Activation of the Cooh-Terminal Src Kinase (Csk) by Camp-Dependent Protein Kinase Inhibits Signaling through the T Cell Receptor

Engagement of the TCR/CD3 complex leads to activation of the Src family tyrosine kinases Lck and Fyn 1,2. These kinases mediate the initial tyrosine phosphorylation of immunoreceptor tyrosine-based activation motifs in the TCR/CD3 subunits (e.g., ζ chain) and elicit a complex series of proximal signaling events. This involves recruitment of the tyrosine kinase Zap-70 to the ζ chain and subsequent tyrosine phosphorylation of lipid raft–associated adaptor molecules such as the linker for activation of T cells (LAT) that, via phosphotyrosine binding, further recruit several downstream, Src homology 2 (SH2) domain–containing signaling molecules (for a review, see reference 3). The Src family of tyrosine kinases are negatively regulated by phosphorylation of a conserved COOH-terminal tyrosine residue (Y505 in Lck, Y528 in FynT) by the COOH-terminal Src kinase, Csk 4,5,6. Although Csk has substantial homology to Src kinases, it lacks the COOH-terminal regulatory tyrosine found in Src kinases 7. Little or no evidence has been presented to demonstrate any enzymatic regulation of Csk 8, such as by other signaling pathways. However, a recently identified LAT-homologous, transmembrane adaptor protein called Csk-binding protein (Cbp; reference 9) or phosphoprotein associated with glycosphingolipid-enriched membrane domains (PAG; reference 10) with the capacity to bind Csk, may allow spatial regulation of Csk activity toward Src kinases in lipid rafts 8, and binding of Csk to Y317 in Cbp/PAG may increase the activity of Csk 11.

cAMP, the levels of which are increased, for example, by prostaglandin E and β-adrenergic stimuli, negatively regulates mitogenic signaling pathways at multiple levels 12,13,14. In normal T cells, cAMP-dependent protein kinase (PKA) type I colocalizes with the TCR–CD3 complex and inhibits TCR-induced cell proliferation via a previously unknown proximal target 15,16,17. In T cells from HIV-infected patients and some patients with common variable immunodeficiency, increased levels of cAMP and hyperactivation of PKA type I contribute to the T cell dysfunction, and PKA type I selective antagonists can improve immune function of patient T cells in vitro up to 300% 18,19,20. Now, we report a novel inhibitory pathway in T cells whereby PKA type I, through activation of Csk leading to inhibition of Lck-mediated ζ chain phosphorylation, shuts down the proximal T cell activation. Furthermore, we demonstrate that the whole PKA type I-Csk-Lck inhibitory pathway spatially is assembled in lipid rafts where the initial T cell activation takes place.

The human leukemic T cell line Jurkat (clone E6.1), Jurkat TAg, a derivate of the Jurkat cell line stably transfected with the SV40 large T antigen 21, and the Lck-deficient JCaM1 cell line 22 were kept in logarithmic growth in RPMI 1640 supplemented with 10% FCS, sodium pyruvate, nonessential amino acids, and monothioglycerol. Human peripheral blood T cells were purified from normal donors by negative selection 18. T cells were activated by the addition of 5–10 μg/ml of anti-CD3ε mAb OKT-3 or by pervanadate treatment. For transfections, cells (2 × 10⁷) in 0.4 ml Opti-MEM were mixed with 2–80 μg of each DNA construct in electroporation cuvettes with a 0.4-cm electrode gap (BioRad Laboratories) and subjected to an electric field of 250 V/cm with 960 μF capacitance. The cells were expanded in complete medium and harvested after 20 h. To obtain only the transfected cell population for functional assays, cells were cotransfected with a plasmid encoding the rat NK cell marker NKR-P1A (a gift from Dr. J.C. Ryan, VA Medical Center, University of California at San Francisco, San Francisco, CA) and purified by positive selection using anti–rat NKR-P1 mAb (clone 3.2.3) and anti–mouse IgG paramagnetic beads which allows release of bead-bound cells by digestion of a DNA linker that attaches the Ab to the bead (Cellection; Dynal).

Immunoprecipitation of Zap-70, Lck, and Csk was as described previously 23. For immunoprecipitation of hemagglutinin epitope (HA)-tagged Csk, transfected cells were disrupted in lysis buffer (50 mM Hepes, pH 7.4, 100 mM NaCl, 5 mM EDTA, 1% Triton X-100, 10 mM sodium pyrophosphate, 1 mM Na₃VO₄, 50 mM NaF, 1 mM PMSF, and 10 μg/ml each of leupeptin, antipain, pepstatin A, and chymostatin). When cells were stimulated with OKT-3, cell lysates were precleared by incubation with protein A/G–Sepharose beads (Sigma-Aldrich) for 1 h at 4°C, and subjected to immunoprecipitation with anti-HA mAb (Babco) or anti-Csk Ab (Santa Cruz Biotechnology, Inc.). After overnight incubation at 4°C, protein A/G–Sepharose was added, and the incubation continued for 1 h. Immune complexes were washed three times in lysis buffer and three times in Csk kinase assay buffer (50 mM Hepes, 5 mM MgCl₂, pH 7.4), followed by Csk kinase assays and Western blot analysis.

Detection of phosphotyrosine by anti-PTyr mAb (4G10; Upstate Biotechnology), and immunoblotting with anti–Zap-70, anti-Lck, anti-HA, anti-Csk, anti-PKA RIα, anti-PKA RIIα, anti-PKA C, and anti-LAT Abs were as before 18,23,24 except that recently developed mAbs directed against human RIα and human RIIα (cat. no. P53620; K. Taskén in collaboration with Transduction Laboratories) and anti-Csk Ab from Santa Cruz Biotechnology, Inc. (SC-286) were used.

The gene-encoding human Csk 25 was subcloned into the expression vector pEF-BOS/HA at NheI-XbaI sites. Csk-S364A, Csk-S364C, and Csk-S339A/S340A/T341A mutants were made by PCR or using a site-directed mutagenesis kit (Quickchange; Stratagene) and verified by sequencing.

Cloning, expression, and purification of human Csk has been reported previously 25 and yielded an enzyme with a specific activity in the range of that of the native purified enzyme. Recombinant purified catalytic subunit of PKA (Cα; reference 26) was a gift from Dr. F. Herberg, Ruhr University, Bochum, Germany.

Csk was incubated with PKA C subunit at 30°C for the indicated time periods in 50 mM Hepes, pH 7.4, 5 mM MgCl₂, 3–5 μM [γ-³²P]ATP (50–320 Ci/mmol). All reactions were stopped by boiling samples in SDS sample buffer, followed by SDS-PAGE. Gels were stained with Coomassie brilliant blue, dried, and subjected to autoradiography.

The tyrosine kinase activity of human Csk was measured as incorporation of [^32P]phosphate into the synthetic polyamino acid poly(Glu,Tyr) 4:1 (Sigma-Aldrich), abbreviated pEY. A standard protocol was followed 25 with reaction volumes of 50 μl containing Hepes buffer, pH 7.4, 5 mM MgCl₂, 200 μM [γ-³²P]ATP (0.15 Ci/mmol), 200 μg/ml pEY, and different amounts of purified Csk. Native or heat-inactivated (65°C for 10 min) C subunit and/or protein kinase inhibitor peptide (protein kinase inhibitor [PKI] 6-22 amide; Sigma-Aldrich) was added where indicated. The incubation temperature was 30°C, and the incubation times were 12–15 min, if not otherwise stated.

Csk was phosphorylated by PKA for 30 min as indicated above and subjected to SDS-PAGE. The band corresponding to phosphorylated Csk was cut from the dried gel and subjected to partial acid hydrolysis in 6 M HCl at 110°C for 2 h. The acid was evaporated under vacuum and the hydrolyzed sample was dissolved in 30 μl H₂O. 10 μl of sample (∼1,000 cpm of ^32P) was separated in two dimensions together with 10 μg each PSer, PThr, and PTyr. Phosphoamino acid standards were stained with ninhydrin, and ^32P-labeled amino acids were detected by autoradiography.

Cell-free supernatants were harvested from Jurkat T cells after 20 h of culture and stored at −80°C. IL-2 levels were determined by ELISA (R&D Systems).

Isolation of lipid rafts or glycolipid-enriched membrane microdomains was performed as described in detail elsewhere 27. In brief, cells were homogenized in 1 ml ice-cold lysis buffer (described above) by 10 pestle strokes in a Dounce homogenizer, loaded at the bottom of a 40–5% sucrose gradient and centrifuged at 200,000 g for 20 h. 0.4-ml fractions were collected from the top.

cAMP treatment of Jurkat T cells (Fig. 1 A) and normal peripheral blood T cells (Fig. 1 B) inhibited and delayed the tyrosine phosphorylation of TCR-ζ chain and Zap-70 after T cell activation by anti-CD3 (OKT3; compare lanes 2 and 8 in Fig. 1 A and lanes 2 and 6 in Fig. 1 B). ζ chain and Zap-70 represent good in vivo substrates for Lck, and their phosphorylation status can be readily assessed in detergent-solubilized extracts as the ζ chain is only loosely associated with lipid rafts in activated T cells 28. Examination of Lck immune precipitates from cAMP-treated Jurkat T cells also showed a 50% decrease in kinase activity in vitro (Fig. 1 C). However, no direct downregulation of Lck or Fyn activity by PKA could be observed in immune precipitates or on purified Lck (data not shown). In contrast, Csk activity was increased two- to threefold after cAMP treatment. Furthermore, cAMP or PGE2 in combination with isobutyl-methylxanthine (IBMX) increased Csk activity similarly in peripheral T cells (Fig. 1 D). Transfection of JCaM1 cells that have a truncated and inactive Lck (Fig. 2, lanes 1–4) with wild-type Lck (lanes 5–8) or Lck-Y505F (lanes 9–12) reconstituted TCR-mediated signaling as evident from anti-CD3–induced ζ chain phosphorylation. Whereas cells with wild-type Lck showed a distinct reduction in anti-CD3–induced phosphorylation of ζ chain when pretreated with 8-CPT-cAMP (top panel, compare lane 8 with lane 6), ζ chain phosphorylation was not inhibited by cAMP in cells with Lck-Y505F (compare lane 12 with lane 10). We conclude that the regulatory site Y505 of Lck is required for cAMP-mediated inhibition of ζ chain phosphorylation. This implicates Csk as a target for regulation by PKA, and we next explored that possibility.

Fully active recombinant Csk 25 was readily phosphorylated by Cα of PKA (Fig. 3 A; lane 1, arrow), whereas no phosphorylation of Csk was detected when incubated with heat-inactivated (65°C for 10 min) Cα (lane 2). Incubation of recombinant Csk with the recombinant catalytic subunit of PKA (Cα) more than doubled the Csk-catalyzed phosphorylation of pEY compared with Csk incubated alone (Fig. 3 B, compare bar 2 with bar 1). This effect was not seen with heat-inactivated Cα (bar 3). Furthermore, the increase in Csk activity in the presence of native Cα was strongly reduced by the addition of PKI, a specific inhibitor of PKA (bar 4). PKA itself did not phosphorylate pEY (data not shown). In the presence of heat-inactivated Cα, Csk activity was constant for the first 10 min and then declined, whereas the activity curve was much steeper in the presence of native Cα and the activity was approximately twofold higher at each time point (Fig. 3 C). Increasing concentrations of Cα subunit led to a saturable increase in activation of Csk, reaching a maximum around a twofold molar excess of C subunit over Csk (Fig. 3 D). Incubation of pEY with increasing concentrations of Csk demonstrated a concentration-dependent increase in phosphate transfer, which was approximately twofold higher at all concentrations in the presence of a fixed amount of native C (Fig. 3 E).

To look at a normal substrate for Csk, heat-inactivated Lck was used as substrate and the activity of Csk in the presence and absence of PKA was examined. When Csk was limiting in the reaction, Csk-mediated tyrosine phosphorylation of Lck was 4.8-fold stronger in the presence than in the absence of PKA (Fig. 4).

Phosphoamino acid analysis of Csk phosphorylated by PKA demonstrated strong labeling on phosphoserine (Fig. 5 A). Tryptic peptide mapping of Csk phosphorylated by PKA revealed two major radioactive spots both of which contained PSer (Fig. 5 B, peptides 1 and 2). The human Csk amino acid sequence contains one putative phosphorylation site that fits the motif preferred by PKA, at amino acids 361–364 in the sequence KKFS. A Csk-S364A mutant was only weakly phosphorylated by PKA (Fig. 5 B, and data not shown), and both major tryptic peptides (1 and 2) were missing compared with wild-type Csk phosphorylated by PKA (Fig. 5 B). The observation that two phosphorylated peptides disappeared by mutation of a single residue is probably because of partial proteolysis by trypsin. To assess the phosphorylation of Csk in intact cells, Jurkat T cells were metabolically labeled with ³²Pi, and anti-Csk immunoprecipitates were analyzed by tryptic peptide mapping (Fig. 5 C). Whereas Csk from untreated cells contained a few weakly labeled phosphopeptides, treatment with cAMP or PGE1 induced the appearance of one strong (peptide 1) and three weaker spots (peptides 2–4). Peptides 1 and 2 comigrated with those in Fig. 5 B, as shown by eluting peptides from the maps with PGE1-induced in vivo–labeled and recombinant Csk and rerunning mixtures with equal amounts of radioactivity on new maps (Fig. 5 D). To determine the site of phosphorylation by PKA in intact cells, Jurkat cells transfected with HA-tagged wild-type and mutant Csk were metabolically labeled and then stimulated with cAMP (Fig. 5 C, right two panels). Tryptic peptide mapping revealed that whereas HA-Csk phosphopeptides (1, 3, and 4) comigrated with those of endogenous Csk, an S364C mutation abrogated labeling of peptide 1. The Csk-S364C mutant was catalytically active both when expressed in Escherichia coli and Jurkat TAg cells (Fig. 6A and Fig. C), whereas Csk-S364A was not. Perhaps Cys, but not Ala, in position 364 permitted a normal folding of Csk. Another site at the activation loop of Csk in the sequence KEASST (amino acids 336–341) could also potentially be phosphorylated by PKA, although not fully consistent with the motif preferred by PKA. This last region is often the site of kinase activation by autophosphorylation 29 or transphosphorylation by another kinase, for example, mitogen-activated protein kinase (MAPK) activation by MAPK kinase 30. The extent of PKA-mediated phosphorylation of a Csk-S339A/S340A/T341A mutant (Csk-AAA) was comparable to that of wild-type Csk, and its tryptic peptide map was identical to that of wild-type (data not shown). A PKA-mediated increase in the kinase activity of this latter mutant and wild-type, but not Csk-S364C, was observed in vitro (Fig. 6 A). Coexpression of wild-type Csk with PKA Cβ showed a 1.8-fold increase in Csk activity compared with Jurkat TAg T cells transfected with the Csk construct together with a vector with Cβ in the reverse orientation (Fig. 6 B). However, in contrast to the 1.8-fold increase in activity of wild-type Csk by treatment of Jurkat T cells with cAMP, the activity of the mutant Csk-S364C enzyme was not affected by cAMP (Fig. 6 C).

To assess the downstream effects of PKA-mediated activation of Csk on T cell activation, we examined TCR-induced IL-2 production in Jurkat T cells (clone E6.1). To avoid dilution by untransfected cells, we developed a protocol for the selection of transfected cells. Cotransfection with DNA encoding the rat NK cell receptor NKR-P1A and magnetic bead selection for receptor allowed purification of cells expressing green fluorescent protein (Fig. 7 A) or Csk (Fig. 7 B). TCR-induced IL-2 production was very sensitive to the levels of expressed Csk, and a 3.5-fold overexpression reduced IL-2 secretion almost down to basal levels (Fig. 7 C). Thus, although the relative effect of cAMP was constant, the magnitude of the inhibition by cAMP was strongly reduced at higher levels of Csk expression (Fig. 7 C, ○). The effect of mutagenesis of S364 in Csk on the cAMP-inhibitable IL-2 production was therefore analyzed at a 1.9:1 ratio of transfected over endogenous Csk (arrow in Fig. 7 C) where changes in the inhibition by cAMP could be measured readily. Expression of Csk-S364C which has no PKA-phosphorylation site, reduced the cAMP inhibition of IL-2 production compared with control or cells expressing wild-type Csk (Fig. 7 D; 30 vs. 50–60% inhibition). The presence of endogenous Csk (1:1.9 versus mutant) explains why the inhibitory effect of cAMP was not totally abrogated. Higher levels of Csk-S364C expression by itself totally inhibited TCR-induced IL-2 production, and the effect of cAMP could not be analyzed. In contrast, Lck overexpression (twofold) by itself did not inhibit IL-2 production, which was fully sensitive to cAMP inhibition. However, Lck-Y505F strongly reduced the inhibitory effects of cAMP on TCR-induced IL-2 production (Fig. 7 E).

We have reported previously the localization of PKA type I with the capped and activated TCR–CD3 complex 17. More recently, the understanding has been developed that proximal signaling events downstream of the TCR occur in specialized cholesterol- and glycolipid-enriched membrane microdomains or lipid rafts where signaling molecules such as Lck and LAT are targeted 27,31. The novel lipid raft–associated Cpb/PAG is shown to interact with Csk in rat brain and in T cells via phosphotyrosine 317 in human PAG (Y314 in rat Cbp; references 9, 10). To analyze the subcellular distribution of components of the novel PKA-Csk-Lck inhibitory pathway mapped here, we purified lipid rafts by sucrose gradient centrifugation and fractionation of Triton X-100 lysates of peripheral blood T cells. Pervanadate treatment of T cells induced a strong tyrosine phosphorylation of the constitutively lipid raft–associated LAT, and increased the phosphotyrosine content of Cbp/PAG and Lck (Fig. 8 A) as well as other proteins not associated with lipid rafts (Fig. 8 A, lanes 9–12). However, both Cbp/PAG and Lck were phosphorylated also in resting peripheral T cells (Fig. 8 A, top). Furthermore, analysis of the same fractions showed that Csk, PKA RIα, and PKA C subunit are present in lipid rafts of both activated (Fig. 8 B) and resting (data not shown) T cells. In contrast, PKA RIIα is not detected in rafts, consistent with our earlier observations showing that PKA type I (RIα₂C₂), and not PKA type II (RIIα₂C₂), mediates the inhibitory effect of cAMP on T cell immune function 16,17,18. Targeting of PKA type I may be mediated by an A-kinase anchoring protein (AKAP; for a review, see reference 32) directed to the RIα subunit and/or by docking of the C subunit e.g., via a caveolin-like protein 33. The constitutive association of Csk with rafts is consistent with the level of tyrosine phosphorylation of Cbp/PAG in resting T cells.

To functionally analyze the effect of PKA on Csk in rafts, we looked at Lck-defective JCaM1 T cells transfected with kinase-dead Lck (Lck-K273M) that cannot be autophosphorylated at Y394 and therefore can only be tyrosine phosphorylated at Y505 (by Csk). Both transfected Lck and endogenous Csk were present in lipid rafts of these cells. Furthermore, when transfected cells were incubated in the presence of forskolin (to stimulate cAMP production), the tyrosine phosphorylation of Lck-K273M isolated from rafts increased 2.3-fold, indicating that Csk activity in rafts was stimulated upon triggering of the cAMP-PKA pathway (Fig. 9). Similar observations were made in whole cell lysates (data not shown).

Csk is present in all human cells as a key regulator of Src kinases 7. The fact that the presence of Y505 in Lck is essential for the inhibitory effect of cAMP on ζ chain phosphorylation and IL-2 production indicates that the PKA-mediated phosphorylation of Csk may be a major mechanism by which cAMP inhibits TCR-mediated T cell activation (Fig. 10). A two- to fourfold increase in Csk activity by phosphorylation of S364 appears to have similarly distinct effects on T cell function as a two- to threefold Csk overexpression, which abolishes activation through the TCR 6 and downstream IL-2 production (Fig. 7 C). Furthermore, the stoichiometry of Csk phosphorylation by PKA in vitro is 0.3–0.5 mol/mol of Csk under optimal conditions, indicating a single site not fully phosphorylated (quite common with bacterially produced protein). In vivo, we anticipate that a specific pool of Csk may be preferentially phosphorylated by colocalized PKA and reaches a higher stoichiometry and extent of activation. In addition, Jurkat and other leukemic T cell lines have higher levels of tyrosine phosphorylation 34 and Src kinase activities 35 than peripheral T cells. Thus, normal T cells appear to have a more controlled and managed Lck activity that may implicate Csk regulation and a PKA-Csk inhibitory pathway to a larger extent than apparent from, for example transfection studies on Jurkat T cells. Indeed, low level labeling of peptide 1 (representing Csk-S364) was seen in the tryptic peptide mapping of Csk from metabolically labeled unstimulated cells (Fig. 5 C) which increased strongly by treatment with PGE1 alone (data not shown). This indicates that this site is phosphorylated under physiological conditions. The mechanism for Csk activation by S364 phosphorylation is currently under investigation in our laboratory, and data in progress indicate that interaction with the intrachain SH3 domain is implicated in the PKA-mediated activation of Csk.

We have recently reported that the T cell dysfunction in HIV can be reversed by inhibition of the increased activity of PKA type I 18, indicating that immunomodulation through cAMP/PKA contributes to the pathogenesis of this immunodeficiency. Inhibition of Lck through activation of Csk provides a molecular mechanism for this effect. Furthermore, PKA-mediated regulation of the activity of various Src kinase family members by phosphorylation of Csk may also provide a molecular mechanism for cAMP-mediated regulation of both B and NK cell activation 36,37. Finally, Csk and Src kinases are expressed in other tissues, including neuronal tissues 38, and the impact of cAMP regulation of Csk in these tissues will be interesting to pursue. The PKA phosphorylation site in Csk is conserved between vertebrates, suggesting that this site may have been subject to selection pressure, but is only partially conserved (RFS or KFT) in Csk homologous kinase (Chk/Lsk/Hyl/Matk) and Csk-type protein kinase (Ctk/Bhk/Ntk).

In conclusion, we report the mapping of a PKA phosphorylation site on Csk and regulation of Csk activity by cAMP/PKA. Localization of both Csk and PKA type I to lipid rafts supports the notion that this novel inhibitory pathway is assembled in membrane microdomains where it can intersect TCR-induced signaling at a proximal level. The presence of adenylyl cyclase that generates cAMP in lipid rafts of S49 lymphoma cells further supports assembly of the cAMP-PKA type I-Csk inhibitory pathway in lipid rafts (Fig. 10; reference 39). The constitutive localization of components of this pathway in lipid rafts may indicate that a tonic level of inhibition of T cell activation is imposed on resting T cells. PKA-mediated activation of Csk provides a molecular mechanism for cAMP-dependent inhibition of lymphocyte activation, and Csk-S364 and/or Lck-Y505 may be future targets for immunomodulating therapies. Furthermore, this mechanism may regulate signaling through Src kinases in general.

The authors are grateful for the technical assistance of Marianne Nordahl, Linda Trobe Dorg, and Scott Williams. We are indebted to Dr. Friedrich Herberg, University of Bochum, Germany for the kind gift of recombinant Cα subunit.

T. Vang, K.M. Torgersen, V. Sundvold, F.O. Levy, B.S. Skålhegg, V. Hansson, and K. Taskén were supported by the Norwegian Cancer Society, The Norwegian Research Council, Novo Nordic Foundation, Anders Jahre's Foundation for the Promotion of Science, and Odd Fellow Medical Fund; T. Mustelin was supported by National Institutes of Health grants AI35603, AI40552, AI41481, and AI48032. T. Vang, V. Sundvold, and K. Taskén are fellows of the Norwegian Cancer Society.