Cooperative Inhibition of  T-Cell Antigen Receptor Signaling by a Complex between a Kinase and a Phosphatase

Cells.

The previously described antigen-specific mouse T-cell hybridoma BI-141 was maintained in RPMI medium supplemented with 10% fetal bovine serum and antibiotics (13). BI-141 derivatives expressing wild-type Csk, ΔSH3 Csk, or the neomycin phosphotransferase alone (Neo) have been reported previously (5). All BI-141 clones were propagated in the presence of G418 (0.6 mg/ml). Cos-1 cells were grown in α-MEM containing 10% fetal bovine serum and antibiotics.

cDNAs and Antibodies.

The cDNAs for wild-type PEP, ΔP1 PEP, wild-type Csk, ΔSH3 Csk, wild-type FynT, kinase-inactive (lysine 295 to arginine 295) Zap-70, and Tac-ζ were reported previously (6, 8, 9, 14, 15). Those coding for the aspartate 195 to alanine 195 (D195A), cysteine 227 to serine 227 (C227S), and arginine 233 to methionine 233 (R233M) mutants of PEP were created by overlap extension PCR. All mutated cDNAs were sequenced to verify that they did not contain unwanted mutations (data not shown). Antibodies directed against PEP, Csk, PTP-PEST, phosphotyrosine, Zap-70, Tac (mAb 7G7), ζ (mAb H146), and Fyn were described elsewhere (6, 8, 9, 15–17).

Transfections.

BI-141 T cells were stably transfected by electroporation (260 V, 960 μF) and selected in the presence of G418 (0.75 mg/ml). Monoclonal cell lines were derived by limiting dilution. Cells were tested for PEP expression by immunoblotting with anti-PEP antibodies. All clones selected for our studies had unchanged levels of TCR, CD3, CD45, and Thy-1, and remained CD4 negative (data not shown). Cos-1 cells were transiently transfected with Lipofectamine-plus (Bethesda Research Laboratories) according to the protocol provided by the manufacturer, using a fixed total amount of DNA (8 μg). Cells were harvested 24–48 h after transfection.

Immunoprecipitations and Immunoblots.

Cells were washed with phosphate-buffered saline and lysed in TNE buffer (50 mM Tris-HCl, pH 8.0, 1% Nonidet P-40, 2 mM EDTA) supplemented with 10 μg/ml each of the protease inhibitors leupeptin, aprotinin, n-tosyl-l-phenylalanine chloromethyl ketone (TPCK), n-p-tosyl-l-lysine chloromethyl ketone (TLCK), and PMSF, as well as the phosphatase inhibitors sodium fluoride (50 mM) and sodium orthovanadate (1 mM). Immunoprecipitations were performed with the appropriate antibodies using equivalent amounts of cellular lysate. Immune complexes were recovered using Staphylococcus aureus protein A (Pansorbin; Calbiochem) or protein A–Sepharose (Pharmacia Biotech). After washing three times in TNE buffer containing sodium orthovanadate (1 mM), proteins were eluted in sample buffer, boiled, and subjected to SDS-PAGE. Immunoblots were performed according to a protocol detailed previously (15), using ¹²⁵I-protein A (Amersham Corp.) unless specified.

Lymphokine Secretion Assays.

Cells were stimulated in triplicate with the indicated concentrations of anti-TCR V_β8 mAb F23.1 coated on 96-well plastic plates. After a 24-h incubation, supernatants were harvested and frozen at −70°C to destroy any carry-over cells. The presence of secreted lymphokines in the supernatant was determined by measuring incorporation of tritiated thymidine in the IL-2–dependent indicator cell line HT-2.

TCR-mediated Protein Tyrosine Phosphorylation.

BI-141 derivatives were activated for the indicated periods of time at 37°C using biotinylated mAb F23.1 (30 μg/ml) and avidin (38 μg/ml). They were then lysed in TNE buffer supplemented with protease and phosphatase inhibitors, and equal amounts of total cellular lysate were subjected to immunoprecipitation using the protocol described above.

PTP Assay.

PEP variants were expressed by transient transfection in Cos-1 cells as explained earlier. After washing with PBS, they were lysed according to the usual protocol, except that the phosphatase inhibitors were omitted. PEP polypeptides were subsequently recovered by immunoprecipitation from 25 μg of cellular lysate. After collecting the immune complexes with S. aureus protein A, they were washed three times in lysis buffer, followed by two washes in phosphatase assay buffer (25 mM Tris-HCl, pH 7.2, 1 mg/ml BSA, 1 mM dithiothreitol). They were resuspended in phosphatase assay buffer and incubated for the indicated periods of time at 30°C with tyrosine phosphorylated myelin basic protein (5 μM). Reactions were stopped by the addition of 20% (wt/vol) trichloroacetic acid, followed by incubation on ice. After centrifugation, the release of ³²PO₄ in the supernatant was measured by scintillation counting. Immunoprecipitates obtained from cells transfected with vector alone were used to measure spontaneous hydrolysis. In our calculations, this amount was subtracted from those of PEP immunoprecipitates. Preparation of tyrosine phosphorylated myelin basic protein was performed using the Protein Tyrosine Phosphatase Assay System (New England Biolabs, Inc.), according to the protocol provided by the manufacturer. All experiments were conducted under linear assay conditions (data not shown).

Substrate Trapping.

Transfected Cos-1 cells were prepared as above, except that they were lysed in a buffer containing 50 mM Tris-HCl, pH 7.5, 1% Triton X-100, 200 mM sodium chloride, 5 mM iodoacetamide, and 1 mM benzamidine, supplemented with leupeptin, aprotinin, TPCK, TLCK, PMSF, and sodium fluoride. Subsequently, immunoprecipitations and immunoblots were performed using the standard protocols.

Metabolic Labeling and Peptide Mapping Studies.

For ³²P_i labeling, cells were incubated for 2 h in phosphate-free DMEM supplemented with 1.0 mCi of ³²P_i (carrier free; NEN Research Products) per ml, and 2% dialyzed FCS. Cells were then washed with phosphate-free medium and lysed in TNE buffer containing protease and phosphatase inhibitors, as described above. FynT polypeptides were recovered by immunoprecipitation and resolved in 8% SDS-PAGE. Cyanogen bromide cleavage was performed as described elsewhere (18). Phospholabeled FynT peptides were resolved in 18% SDS-polyacrylamide gels and detected by autoradiography.

Impact of PEP on Antigen Receptor–induced Lymphokine Secretion.

To evaluate the role of PEP in T lymphocytes, we opted to use the antigen-specific mouse T-cell line BI-141, as it has proven very useful in ascertaining the role and regulation of cytoplasmic PTKs in T lymphocytes (6, 8, 15, 19). Cells were transfected by electroporation with a cDNA coding for mouse PEP, and stable monoclonal cell lines were selected after growth in G418-containing medium. Derivatives overexpressing PEP were identified by immunoblotting of anti-PEP immunoprecipitates with anti-PEP antibodies. Representative clones are shown in Fig. 1 A. In comparison with control neomycin-resistant BI-141 cells (Neo; lanes 1–3), these cells (lanes 4–6) contained three to four times greater amounts of a 110-kD polypeptide consistent with PEP.

Next, it was determined whether PEP overexpression resulted in an increase in the abundance of Csk-PEP complexes (Fig. 1, B and C). When anti-Csk immunoprecipitates were probed by immunoblotting with anti-PEP antibodies (Fig. 1 B), we observed that PEP overexpression (lanes 5–8) induced a three- to fourfold increase in the quantity of PEP that was coimmunoprecipitated with Csk. As was the case for control Neo cells (lanes 1–4), ∼50–80% of PEP molecules contained in these cells were associated with Csk (lanes 5–8). In a similar way, enforced expression of PEP induced a proportional augmentation in the amount of Csk that could be coimmunoprecipitated with PEP (Fig. 1 C; compare lanes 2 and 6). Whereas ∼5% of Csk was associated with PEP in control Neo cells (lane 2), this fraction was elevated to ∼15% in PEP-overexpressing cells (lane 6). PEP overexpression did not affect the ability of Csk to associate with PTP-PEST, a PEP-related PTP expressed ubiquitously (compare lanes 3 and 7) (16, 20). Importantly, all cell lines used in our experiments expressed unaltered levels of TCR, CD3, and CD45 (data not shown).

The effect of PEP on antigen receptor–mediated lymphokine secretion, a late marker of T-cell activation (21, 22), was first evaluated (Fig. 1 D). Cells were stimulated with increasing concentrations of anti-TCR mAb F23.1 immobilized on plastic, and the release of lymphokines in the supernatant was monitored using a standard bioassay. This analysis revealed that, compared to Neo cells, cells overexpressing PEP had dramatically reduced lymphokine production in response to TCR stimulation. This diminution was observed in all clones tested, and at all concentrations of anti-TCR mAbs.

Structural Requirements for the Inhibitory Effect of PEP on T-Cell Activation.

To begin elucidating the mechanism by which PEP affected TCR signaling, the relevance of its phosphatase activity was determined. To this end, catalytically defective PEP mutants in which either cysteine 227 was replaced by serine (C227S), or arginine 233 was substituted by methionine (R233M), were created. As described for other phosphatases (23–25), mutation of these highly conserved residues located in the catalytic domain strongly diminished the ability of PEP to dephosphorylate the substrate myelin basic protein in an immune complex phosphatase assay (Fig. 2 B; data not shown). When introduced in BI-141 cells, the two PEP mutants (Fig. 2 A; lanes 7–11) were expressed in quantities comparable to those of wild-type PEP (lanes 4–6). Furthermore, they were able to associate with Csk with a similar stoichiometry (Fig. 2 C; lanes 9–12). However, they were incapable of inhibiting TCR-triggered lymphokine production (Fig. 2 D).

Then we examined the importance of the association of p50^csk with PEP by testing the function in BI-141 cells of a PEP mutant carrying a deletion in the proline-rich region mediating binding to Csk (ΔP1 PEP; Fig. 3 A) (9, 26). While this mutation did not affect the ability of PEP to dephosphorylate the exogenous substrate myelin basic protein in vitro (Fig. 2 B), it significantly reduced (approximately fivefold) its capacity to be coimmunoprecipitated with Csk (Fig. 2 C; lanes 13–16). By opposition to our earlier finding in Cos-1 cells (9, 26), there appeared to be a partial association of ΔP1 PEP with Csk in BI-141 cells. Although the exact basis for this difference is undetermined, it is possible that other structural determinants in PEP (such as the proline-rich domain P2) mediated binding to Csk in the T-cell line. In support of this notion, we reported elsewhere that PEP P2 could also associate with the Csk SH3 domain in vitro, albeit with a greatly reduced apparent affinity in comparison with P1 (26). Alternatively, it is conceivable that the presence of ΔP1 PEP in BI-141 cells augmented by an unknown mechanism the ability of endogenous PEP molecules to associate with Csk. In any case, we observed that ΔP1 PEP failed to inhibit antigen receptor–mediated lymphokine production in BI-141 cells (Fig. 3 B), indicating that the small (approximately twofold) increase in the abundance of Csk-associated PEP in these cells was insufficient to inhibit TCR signaling. The defective function of ΔP1 PEP was reminiscent of the inability of an SH3 domain–deleted version of Csk (ΔSH3 Csk) to suppress TCR-mediated lymphokine secretion in BI-141 cells (8; Fig. 3 B). From these data, we deduced that the inhibitory effect of PEP on TCR-mediated lymphokine secretion not only requires its phosphatase activity, but also is correlated with the capacity to associate with Csk.

Impact of PEP on Antigen Receptor–mediated Protein Tyrosine Phosphorylation.

The influence of PEP overexpression on TCR-mediated protein tyrosine phosphorylation was investigated next (Fig. 4). BI-141 cells were stimulated for various periods of time at 37°C with biotinylated anti-TCR mAb F23.1 and avidin. After stimulation, they were lysed in detergent-containing buffer, and tyrosine phosphorylated proteins were immunoprecipitated and detected by antiphosphotyrosine immunoblotting (Fig. 4 A). In contrast to Neo cells (lanes 1–6), cells overexpressing PEP (lanes 7–12) demonstrated a clear reduction in TCR- induced protein tyrosine phosphorylation. Tyrosine phosphorylation of most, if not all, TCR-regulated substrates was diminished approximately fourfold. The impact of PEP on tyrosine phosphorylation of Zap-70 and the ζ subunit of TCR, the two most proximal substrates in TCR signaling (21, 22), was also specifically evaluated. Antiphosphotyrosine immunoblotting of Zap-70 immunoprecipitates (Fig. 4 B) showed that PEP (lanes 5–8) markedly reduced the phosphotyrosine content of Zap-70 in TCR-stimulated BI-141 cells. Tyrosine phosphorylation of c-Cbl, a 120-kD adaptor molecule associated with Zap-70 (27, 28), was also lowered. In a similar way, PEP diminished tyrosine phosphorylation of the ITAMs of ζ (Fig. 4 C), the first biochemical change known to occur during T-cell activation (21, 22).

Lastly, the structural requirements for the diminution of TCR-induced protein tyrosine phosphorylation by PEP were ascertained (Fig. 4 D). This analysis showed that, unlike wild-type PEP (lanes 5–8), the two phosphatase-inactive variants of PEP (lanes 9–14), as well as ΔP1 PEP (lanes 15–18), were unable to inhibit TCR-triggered protein tyrosine phosphorylation.

Cooperative Inhibition of Protein Tyrosine Phosphorylation by PEP and Csk Is Enhanced by Their Physical Association.

These findings demonstrated that, in a manner analogous to Csk (6), PEP was a negative regulator of antigen receptor signaling in T cells. Together with previously published results (8), they also indicated that the domains mediating the physical association between Csk and PEP were critical for the respective inhibitory impacts of these two molecules. On this basis, it seemed plausible that Csk and PEP needed to be associated with each other in order to inhibit TCR-mediated signal transduction events. To further test this possibility, attempts were made to recreate the influence of Csk and PEP on TCR-mediated tyrosine phosphorylation events in a heterologous system (Fig. 5). The Cos-1 cell system was chosen for these studies, because it has been extensively validated to dissect the proximal events of TCR signaling (17, 29, 30). Cells were transiently transfected with cDNAs coding for a chimeric receptor containing the cytoplasmic portion of ζ (Tac-ζ), the Src family kinase FynT, and Zap-70. A kinase-inactive version of Zap-70 (lysine 295 to arginine 295 mutant) was used in these assays to ensure that baseline protein tyrosine phosphorylation was caused exclusively by FynT.

As described elsewhere (29), expression of wild-type FynT allowed tyrosine phosphorylation of the ζ chimera (Fig. 5 A; first panel, lane 1) and Zap-70 (second panel, lane 1). Neither substrate was phosphorylated in cells lacking FynT (data not shown). In the presence of FynT, Tac-ζ was also detectably associated with tyrosine phosphorylated Zap-70 molecules (first and second panels, lane 1). In cells overexpressing p50^csk alone (lane 2), there was a small (<25%) diminution in the phosphotyrosine content of Tac-ζ and Zap-70. This decrease was more marked (30– 40%) in cells containing PEP alone (lane 3), although residual phosphorylation of Tac-ζ and Zap-70 was clearly observed. However, in contrast, cells coexpressing Csk and PEP (lane 4) displayed a total abolition of tyrosine phosphorylation of both the chimeric receptor (first panel) and Zap-70 (second panel). Similar results were obtained in cells transfected with the Src-like enzyme Lck rather than FynT (data not shown).

Since this system recapitulated the effects of Csk and PEP on the initial events of TCR signaling, we wanted to test whether the ability of these enzymes to associate was important for their cooperation. Thus, mutants that were defective in this association (ΔSH3 Csk and ΔP1 PEP; 9) were evaluated (Fig. 5 B). Like wild-type p50^csk or PEP alone (Fig. 5 A, lanes 2 and 3), ΔSH3 Csk (Fig. 5 B, lane 3) or ΔP1 PEP (lane 4) alone had a partial effect on tyrosine phosphorylation of Tac-ζ (first panel) and Zap-70 (second panel). However, contrary to wild-type Csk and PEP together (lane 2), coexpression of ΔSH3 Csk and ΔP1 PEP (lane 5) could not eliminate tyrosine phosphorylation of Tac-ζ and Zap-70. Together, these data supported the notion that the optimal ability of PEP and Csk to inhibit protein tyrosine phosphorylation by Src family kinases required their physical association.

Identification of Potential PEP Substrates.

The ability of PEP overexpression in BI-141 cells to repress tyrosine phosphorylation of most TCR-regulated substrates, including Zap-70 and ζ, indicated that this phosphatase acted on a proximal step of TCR signaling. However, unfortunately, in light of the complex interplay between Src family kinases, ITAMs, and Zap-70, it was difficult to identify the immediate targets of PEP using the T-cell system. To help identify the PEP substrates, substrate-trapping experiments were conducted in the Cos-1 system (Fig. 6). Cells were transfected as described in Fig. 5, except that a PEP variant carrying a replacement of a conserved aspartate in the catalytic domain (aspartate 195 to alanine 195 mutant; D195A PEP) was used. Mutation of the equivalent residue in other PTPs has been shown to abolish their catalytic activity while preserving the ability to associate with substrates (24, 31–33). Such a property has provided a very useful tool in identifying targets of PTPs.

Immunoblotting of total cell lysates with antiphosphotyrosine antibodies showed that, when compared with wild-type PEP (Fig. 6 A, lane 3), D195A PEP (lane 5) was less efficient at repressing protein tyrosine phosphorylation in Cos-1 cells. Nonetheless, it was clearly able to reduce the tyrosine phosphorylation of certain substrates, especially Tac-ζ. Furthermore, the mutant could cooperate with Csk (lane 6) to inhibit protein tyrosine phosphorylation, albeit to a lesser degree than wild-type PEP (lane 4). The ability of D195A PEP to repress protein tyrosine phosphorylation may appear paradoxical, especially since the catalytic activity of this mutant was found to be abolished in in vitro phosphatase assays (data not shown). However, these results are consistent with the findings of others (33). Seemingly, substrate-trapping mutants of other PTPs can interfere with protein tyrosine phosphorylation as a result of steric hindrance.

To identify which proteins were the immediate substrates of PEP, PEP molecules were recovered by immunoprecipitation and probed by antiphosphotyrosine immunoblotting (Fig. 6 B, top panel). This experiment revealed that D195A PEP (lanes 5 and 6) associated with a major tyrosine phosphorylated polypeptide of ∼70 kD. This association occurred independently of coexpression of Csk (compare lanes 5 and 6). Small quantities of a ∼59-kD phosphotyrosine–containing protein were also present in PEP immunoprecipitates. Unlike D195A PEP, wild-type PEP (lanes 3 and 4) was not associated with any detectable tyrosine phosphorylated protein. This difference was not caused by variations in the levels of expression of the two forms of PEP, as revealed by reprobing of the immunoblot membrane with anti-PEP serum (bottom panel).

To characterize the tyrosine phosphorylated proteins associated with D195A PEP, anti-PEP immunoprecipitates were also probed by immunoblotting with antibodies against either Zap-70, FynT, or Tac-ζ (Fig. 6 C). This analysis revealed that D195A PEP (lanes 5 and 6) formed a stable complex with Zap-70 (first panel) and FynT (second panel). However, it was not detectably associated with Tac-ζ (third panel). It is noteworthy that the ability of D195A PEP to associate with FynT and Zap-70 was not influenced by the presence of Csk (compare lanes 5 and 6), implying that Csk did not mediate these interactions. When compared with D195A PEP, wild-type PEP (lanes 3 and 4) bound much smaller quantities of Zap-70 and FynT, in keeping with the results obtained with substrate-trapping mutants of other phosphatases (31, 32). Thus, the results of these experiments suggested that PEP inhibited TCR signaling by dephosphorylating Zap-70 and FynT, but not ζ.

PEP Induces Dephosphorylation of the Positive Regulatory Site of FynT, Tyrosine 417.

Like other Src family kinases, FynT is regulated by phosphorylation of two distinct tyrosine residues (5, 34, 35). Phosphorylation of the major site of autophosphorylation, tyrosine 417, is involved in the activation of FynT. In contrast, phosphate occupancy of the COOH-terminal tyrosine residue, tyrosine 528, which is mediated by Csk, causes an inhibition of the catalytic function of FynT. To assess which of these residues was(were) targeted for dephosphorylation by PEP, peptide mapping studies were conducted (Fig. 7). Cos-1 cells were transfected with cDNAs coding for wild-type FynT, Tac-ζ and kinase-inactive Zap-70 as above, in the absence or presence of wild-type PEP. Subsequently, they were labeled metabolically with ³²P_i, and FynT molecules were isolated by immunoprecipitation (Fig. 7 A). Gel electrophoresis revealed that FynT obtained from cells lacking PEP (lane 1) migrated as a phosphorylated product of 59 kD, in addition to a species of ∼65 kD (FynT*). The existence of a similar form of FynT, having a retarded electrophoretic mobility, was shown previously to correlate with FynT activation (36). In comparison, FynT polypeptides recovered from cells expressing PEP (lane 2) were resolved as a single phosphorylated product of 59 kD. Little or no FynT* was found in these cells.

To identify more precisely the sites of phosphorylation on these molecules, the phosphorylated products were cleaved with cyanogen bromide and the resulting fragments were resolved in 18% SDS-PAGE gel (Fig. 7 B). Previous studies demonstrated that cleavage of FynT with cyanogen bromide produces three major phosphorylated peptides: C1 (21 kD), which contains NH₂-terminal sites of serine and threonine phosphorylation; C2 (9.6 kD), which possesses tyrosine 417; and C3 (4.2 kD), which bears tyrosine 528 (37–39). To help identify the C2 fragment, a peptide map of constitutively activated tyrosine 528 to phenylalanine 528 (F528) FynT was performed in parallel (lane 5). This mutant is known to be extensively phosphorylated at tyrosine 417 (38).

We observed that the 59-kD FynT molecules in cells lacking PEP (Fig. 7 B, lane 1) were phosphorylated within the tyrosine 528–containing C3 peptide and to a lesser, albeit appreciable extent, the tyrosine 417–bearing C2 peptide. Phosphorylation within the NH₂-terminal C1 fragment was also present. The 65-kD products (lane 3) were more prominently phosphorylated within the tyrosine 417–containing C2 peptide, relative to the C3 fragment. This finding is consistent with the notion that FynT* represented activated FynT molecules. In cells expressing PEP (lane 2), there was no reduction in the extent of phosphorylation of p59^fynT within the tyrosine 528–bearing C3 peptide. However, phosphorylation at tyrosine 417 was abolished. Furthermore, the existence of FynT* was abrogated (lane 4). Even though the global extent of phosphorylation in the C1 fragment was not altered by the presence of PEP, the electrophoretic mobility of this peptide was noticeably faster (lane 2). The exact basis for this difference is not known. Thus, in combination, these data indicated that PEP caused dephosphorylation of tyrosine 417, but not of tyrosine 528, of FynT.