Phosphorylation of extracellular domains of T-lymphocyte surface proteins. Constitutive serine and threonine phosphorylation of the T cell antigen receptor ectodomains.

The extracellular accumulation of ATP after activation of T-lymphocytes, as well as the presence of ecto-protein kinases in these cells, led us to propose that T cell surface receptors could be regulated through the reversible phosphorylation of their extracellular domains (ectodomains). Here, in a model system, we used T cell transfectants which express T cell antigen receptor chains lacking intracellular and transmembrane protein domains and 32Pi metabolic labeling of cells to definitively demonstrate phosphorylation of ectodomains of T cell surface proteins. We show that αβTCR ectodomains were phosphorylated intracellularly and constitutively on serine and threonine residues and were then expressed on the T cell surface in phosphorylated form. TCR ectodomains also could be phosphorylated at the cell surface when extracellular [γ-32P]ATP or [γ-32P]GTP were used as phosphate donors with the same cells. Consensus phosphorylation sites for serine and threonine protein kinases were found to be strongly evolutionary conserved in both α and β TCR chains constant regions. These results are consistent with the hypothesis, where T cell surface proteins which are phosphorylated intracellularly on their ectodomains, could subsequently be expressed at the cell surface and then be reversibly modified by ectoprotein phosphatase(s) and by ectokinase(s). Such modifications may change T cells cognate interactions by, e.g. affecting TCR-multimolecular complex formation and antigen binding affinity. It is suggested that αβTCR ectodomain phosphorylation could serve as a potential mechanism for regulation of αβTCR-mediated T-lymphocytes response.

T cell differentiation and effector functions involve multiple cell-cell contacts and interactions that are mediated by cell surface-located recognition structures, cell adhesion molecules, and growth factor receptors. Recognition of antigenic peptides/ MHC 1 class I and II complexes is the function of extracellular variable regions of the ␣␤ T cell receptor (␣␤TCR), while cytoplasmic domains of ␣␤TCR⅐CD3 complex molecules are considered critical for transmembrane signaling (1,2).
In addition, extracellular domains of cell surface proteins are involved in multimolecular complex formation as underscored by studies of TCR/CD3 subunit assembly (3) and of CD4-TCR interactions that contribute to TCR recognition of MHC class II-presented peptides (4). It was also demonstrated recently that CD45 ectodomains regulate CD4-␣␤TCR associations (5).
These considerations and our earlier studies of the role of extracellular ATP in T cell effector functions (6) led us to propose that properties of T cell surface receptors, including cell adhesion proteins and recognition molecules, could be regulated through phosphorylation of their extracellular domains (7,8) in a manner now accepted as a mechanism for regulation of enzyme-substrate interactions (9).
The precedent for such a mechanism was recently provided by studies of CD36 on the platelets surface. It was demonstrated that the specificity of that receptor for collagen or thrombospondin is regulated by ectophosphorylation of its ectodomain (10).
Analysis of published amino acid sequences of lymphocyte surface proteins (data not shown) revealed the presence of consensus protein kinase phosphorylation sites in the extracellular domains of the majority of functionally important surface proteins, including ␣␤TCR. ␣␤TCR was also found to be ectophosphorylated by extracellular [␥-32 P]ATP during preliminary screening of immunoprecipitates of T cell surface proteins with a panel of monoclonal antibodies (8). 2 This suggested that ␣␤TCR ectodomain phosphorylation could affect antigen recognition and/or interactions of ␣␤TCR molecules with ectodomains of other functionally important molecules (e.g. CD4/CD8, CD3, or CD45) and thereby regulate T cell cognate interactions and effector functions (8).
The unambiguous demonstration of T cell surface proteins ectodomain phosphorylation is an important requirement for the further testing of this hypothesis since the studies of extracellular protein kinases (ectokinases) in other cellular systems (11,12) were not complemented by definitive description of their cell surface substrates due to known caveats of ectophosphorylation assays using extracellular [␥-32 P]ATP (13).
In this study we provide the first direct evidence of the T cell surface protein (␣␤TCR) ectodomain phosphorylation. This was done by using T cell transfectants that express the recombinant intracellular tail and transmembrane domain-lacking ␣␤TCR molecules after the long-term metabolic 32 P i labeling.

MATERIALS AND METHODS
Recombinant Proteins, Cells, and mAbs-The extracellular domains of ␣␤TCR cloned from the anti-H-2L d cytolytic clone, 2C were expressed as phosphatidylinositol-linked molecules on transfected BW 5147 lymphoma cells (14) and released from the cell surface using phosphatidylinositol-phospholipase C (PLC, Sigma). The 2C-TCR transfectants were maintained in AMEM media (Biofluids, Rockville, MD), releasable 2C-TCR and soluble MHC class I molecules (sH-2L d ) were purified by immunoaffinity chromatography as described previously (15). 2B4 T helper hybridoma cells were maintained in RPMI 1640 with standard supplement and 10% fetal calf serum (Biofluids, Rockville, MD). Thymocyte clone 4F7 was established by limiting dilution from spontaneous thymoma formed in p53KO mice and maintained as 2B4 cells. 3 Normal thymocytes and lymph node cells were isolated from DBA/2 mice using standard procedures. mAb 1B2 (16), directed against the antigen combining site of the cloned 2C-TCR were purified using protein A-Sepharose chromatography. Purified H57-597 hamster mAb against the common epitope of mouse TCR␤ chain were purchased from Pharmingen (San Diego, CA).
Real Time Surface Plasmon Resonance-All binding experiments were performed at 25°C using Pharmacia BIAcore TM (Pharmacia, Uppsala, Sweden) as described (15). Briefly, the soluble 2C-TCR was diluted in 10 mM sodium acetate at pH 4.6 and was coupled to the biosensor surface with standard amine coupling. Binding abilities of chemically immobilized 2C-TCR were determined by using soluble H-2L d MHC class I molecules containing bound endogenous peptides (no added peptide) or complexed with the specific p2Ca (LSPFPFDL) or pMCMV (YPHFMPTNL), nonspecific control peptides. All peptides were synthesized at the Laboratory of Molecular and Structural Biology (National Institute of Allergy and Infectious Diseases, National Institutes of Health) and are referred to by single-letter code.
Treatment of Cells with Protein Phosphatases-Acid phosphatase (Sigma) was dialyzed against Hanks' balanced salt solution before experiment and pretested for its phosphatase activity. Cells were incubated with or without (control) 200 units/ml of purified preparations of acid phosphatases 10 min, 37°C at pH 7.1 in conditions shown to be both sufficient for the demonstration of enzymatic activity of acid phosphatase and for maintenance of the cell viability during the treatment. Phosphatase-pretreated cells were used in ectophosphorylation assay as described above.
Phospholipase C Treatment-Labeled cells were washed twice and resuspended in RPMI 1640 with standard supplement and 100 nM phosphatase inhibitor microcystin LR (Calbiochem, La Jolla, CA). The cell suspension (10 -20 ϫ 10 6 /ml) was incubated at 37°C for 1.5 h in T-80 flasks (15-20 ml) in the presence of 0.25-0.5 units/ml of phospholipase C (Sigma). In some experiments, 10 ϫ 10 7 cells/ml were treated with 1 unit/ml of phospholipase C for 2 h at 37°C. The supernatants of PLC-treated cells were harvested, precleared by centrifugation and by incubation with protein A-or G-Sepharose for 1 h at 4°C before immunoprecipitation. Control for efficiency and specificity of PLC treatment was monitored by fluorescence-activated cell sorter analysis as described below.
Immunoprecipitation and SDS-PAGE-Cell lysate was prepared with buffer containing 1% Nonidet P-40 (Pierce, Rockford, IL), 150 mM NaCl, 50 mM Tris (pH 7.5), and 1 mM phenylmethylsulfonyl fluoride, 10 mg/ml leupeptin, antipain, aprotinin, 0.4 mM orthovanadate, and 10 mM sodium fluoride (all from Sigma), and 2 mM Pefablock SC from Boehringer Mannheim. Purified mAbs were preabsorbed on protein A-or protein G-Sepharose (Pharmacia) (20 g of antibody per 50 l of 50% Sepharose gel) by incubation for 2 h or overnight at 4°C. Precipitation of immunocomplexes was done by 2 h incubation of mAb-precoated Sepharose with cell lysates or supernatants of PLC-treated cells. Immunoprecipitates were washed 4 -6 times with lysis buffer, than were boiled with SDS sample buffer and eluted immunocomplexes were analyzed by SDS-PAGE under reducing and nonreducing conditions using the standard Laemmli method and 4 -20% gradient minigel system electrophoresis (Integrated Separation System, Natick, MA). 14 C-Labeled standard markers (Amersham Life Science, Inc.) were used to identify molecular weights of phosphoproteins. After electrophoresis, gels were stained with Coomassie Brilliant Blue, destained, dried, and subjected to autoradiography using X-Omat AR or BIOMAX MS film (Eastman Kodak, Rochester, NY).
Phosphoamino Acid Analysis-Phosphoamino acid analysis was carried out according to the standard method (19). Briefly, SDS-PAGEseparated proteins were transferred onto a polyvinylidene difluoride membrane (Millipore, Bedford, MA) and after overnight autoradiography, phosphoprotein bands were excised and hydrolyzed in boiling 5.7 N HCl at 110°C for 60 min. The samples were dried in a Speed-Vac Concentrator (Savant, Hicksville, NY), dissolved in the buffer mixed with unlabeled phosphoserine, phosphothreonine, and phosphotyrosine (Sigma) and subjected to two-directional electrophoresis (pH 1.9 and 3.5) on TLC plates (Uniplate TM Avicell, Analtech Inc., Newark, DE). No Thr-phosphorylation sites and only one possible Ser-phosphorylation site were found in the amino acid sequence of Thy-1-derived region (14) of chimeric 2C-TCR molecules.

RESULTS
In preliminary experiments, under conditions that favor phosphorylation of ectodomains rather than intracellular sites (20), we observed that TCR, CD45, and HSA proteins expressed on the cell surface of living cells were phosphorylated during short incubation with extracellular [␥-32 P]ATP (data not shown). However, such assays could be misleading, despite extensive controls (13), and we developed new and more stringent criteria to unambiguously demonstrate phosphorylation of the extracellular domains of such surface proteins.
We focused our studies on the ␣␤TCR due to its key role in T cell activation as well as because of the availability of cellular and molecular reagents that allow critical controls. First we examined transfectant T cells expressing recombinant GPIanchored ␣␤TCR molecules from 2C cytolytic T cell clone ( Fig.  1). For such a protein, phosphorylation can only occur on the extracellular part of the molecules allowing localization to the ectodomain. The release of the TCR by PLC treatment provides additional assurance that we are examining only cell surface expressed molecules.
Isolation and Immunopurification of ␣␤TCR Ectodomains-Specific removal of GPI-linked TCR molecules from the cell surface was confirmed by flow cytometry and 125 I-surface labeling. The treatment of 2C-TCR cells with PLC results in the decrease of anti-TCR staining with mAb but did not affect other, non-GPI-linked surface antigens, such as CD44 (Fig.  1B). In parallel the appearance of 125 I-labeled 2C-TCR molecules was observed in supernatants of PLC-treated cells (Fig.  1C, inset). PLC treatment of 4F7 T cells was shown to effectively remove another GPI-linked surface protein (i.e. Thy-1, HSA) without affecting expression of non-GPI-linked proteins (e.g. TCR and CD45; data not shown).
We also confirmed the antigen-specific binding of PLCreleased soluble 2C-TCR molecules that were purified using clonotypic anti-2C-TCR mAb 1.B2 (Fig. 1C). Indeed, the purified 2C-TCR molecules were able to bind to H-2L d molecules loaded with specific (p2Ca), but not with control (pMCMV) peptide, as detected by plasmon resonance assay (Fig. 1C). The soluble 2C-TCR was also found to bind to clonotypic 1B.2 mAb in a parallel experiment (data not shown). Taken together, these results justified the use of a procedure involving PLC pretreatment of 2C-TCR transfectants followed by immunoprecipitation with mAb to TCR for the specific purification of 2C-␣␤TCR molecules in order to study their ectodomain phosphorylation.
Detection of ␣␤TCR Ectodomain Phosphorylation-After 4 h of metabolic labeling with 32 P i followed by cell lysis and immunoprecipitation with the clonotypic anti-2C-TCR mAb we observed phosphorylated TCR molecules on SDS-PAGE gel. 32 P-Labeled 2C-TCR is seen as disulfide linked heterodimer that comigrates with 125 I-surface-labeled TCR ( Fig. 2A).
Phosphorylated 2C-TCR molecules were released from the cell surface by PLC treatment after 16 h of metabolic cell labeling and were detected in the supernatants by immunoprecipitation with clonotypic 1B.2 mAb and SDS-PAGE under reducing and non-reducing conditions similar to 125 I-surfacelabeled TCR (Fig. 2, panels C and D, lanes 4). In control samples no ␣␤TCR bands were observed when cells were not exposed to PLC (Fig. 2, C and D, lanes 3 and 5) or if no 1B2.mAb was used (lanes 1 and 2). H57-597 mAb anti-mouse TCR common epitope was used with the same results (data not shown) as clonotypic mAb 1B.2. These results are consistent with the conclusion that the 32 P i -labeled proteins detected in anti-TCR mAb immunoprecipitates (Fig. 2) were, indeed, the GPI-linked ␣␤TCR chains.
The detailed time course of the cell labeling with 32 P i and PLC-treatment (Fig. 2B) demonstrated that 32 P-labeled ␣␤TCR molecules could be detected on the cell surface as early as after 1 h and established 16 h as the best and most convenient time of 32 P i labeling to observe the 32 P-phosphorylated and PLCreleasable ␣␤TCR molecules.
Phosphoamino acid analysis of SDS-PAGE separated TCR protein bands both under reducing (Fig. 3, panels A and B) and non-reducing (Fig. 3, panels C and D) conditions revealed the presence of phosphoserine and phosphothreonine, but not phosphotyrosine. The same results were obtained by phosphoamino acid analysis of TCR isolated by immunoprecipitation from either total cell extracts of 32 P i metabolically labeled T cells (data not shown) or from the PLC-releasable fraction (Fig. 3). Detection of phosphoserine and phosphothreonine also excluded the possibility that the radioactivity observed on autoradiographs (Fig. 2) in ␣␤2C-TCR chains was due only to the 32 P-labeled phosphoinositol, which was released with ␣␤TCR ectodomains after treatment with PLC. It is of interest that some yet-to-be identified T-cell phosphoproteins were co-immunoprecipitated with TCR (Fig. 2) and were ectophosphorylated by [␥-32 P]ATP and [␥-32 P]GTP (Fig. 5).

Phosphorylation of 32 P i -Labeled ␣␤TCR in Normal T Cells and the Presence of Evolutionary Conserved Phosphorylation
Sites-While the use of 2C-TCR transfectants provided a convenient model system, we also studied the phosphorylation of ␣␤TCR in other cell lines. The 32 P i labeling of ␣␤TCR in phosphorylation assays, similar to those used in demonstration of CD3 chain phosphorylation (21), has not yet been reported. Fig.  4A demonstrates phosphorylated ␣␤TCR from 2B4 T cells after 16 h of 32 P i metabolic labeling (Fig. 4A, lane 3) which comigrate with 125 I-labeled ␣␤TCR (Fig. 4B, lane 2) molecules. The strikingly evolutionary conserved (between rabbit, human, and mouse) protein kinase phosphorylation sites were revealed in both C␣TCR and C␤TCR ectodomains (Fig. 4C) by analyzing their published (22,23) amino acid sequences. Interestingly, the CKII phosphorylation site in C␤TCR is located in the solvent-exposed insertion absent in C domains of Ig, as indicated in the x-ray crystallographic structure of the TCR ␤ chain (24).
Phosphorylation of ␣␤TCR ectodomains in normal T cells and in GPI-TCR transfectants was also studied in an ectophosphorylation assay by using extracellular [␥-32 P]ATP or [␥-32 P]GTP as a phosphate donor. We observed ␣␤TCR phosphorylation by extracellular [␥-32 P]ATP in thymocytes, lymph node cells (Fig. 5, panel A), and T cell lines (Fig. 5, panels B-D).
To avoid misinterpretation (13) we introduced two new and more stringent controls: (i) the requirement to confirm the extracellular [␥-32 P]ATP-mediated phosphorylation of intact cells by using transfectants with transmembrane domain and cytoplasmic tail lacking TCR molecules; (ii) the requirement for [␥-32 P]ATP-mediated ectophosphorylation to be affected by pretreatment of cells with membrane-impermeable protein phosphatases. These requirements are satisfied in experiments where we observed the phosphorylation of 2C-TCR in transfectants after incubation with [␥-32 P]ATP (Fig. 5, panel D) similar to the phosphorylation after 32 P i metabolic labeling of the cells (Fig. 2). The pretreatment of T cells with protein phosphatases dramatically improved the [␥-32 P]ATP-mediated ectophosphorylation of ␣␤TCR as was expected if the majority of ␣␤TCR ectodomain-phosphorylation sites has been constitutively phosphorylated (Fig. 5, panel B).
Incubation of intact cells with both extracellular [␥-32 P]ATP or [␥-32 P]GTP resulted in phosphorylation of proteins that were immunoprecipitated with clonotypic anti-TCR mAb ( Fig. 5 and data not shown). The abilities of both [␥-32 P]ATP and [␥-32 P]GTP (Fig. 5, lanes 2 and 4) to serve as phosphate donors in ectophosphorylation reactions are hallmarks of the casein II kinases (25). In our earlier studies of ectophosphorylation assays with specific inhibitors and enhancers of CKII, 4 we found that a casein kinase II-like enzymatic activity was the major ectokinase activity in T cells.
The results of the experiments presented above strongly suggest that ectodomain phosphorylation may take place both intracellularly and extracellularly. Importantly, the intracellular phosphorylation of ␣␤TCR is constitutive, since no additional stimuli were required to observe it in metabolically 32 P ilabeled T cells. The demonstrated effects of extracellularly added protein phosphatases (Fig. 5, panel B) support the possibility of the ectodomain phosphorylation being a reversible process. DISCUSSION We describe here the constitutive phosphorylation of T cell surface protein ectodomains using recombinant ␣␤TCR T cell transfectants as a model system. The possibility is raised that such phosphorylation may reflect the functioning of the previously unexplored mechanism of regulation of TCR-mediated cognate T cell's interactions. Indeed, T lymphocytes seem to possess the complete system of reversible extracellular phosphorylation/dephosphorylation as evidenced by the extracellular ATP accumulation in TCR-triggered T cells (26), by the description of T cell-associated ATPases (6), ectokinases, and protein phosphatases (8) 5 and as demonstrated here by ectodomain phosphorylation of functionally important T cell surface protein (␣␤TCR). The function of the highly active ecto-ATPase activities on T cells (6) could be to provide an additional level of regulation of the extracellular domains phosphorylation and signaling through purinergic receptors by limiting the concentration of the extracellular ATP near the cell surface.
Extracellular domains of ␣␤TCR have been the subject of intensive investigations, since they contain the antigen recognition and binding sites (1) and regulate antigenic peptide recognition due to their interactions with CD4 ectodomains (4) and/or other surface proteins which form the multimolecular TCR⅐CD3 complex. The involvement of ␣␤TCR ectodomainmediated interactions is also implied by considerations of the 4 S. A. Redegeld, P. Smith, and M. Sitkovsky, unpublished observations. 5 F. Redegeld, unpublished observations. role of ␣␤TCR dimerization and aggregation (27)(28)(29), lattice formation (30) in transmembrane signaling in T cells. In addition, ␣␤TCR-CD3 ectodomain interactions were shown to be important for the assembly of TCR⅐CD3 molecular complexes (3).
The possibility of the role of extracellular ATP in extracellular domain phosphorylation of functionally important proteins in T cell's effector functions was first raised during earlier studies of the biochemical mechanisms of CTL-target cell interactions (7,8,26) and it was important to explore whether reversible ␣␤TCR ectodomain phosphorylation could be the fine-tuning mechanism of TCR-mediated immune responses.
The ectophosphorylation in different cellular systems has been studied over the last several decades (reviewed in Refs. 20 and 31), and elaborate criteria were developed to avoid many potential problems in interpretations of ectophosphorylation assays. Nevertheless, none of these criteria were sufficient, and intracellularly-cytoplasmic tail-phosphorylated proteins could be still misinterpreted as "ectophosphorylated" (13). 6 Thus, the definitive demonstration of phosphorylation of ectodomains had yet to be provided in any cellular system, and a new approach was needed to demonstrate the presence of phosphorylated amino acid residues in ␣␤TCR ectodomains.
The demonstration of ␣␤TCR ectodomain phosphorylation was accomplished here using T cells stably transfected with transmembrane domain-and cytoplasmic tail-lacking GPI anchor-linked molecules of 2C-␣␤TCR and labeled both in a [␥-32 P]ATP ectophosphorylation assay and in a long-term 32 P i metabolic labeling assay (Figs. [1][2][3][4][5]. This allowed us to independently confirm experiments with 32 P i -labeled 2C-TCR transfectants or normal T cells with results of ectophosphorylation assay obtained by incubating intact cells with [␥-32 P]ATP/GTP. The important differences between ␣␤TCR ectodomains and intracellular domains of the TCR⅐CD3 complex phosphorylation were revealed here by phosphoamino acid analysis (Fig. 3). In contrast to well documented intracellular domain tyrosine phosphorylation (2), only the phosphoserine and phosphothreonine, but no phosphotyrosine, were detected in our studies of ␣␤TCR ectodomains.
The interchangeable use of ATP and GTP (Fig. 5) as phosphate donors in ␣␤TCR ectophosphorylation has important implications, since it not only implicates ectokinase with CKIIlike properties in ␣␤TCR ectodomain phosphorylation but it also eliminates the possibility of the opening of ATP 4Ϫ -gated nonspecific membrane pores-P2z receptors (8). These results are consistent with experiments demonstrating the effects of CKII inhibitors and enhancers on the ectophosphorylation of surface proteins on T cells (data not shown).
Taken together, the results described above and studies of the effects of pretreatment of cells with protein phosphatases and protein kinases ( Fig. 5 and data not shown) are consistent with the model in which ectodomains of TCR are constitutively intracellularly serine/threonine phosphorylated and then expressed on the cell surface. The detection of both phosphoserine and phosphothreonine indicates the possibility that there are several phosphorylation sites in each of the ␣␤ TCR chains. Subsequently, ␣␤TCR ectodomains can be dephosphorylated extracellularly by PP1 and PP2A ectoprotein phosphatase(s) (data not shown) and further rephosphorylated by CKII-like or other ectoenzyme with extracellular ATP/GTP as a phosphate donor. The implication of ecto-CKII-PKase in ␣␤TCR ectophosphorylation is especially intriguing in view of strikingly evolutionary conserved (between rabbit, mouse and human) consen-sus phosphorylation sites for CKII in the C␤ domain of ␣␤TCR and of tandem PKC and CKII phosphorylation sites in the C␣ domain (Fig. 4C).
The functional role of ␣␤TCR ectodomain phosphorylation remains to be definitively established and the most promising approach appears to be in testing the effects of mutation of the ectophosphorylation site on cognate interactions of T cells transfected with mutated ␣␤TCR and studies of the effects of phosphorylation of soluble recombinant TCR on TCR/MHC class I-peptide interactions using plasmon resonance techniques. The extracellular phosphorylation system described here may have important immunopharmacologic implications, since the well defined phosphorylation sites in ectodomains of functionally important surface proteins and ectoenzymes (protein kinases and protein phosphatases) would provide attractive targets for immunomodulation due to their surface location.
In conclusion, we would like to speculate that ␣␤TCR chains have more than one phosphorylated residue in each chain and, depending on their location, they may influence associations of ␣␤TCR chains with CD3 molecules and/or associations of ␣␤TCR chains with CD4/CD8 co-receptors or other molecules comprising the TCR⅐CD3 complex. It should be also tested whether the ␣␤TCR ectodomain phosphorylation may affect recognition and binding properties of ␣␤TCR variable regions thereby providing an attractive mechanism of "On/Off" switch in regulation of formation and separation of cells in CTL/target cell or Th cell/antigen presenting cells conjugates.