Fig. 7A, first panel, lane 1), cells exposed to ionomycin (Fig. 7A, first panel, lane 2) exhibited a twofold increase in the association of PAG with FynT. This finding was reproduced in other experiments (data not shown). In contrast, there was no change in the interaction of PAG with Csk (Fig. 7A, second panel). Moreover, while the total cellular levels of FynT (Fig. 7A, fourth panel) and Csk (Fig. 7A, fifth panel) were unchanged, there was a decrease, rather than an increase, in the abundance of PAG (Fig. 7A, third panel). While the significance of the latter finding is not known, it did not reflect a change in the distribution of PAG in lipid rafts, as cells were lysed in a buffer containing maltoside that efficiently extracts lipid raft-associated proteins. The efficient induction of anergy by ionomycin treatment was confirmed by stimulating T cells in vitro with anti-CD3 plus anti-CD28 antibodies and measuring the subsequent reduction of proliferation and IL-2 secretion (Fig. 7B). Thus, an increase in the association of PAG with FynT, but not Csk, may indeed be linked to the induction and/or maintenance of T-cell anergy.

We also studied whether similar changes occurred in an in vivo model of T-cell anergy (Fig. 7C and D). For this purpose, transgenic mice expressing the class I-restricted, LCMV gp33-specific TCR P14 were injected or not intravenously on day 0 and day 3 with soluble gp33 peptide. On day 4, CD8⁺ T cells were purified, and the extents of the associations of PAG with FynT and Csk were examined (Fig. 7C). We observed that the amount of PAG-associated FynT was increased approximately twofold in CD8⁺ T cells from peptide-treated mice (Fig. 7C, first panel, lane 2), compared to that in CD8⁺ T cells from control PBS-injected mice (Fig. 7C, first panel, lane 1). In contrast, the amount of PAG-associated Csk was dramatically reduced (Fig. 7C, second panel). The increase in PAG-associated FynT in CD8⁺ T cells from peptide-injected mice correlated with an increment in the abundance of FynT in total cell lysates (Fig. 7C, fourth panel), despite the fact that, as was the case for the ionomycin-triggered anergy (Fig. 7A), there was a pronounced reduction of the abundance of PAG in T cells from peptide-injected mice (Fig. 7C, third panel, lane 2). However, there was no change in the abundance of Csk in these lysates (Fig. 7C, fifth panel). To verify the efficiency of the protocol used for anergy induction in vivo, CD8⁺ T cells from the treated mice were also stimulated in vitro with gp33 peptide plus irradiated syngeneic splenocytes, and T-cell proliferation and IL-2 secretion were determined (Fig. 7D). This experiment demonstrated that, compared to CD8⁺ T cells from control mice, CD8⁺ T cells from peptide-injected mice exhibited a reduction of T-cell proliferation and, to a greater extent, IL-2 secretion.

Evidence that PAG-associated FynT can promote T-cell anergy.

Considering these observations, we addressed the effect of PAG-associated FynT on T-cell anergy by evaluating the impacts of PAG Y314F and the other PAG polypeptides in two different models of anergy (Fig. 8 and 9). First, we examined the influence of these PAG molecules in an in vitro model of T-cell anergy (Fig. 8) (14). CD4⁺ T cells from the various mice were first activated in vitro for 2 days by a combination of anti-CD3 and anti-CD28, subsequently expanded for 3 days in the presence of IL-2, and then stimulated or not for 16 h by anti-CD3 alone to induce anergy. After an additional rest period of 30 h, cells were restimulated for 48 h with anti-TCR and anti-CD28 and assayed for proliferation or IL-2 secretion.

FIG. 8.

Effects of various PAG polypeptides and FynT on anti-CD3-induced T-cell anergy in vitro. CD4⁺ T cells from nontransgenic (control) mice or transgenic mice expressing the indicated PAG polypeptides, with or without FynT, were first activated and then rested in IL-2, as detailed in Materials and Methods. They were then treated or not with anti-CD3 (1 μg per ml), rested for an additional 30 h, and then restimulated with anti-CD3 and anti-CD28. Proliferation was monitored by assaying tritiated thymidine incorporation (A and C), while IL-2 secretion was determined by an enzyme-linked immunosorbent assay (B and D). The anergy index is the ratio between the thymidine incorporation or IL-2 secretion levels of untreated and anti-CD3-treated cells. CPM, counts per minute; PAG wt, wild-type PAG. Standard deviations for the triplicate values in the assays are shown. (A and B) Impacts of various PAG polypeptides. (C and D) Role of FynT in the effect of PAG Y314F.

FIG. 9.

Effects of PAG Y314F and FynT on superantigen-induced T-cell anergy in vivo. Nontransgenic (control) mice or transgenic mice expressing the indicated PAG polypeptides, with or without FynT, were injected intraperitoneally with SEB (100 μg) or PBS alone. After 10 days, CD4⁺ T cells were isolated from spleen and incubated for 48 h with SEB (0.12 μg per ml) in the presence of irradiated splenocytes from C57BL/6 mice as antigen-presenting cells. Proliferation was monitored by assaying tritiated thymidine incorporation (A), while IL-2 secretion was determined by an enzyme-linked immunosorbent assay (B). The anergy index is the ratio between the thymidine incorporation or IL-2 secretion levels of PBS-injected and SEB-injected mice. The proportions of Vβ8.1/8.2-positive cells in purified CD4⁺ T cells from the various mice were as follows (data not shown): for PBS-treated mice, 17.9% (control), 17.4% (PAG Y314F), 17.7% (fyn^−/−), and 17.1% (PAG Y314F fyn^−/−), and for SEB-treated mice, 13.8% (control), 15.3% (PAG Y314F), 15.5% (fyn^−/−), and 13.0% (PAG Y314F fyn^−/−). CPM, counts per minute. PAG wt, wild-type PAG. Standard deviations for the triplicate values in the assays are shown.

As expected, in control mice, anergy induction by anti-CD3 suppressed T-cell proliferation (Fig. 8A) and IL-2 secretion (Fig. 8B) upon restimulation with anti-TCR plus anti-CD28. The ratio of thymidine incorporation in cells treated with medium alone over that in cells exposed to anti-CD3, called the “anergy index,” was 5.6 (Fig. 8A). The anergy index was observed to be reduced to 2.5 in cells expressing augmented levels of wild-type PAG. This small reduction was mostly caused by a reduction of proliferation by nonanergized cells, reflecting the inhibitory effect of wild-type PAG on TCR signaling. More significantly, the anergy index was augmented to 114 in cells expressing PAG Y314F. This was primarily due to a decrease in the proliferation of anergized cells, although an increase in the proliferation of nonanergized cells was also seen. In contrast, the anergy index was only minimally increased (8.9) in cells expressing PAG 9YF. This slight increment solely reflected an augmented proliferation of nonanergized cells. Similar results were obtained when IL-2 secretion, instead of proliferation, was assayed (Fig. 8B).

These data showed that PAG Y314F, but not wild-type PAG or PAG 9YF, markedly stimulated the induction of anergy in response to anti-CD3. To examine whether this effect was mediated by FynT, similar experiments were conducted with mice expressing PAG Y314F that were bred or not with fyn^−/− mice (Fig. 8C and D). In PAG Y314F-expressing T cells, a lack of FynT expression caused a reduction of the anergy index from 79 to 1.4 in proliferation assays (Fig. 8C). The index became essentially identical to that seen in T cells from normal mice (1.7). These results indicated that, as was the case for the calcium fluxes, the augmented anergy caused by PAG Y314F was mediated by FynT. Equivalent results were obtained when IL-2 secretion was measured (Fig. 8D).

We also studied the impact of PAG Y314F, in the presence or in the absence of FynT, on anergy induction in vivo (Fig. 9) (18). Mice were injected with the superantigen SEB or with PBS, and after 10 days, CD4⁺ T cells were restimulated in vitro with various concentrations of SEB. They were then assayed for proliferation (Fig. 9A) or IL-2 secretion (Fig. 9B). Importantly, staining of purified CD4⁺ T cells from the diverse animals indicated that there was no marked difference between the relative abundances of Vβ8.1/8.2-positive T cells in mice injected with PBS or those injected with SEB (data not shown). This T-cell subset is known to mediate responses to SEB.

According to our expectation, injection of control mice with SEB caused a reproducible inhibition of superantigen-induced T-cell proliferation (Fig. 9A) or IL-2 release (Fig. 9B) in the restimulation assay, in comparison to injection with PBS alone. Here, the anergy index corresponds to the ratio between the thymidine incorporation or IL-2 secretion levels of PBS-injected and SEB-injected mice at a given concentration of SEB. In normal mice and at 0.12 μg per ml of SEB, these values were 1.7 for proliferation (Fig. 9A) and 4.0 for IL-2 production (Fig. 9B). They were enhanced to 4.0 and 10.0, respectively, in mice expressing PAG Y314F. Similar differences were seen when higher concentrations of SEB were used for restimulation (data not shown). As was the case for the anti-CD3-triggered anergy, the impact of PAG Y314F on SEB-induced anergy was eliminated when mice were crossed with fyn^−/− mice. The values of the anergy index for these mice were 1.0 for proliferation (Fig. 9A) and 3.4 for IL-2 secretion (Fig. 9B). It is noteworthy that SEB-mediated anergy was also abrogated in FynT-deficient mice in the absence of PAG Y314F, supporting a role for FynT in anergy induction even in the absence of PAG Y314F. Therefore, the results shown in Fig. 8 and 9 were consistent with the hypothesis that increased PAG-associated FynT, in the absence of augmented PAG-associated Csk, can promote T-cell anergy.

DISCUSSION

Previous reports provided evidence, using overexpression studies and RNA interference experiments, that PAG is an effective negative regulator of cellular processes mediated by Src-related PTKs (4, 6, 21, 25). This function was believed to be largely mediated by the binding of PAG to the inhibitory PTK Csk by way of Y314 of PAG. It was also reported that, in addition to Csk, PAG can associate with the Src-related kinase FynT (4). Earlier studies performed with T cells from FynT-deficient mice indicated that this interaction was critical for PAG tyrosine phosphorylation and, consequently, the interaction of PAG with Csk (37).

In the present report, we wanted to understand better the role and regulation of the PAG-FynT interaction. Our data showed that, like Csk, FynT was constitutively associated with PAG in unstimulated T cells. Moreover, as was the case for Csk, FynT became rapidly dissociated from PAG following TCR engagement. Interestingly, time course analyses indicated that dissociation of the PAG-FynT complex preceded PAG dephosphorylation, cleavage of the PAG-Csk interaction, and induction of overall protein tyrosine phosphorylation following TCR stimulation. Using an antibody specifically recognizing phosphorylated Y314 of PAG, it was also established that phosphorylation of Y314 was reduced in FynT-deficient T cells, extending previous findings that FynT was required for global PAG tyrosine phosphorylation (37). Lastly, we found that a PAG mutant (PAG 8YF) that cannot bind FynT but has a preserved Csk-binding site (Y314) was no longer phosphorylated at Y314 and was unable to interact with Csk in T cells. Presumably, the lack of phosphorylation of PAG 8YF at Y314 was due to the inability of this mutant to associate with FynT.

Based on these findings and previously published results (4, 6, 29), we propose that constitutive binding of PAG to FynT in resting T cells is critical for phosphorylation of PAG at Y314 and subsequent binding of PAG to Csk. We also postulate that, following TCR stimulation, dissociation of the PAG-FynT complex is instrumental in causing loss of PAG phosphorylation at Y314, disruption of the PAG-Csk association and, ultimately, induction of TCR-triggered protein tyrosine phosphorylation. Although the mechanism of PAG-FynT dissociation in response to TCR engagement is not known, possibilities include changes in the conformation of PAG or FynT, dephosphorylation of PAG at the tyrosines other than Y314, or posttranslational modifications of FynT, such as serine/threonine phosphorylation. Obviously, future studies are warranted to elucidate the mechanism of this dissociation.

The precise mechanism by which FynT interacts with PAG also remains to be clarified. Since this association required tyrosines in the cytoplasmic domain of PAG, it may be mediated, at least in part, by direct binding of the FynT SH2 domain to sites of tyrosine phosphorylation in PAG. In keeping with this, we have observed that recombinant FynT SH2 domains can interact with PAG in vitro and that mutation of all tyrosines of PAG other than Y314 (PAG 8YF mutation) abrogates this association (our unpublished results). It seems, however, that more than one tyrosine would be implicated, as mutation of the first five or the last three tyrosines of PAG does not abolish the association of PAG with FynT SH2 domains in vitro. It is possible that additional mechanisms, including binding of the FynT SH3 domain to proline-rich sequences in PAG or participation of intermediary proteins, are also implicated. Future studies are needed to address exhaustively these various possibilities.

We were interested in examining whether PAG-associated FynT may mediate other functions, in addition to allowing the recruitment of Csk via phosphorylation of Y314. To this end, the functional impact in normal T cells of the expression of wild-type PAG, a PAG mutant that can bind FynT but not Csk (PAG Y314F), or a PAG mutant that can bind neither FynT nor Csk (PAG 9YF) was examined. We observed that PAG Y314F caused a pronounced increase in TCR-triggered calcium fluxes. Importantly, this effect was not seen in T cells expressing PAG 9YF. Furthermore, it was eliminated when mice expressing PAG Y314F were crossed with fyn^−/− mice. Hence, it is likely that PAG-associated FynT was responsible for this augmentation of TCR-triggered calcium fluxes. In contrast to the enhanced calcium response, PAG Y314F had no effect on phosphorylation of IκB, a marker of NF-κB activation, and caused a reduction of Erk activation. Therefore, the stimulatory effect of PAG Y314F on calcium fluxes was rather specific. Although the basis for the concomitant inhibition of Erk activation remains to be established, it is possible that this effect was due to inhibition of Ras activation by calcium-triggered recruitment of CAPRI, a calcium-regulated Ras-GTPase-activating protein (2).

The mechanism by which PAG-associated FynT caused an increase in TCR-triggered calcium fluxes needs to be determined. Surprisingly, we did not detect any increase in TCR-induced tyrosine phosphorylation of phospholipase C-γ1 in T cells expressing PAG Y314F (our unpublished results). We were also not able to notice any change in the tyrosine phosphorylation of the inositol triphosphate receptor, which was previously reported to be a target of FynT (13). Moreover, we did not observe any impact on the increase of intracellular inositol triphosphate in response to TCR stimulation. Therefore, it is likely that another mechanism is involved, and future studies are warranted to resolve this issue.

In the light of the selective impact of PAG Y314F on TCR-triggered calcium fluxes, we examined the possibility that PAG-associated FynT may be implicated in the induction of T-cell anergy. This suggestion was prompted by earlier work indicating that TCR-mediated activation of calcium signaling in the absence of sufficient parallel activation of other pathways, such as NF-κB, can promote T-cell anergy (17). In a physiological setting, this is presumably occurring when the TCR is engaged by antigen in the absence of adequate costimulation through CD28. Preliminary support for the idea that PAG-associated FynT may indeed be implicated in anergy induction was obtained with our observation that the abundance of PAG-associated FynT, but not PAG-associated Csk, was augmented in T cells anergized in vitro with ionomycin or in vivo with soluble peptide antigen. Furthermore, by studying two different anergy models, anergy triggered by anti-CD3 antibodies or by superantigen (SEB), we observed that PAG Y314F, but not wild-type PAG or PAG 9YF, potently augmented anergy induction in T cells. As was the case for the enhanced calcium fluxes, the effect of PAG Y314F on anergy was eliminated in T cells lacking FynT, implying that it was mediated by PAG-associated FynT. Consequently, it seems likely that PAG-associated FynT can participate in anergy induction in T cells.

One issue that arises from these results is whether PAG Y314F mimics a bona fide effect of PAG-associated FynT on anergy. We believe this to be the case for the following reasons. First, previous studies with several anergy models revealed that FynT expression was greatly enhanced in anergic T cells (3, 9, 23, 30, 31). Second, expression of FynT was shown to be required for anergy induction in an unusual population of double-negative αβ T cells (30). This finding was extended by our demonstration that FynT expression was also important for superantigen-induced anergy in conventional αβ T cells, even in the absence of PAG Y314F. And third, in the ionomycin-induced and the soluble-peptide-triggered models of T-cell anergy, we documented an increased abundance of endogenous PAG-FynT complexes but not PAG-Csk complexes. Hence, it appears that FynT molecules bound to endogenous PAG can also participate in the induction of T-cell anergy. While it is difficult to demonstrate this idea further at this time, one possible approach would be to introduce “knock-in” germ line mutations of all tyrosines other than Y314 in the mouse pag gene. Nevertheless, since complete ablation of PAG expression in the mouse had little or no effect on T-cell functions (presumably because of redundancy or compensation by other signaling pathways) (8, 36), this approach may not be very informative.

These observations also imply that a proper balance between PAG-associated FynT and PAG-associated Csk is crucial to avoid induction of gratuitous T-cell anergy. In the presence of adequate recruitment of Csk to PAG, PAG-associated FynT is presumably maintained relatively inactive, thereby preventing anergy induction. This notion is in agreement with our observation that, contrary to PAG Y314F, wild-type PAG did not promote T-cell anergy. In this setting, the effect of PAG-associated Csk was probably dominant. The situation would change when PAG-associated FynT dominates over PAG-associated Csk, as seen with PAG Y314F or in ionomycin-triggered anergic T cells. Under physiological circumstances, this may occur because of a change in the relative abundances or activities of FynT and Csk or an alteration of their affinities for PAG.

Initial reports suggested that the exclusive role of PAG was to recruit the negative regulator of Src family kinases, Csk, to lipid rafts (4, 6, 16, 29). According to these studies, PAG-associated FynT was simply implicated in the phosphorylation of PAG at Y314, an event that enables binding of PAG to Csk. While the data presented here further supported the importance of PAG-associated FynT for Csk recruitment, they also provided evidence that PAG-associated FynT can transduce other biochemical signals during TCR signaling. These signals lead to a marked increase in TCR-induced calcium fluxes and can, under certain circumstances, promote T-cell anergy. In this manner, PAG is reminiscent of other lipid raft-associated adaptors, such as NTAL/LAB, SIT, and LAT, which can have dual, and at times opposing, functions in immune cells (26).

Our finding that PAG can generate multiple signals may also explain in part why little or no phenotype was observed in PAG-deficient mice (8, 36). Removal of opposing signals, as a result of PAG deficiency, may cause no appreciable net effect on cell signaling. The generation of knock-in mice in which PAG can bind only FynT or Csk may help address this intriguing idea. However, as alluded to already (8, 36), it is likely that other mechanisms of recruitment of Csk and FynT can compensate in the absence of PAG. Indeed, Csk can be recruited by several other molecules, including Dok-related adaptors, LIME, SIT, and paxillin (34). Likewise, FynT can be recruited by the TCR, the adaptor SAP, and others (24, 32). Clearly, more work remains to be done to elucidate the cooperative importance of these various mechanisms in the regulation and functions of Csk and FynT, not only in T cells but also in other cell types.