INTRODUCTION

Over 70 molecules are known to be specifically regulated during the process of T cell activation and differentiation (1). Many of these are up-regulated or expressed de novo on the surface of activated T cells, and detailed study of the structure and function of such activation antigens has been instructive toward an understanding of the immune response. In many cases elucidation of the regulation of expression and, in particular, the structure of activation antigens has led to the identification of a ligand or counter receptor, frequently expressed as an activation antigen on a non-T cell (reviewed in Refs. 2-4). Engagements between counter receptors are now established as crucial events during thymic differentiation and in the activation and proliferative response of T cells to antigenic stimulation (2). In addition, transfection studies with such counter receptors has greatly facilitated analyses of the biochemical signaling pathways that drive T cell proliferation.

More elusive has been the elucidation of the processes that drive the differentiation of precursor cytotoxic T lymphocytes (pCTL)¹ into mature CTL effector cells. In particular, it is not resolved whether the signals that drive proliferation also promote maturation and differentiation or whether discrete signals are required (reviewed in Ref. 5). Certainly there is evidence to suggest that primed pCTL may be able to undergo differentiation in the absence of further proliferation (reviewed in Ref. 6). One activation antigen, termed T lineage-specific activation antigen 1 (TLiSA1), has provided data pertinent to this question. A monoclonal antibody (mAb) termed Leo A1 and F(ab¹)₂ fragments of this mAb and of a polyclonal antibody to TLiSA1 inhibited the differentiation of human pCTL into effector cells when present during mixed lymphocyte culture but had no effect on cell proliferation or cytolytic function (7, 8). The mAb also inhibited the interferon-induced differentiation of human pCTL clones into cytotoxic effector cells, an effect that did not involve blocking of interferon binding (9).

The function of this antigen is not known, but a possible role in transmembrane signaling was suggested with the identification of apparently the same glycoprotein antigen on the surface of platelets (10). LeoA1 mAb directly induced persistent platelet aggregation independent of granule secretion in an Fc receptor-dependent process thought to involve the activation of protein kinase C, and during activation the platelet antigen, termed platelet and T cell antigen 1 (PTA1), itself became phosphorylated (10-12).

In a previous study we demonstrated that the expression of TLiSA1 on the surface of mixed lymphocyte culture or PHA-generated T cell blasts could be regulated with cytokines (8). Antigen expression was up-regulated with added interleukin-2 or tumor necrosis factor , and down-regulated with transforming growth factor (TGF) 1 (8). Notably, Ranges et al. (13) have reported that TGF1, added from the beginning of mixed lymphocyte culture, inhibited the development of murine CTL, a finding confirmed with human T cells in our study (8). This inhibitory effect of TGF1 was partially reversed by the addition of tumor necrosis factor  (8, 13), and it was also found that this cytokine prevented the suppressive effect of TGF1 on TLiSA1 expression (8). During the course of further studies to investigate the mechanism of such regulation, we made the unexpected finding that phorbol ester (TPA) up-regulated TLiSA1 surface expression, but this effect of TPA on Jurkat cells was abolished by co-induction with the calcium ionophore, A23187. Because the dual signal of TPA plus ionophore is an absolute requirement for the transcriptional activation of interleukin-2 and is necessary for interleukin-2-dependent T cell proliferation (1), whereas phorbol ester used alone can induce growth arrest and differentiation of HL60 cells and also the differentiation of sensitized splenic T cells into CTL effector cells (14, 15), such unusual regulation of expression may be of significance for the functional role of TLiSA1.

Clearly, further understanding of the function of TLiSA1 requires knowledge of its primary structure, its regulation of expression, and its relationship to PTA1 of platelets. In this report we show that the structure of TLiSA1 predicted from cDNA cloning is that of an immunoglobulin superfamily (IgSF) member, related to other IgSF members with adhesive function but with the unusual structure of two V domains and no C domains. The antigenic identity between TLiSA1 and PTA1 is confirmed with a mAb raised against the platelet protein and substantiated by extensive peptide sequencing of the PTA1 molecule and nucleotide sequencing of the platelet mRNA. Together, these data strongly suggest that TLiSA1/PTA1 is a signaling molecule, likely triggered by engagement of an extracellular ligand or cellular counter receptor. Intriguingly, the unusual regulation of expression of TLiSA1 antigen on the surface of stimulated Jurkat cells is shown to occur at the transcriptional level; TLiSA1 mRNA is induced with the phorbol ester, TPA, but this induction is abrogated or greatly reduced with the combined signal of TPA plus ionophore. However, the effect of ionophore does not appear to be mediated through calcineurin because it is not reversed in the presence of FK506. These results may facilitate further understanding of the processes involved in CTL differentiation.

EXPERIMENTAL PROCEDURES

Antibodies and Reagents

The preparation and specificity of the LeoA1 mAb against TLiSA1 have been described previously (7). The second generation anti-PTA1 mAb, NEWE1, was prepared by standard fusion techniques after immunizing BALB/c mice with immunopurified platelet PTA1 antigen. Antigen purification using LeoA1 mAb coupled to Sepharose CL-4B was by a batch procedure from platelet lysate as described (10). Hybridoma clones were screened by enzyme-linked immunosorbent assay using purified PTA1 antigen. After subcloning the mAb was purified from ascites fluid by DEAE chromatography and NEWE1 was typed as IgG1. Phorbol ester, TPA, A23187, biotinamidocaproae N-hydroxysuccinimide ester, lactoperoxidase, apyrase, prostaglandin E₁, 2-mercaptoethanol, neuraminidase, and the detergent Nonidet P-40 were purchased from Sigma, and PHA was from Wellcome (Dartford, UK). Nitrocellulose for immunoblotting was from Sartorious (Goettingen, Germany). Enhanced chemiluminescence detection reagents were from NEN Life Science Products. Rabbit anti-mouse IgG was from Dako (Carpinteria, CA). Na¹²⁵I was purchased from Australian Radioisotopes (Sydney, Australia). Streptavidin-biotin-horseradish peroxidase and hyperfilm-MP were from Amersham Corp. All other chemicals were from Ajax Chemicals (Auburn, Australia) or BDH (Kilsyth, Australia). FK506 was kindly donated by the Fujisawa Pharmaceutical Company (Osaka, Japan). The c-myc and c-jun cDNA were provided by Dr. Suzannae Cory (Walter and Eliza Hall Institute, Melbourne, Australia).

Preparation of Stimulated T Cells

PHA-T lymphoblasts were generated by culturing freshly isolated peripheral blood mononuclear cells for 3-10 days in the presence of 2 µg/ml PHA in tonsil-conditioned medium. Tonsil-conditioned medium was prepared by culturing single cell suspensions of tonsil and teasing freshly isolated tissue through a wire mesh in tissue culture medium containing 5 µg/ml PHA for 18 h. Tonsil-conditioned medium was recovered by centrifugation at 350 × g for 5 min, filter sterilized, and stored at 20 °C.

Stimulation with Pharmacologic Agents

Jurkat cells were maintained in RPMI 1640 tissue culture medium (Cytosystems, Sydney, Australia) containing 10% fetal calf serum (Flow Laboratories, Washington, DC), 1% glutamine, 5 × 10⁵ M 2-mercaptoethanol, penicillin (100 µ/ml), and streptomycin (100 µg/ml) at 37 °C in 10% CO₂ in air. Prior to stimulation, Jurkat cells in the logarithmic growth phase or PHA blasts prepared as above were harvested, washed, and resuspended to 2-5 × 10⁶ cells/ml. Unless stated otherwise, the phorbol ester, TPA, was added at 50 ng/ml, PHA at 2 µg/ml, ionophore A23187 at 0.3 µM, and FK506 at 10 ng/ml, and where combinations were used these were added together at the same concentrations as above. Except in time-course experiments, the cells were treated for 12 h before harvest for binding assays or Northern blot analysis.

Competitive Binding Assay

The binding of ¹²⁵I-LeoA1 antibody to resting and pharmacologically activated PHA-induced blasts and Jurkat cells was assessed as described (8). Scatchard analysis of the data was performed using a LIGAND and EBDA program modified for microcomputers.

Transfection Studies and Flow Cytometry

TLiSA1 cDNA was inserted into the XbaI site of the EF-BOS expression vector (16) and used to transfect by electroporation COS-7 epithelial and unstimulated Jurkat T cells. At 72 h post-transfection cells were harvested and stained with LeoA1 mAb and analyzed for TLiSA1 expression on a Becton Dickinson FACScan flow cytometer. Cells transfected with EF-BOS vector alone were used as a control.

Cell Labeling and Immunoprecipitation

Fresh human platelets from consenting adult donors were isolated from 3.13% sodium citrate anticoagulated venous forearm blood (1:9 v/v) collected through a 19 G needle after discarding the initial 10 ml of the transfusion. The whole blood samples were spun at 120 × g for 10 min to prepare platelet-rich plasma. Prostaglandin E₁ (1 µM) was added to the platelet-rich plasma, and the platelets were pelleted by centrifugation at 200 × g for 15 min. Platelet pellets were washed in buffer A (138 mM NaCl, 2.9 mM KCl, 0.5 mM MgCl₂, 12 mM NaHCO₃, 0.3 mM NaH₂PO₄, 0.5 mM glucose, 10 mM HEPES, pH 7.4) plus 2 µM prostaglandin E₁ and 0.2 units/ml apyrase before resuspension into phosphate-buffered saline for subsequent surface iodination. Jurkat cells were prepared as described above and stimulated with TPA for 12 h prior to harvest. Surface proteins on platelets and Jurkat cells were labeled using lactoperoxidase catalyzed iodination as described previously (10). Following labeling, cells and platelets were lysed in 1% Nonidet P-40 lysis buffer (1% Nonidet P-40, 150 mM NaCl, 10 mM Tris, pH 7.4, containing 2 mM phenylmethylsulfonyl fluoride, 20 mM iodoacetamide, and 10 µg/ml soybean trypsin inhibitor) for 60 min on ice.

COS-7 and unstimulated Jurkat cells transfected with either TLiSA1 cDNA in pEF-BOS or with pEF-BOS vector alone were surface labeled by biotinylation. Briefly, 2.10⁷ transfected cells were harvested by trypsinization and resuspended into 3.0 ml of borate buffer (10 mM Na₂B₄O₇, pH 8.8, 150 mM NaCl). Biotinamidocaproae N-hydroxysuccinimide ester was added to 50 µg/ml, and the cells stood at room temperature for 15 min. Ammonium chloride was added to 10 mM to stop the reaction, and the cells were washed in buffer A before lysis in Nonidet P-40 lysis buffer for 60 min on ice.

After lysis whole cell lysates were spun at 15,000 × g for 15 min at 4 °C to remove insoluble material. Soluble lysates were precleared with rabbit anti-mouse IgG coupled to Sepharose 4B beads for 2 h at 4 °C. Antigen was precipitated by addition of a primary precipitating mAb to the precleared lysate. Immunoprecipitation was completed by the addition of rabbit anti-mouse IgG Sepharose-4B beads 30 min prior to completion of the procedure. All antigen-bead conjugates were washed with two consecutive cycles of RIPA buffer 1 (25 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.05% SDS) and RIPA buffer 2(25 mM Tris-HCl, pH 7.4, 500 mM NaCl, 1% Tritox X-100, 0.05% sodium deoxycholate) before analysis of SDS-PAGE. Biotinylated samples were additionally electrophoretically transferred to nitrocellulose for visualization. Radiolabeled and biotinylated proteins were detected by autoradiography.

Radiolabeled PTA1 antigen to be treated with neuraminidase was eluted from Sepharose beads in 0.1% SDS in 10 mM Tris-HCl, pH 7.4, and dithiothreitol added to a concentration of 1 mM. Nonidet P-40 was added to 19%, and PTA1 protein was precipitated with trichloroacetic acid. Trichloroacetic acid precipitates were then solubilized by boiling in 30 µl of 10 mM Tris-HCl, pH 7.0, 1% SDS, 1% -mercaptoethanol, diluted to 500 µl with 100 mM sodium acetate, pH 5.5, 0.3 M NaCl, 0.2% CaCl₂, 1 mM phenylmethylsulfonyl fluoride, and digested with 100 milliunits neuraminidase for 4 h at 37 °C. A further 100 milliunits of neuraminidase was added, and the digestion continued for an additional 12 h. Prior to analysis by O'Farrell two-dimensional gel electrophoresis, the digested protein was precipitated with trichloroacetic acid and solubilized in sample buffer. Modified PTA1 antigen was visualized by autoradiography.

Protein Purification and Sequence Analysis

PTA1 antigen from platelets was purified by affinity chromatography of platelet lysate in a batch procedure with LeoA1 mAb directly coupled to Sepharose 4B beads as described previously (10). To obtain the amino-terminal sequence, purified antigen (5 µg/track) was electroblotted from an 11-20% gradient SDS-polyacrylamide gel onto Immobilon P membrane (Millipore Corp., Bedford, MA) in a transfer buffer containing 25 mM Tris, 10 mM glycine, 0.5 mM dithiothreitol. The protein was visualized by staining with 0.2% Amido Black and excised, and the amino-terminal sequence was determined as described (17). The internal amino acid sequence of purified platelet PTA1 antigen was obtained similarly (18) except that the protein was analyzed by nongradient SDS-PAGE and electrophoretically transferred to nitrocellulose in a buffer containing 20 mM Tris, 150 mM glycine, 20% methanol, 0.1% SDS.

cDNA Cloning and Sequencing

A cDNA library prepared from the Jurkat T cell line in the vector pCDM8 was expressed in COS-7 cells and screened by panning with the monoclonal antibody LeoA1 (7) as described previously (19, 20). Episomal DNA was recovered from the selected cells by the Hirt procedure and transformed into Escherichia coli MC1061/p3 (19, 20). A cDNA insert of approximately 1.4 kilobases was subcloned from the recovered pCDM8 into pBluescript for sequencing by dideoxy chain termination (21) using Sequenase (Amersham Corp.). The clone was restriction mapped, and specific fragments plus randomly selected HaeIII and Sau3A1 fragments were subcloned and sequenced in both orientations with multiple coverage. Oligonucleotides were designed from relevant regions of the sequence during compilation, and these were used as sequencing primers to close gaps along with automated sequencing.

mRNA was purified from platelets, prepared as described above, and resuspended in phosphate-buffered saline using oligo(dT)-coated magnetic beads (Dynal, UK) after lysis with an equal volume of 20 mM Tris-HCl, pH 7.5, 280 mM NaCl, 10 mM Kcl, 2% Nonidet P-40. The beads were recovered, and the mRNA was reverse transcribed in situ with Moloney murine leukemia virus reverse transcriptase (Promega) and then subjected to 30 cycles of polymerase chain reaction with Taq polymerase (Promega) and oligonucleotides corresponding to nucleotides 66 to 42 and 1084 to 1108 of the sequence as shown in Fig. 1A. The polymerase chain reaction fragment was cloned into pTAg (Ingenius) and sequenced as above.

18

1

The sequences were assembled using the University of Wisconsin Genetics Computer Group (GCG) sequence analysis package (22). Homology searches of sequence data bases were made with the BLAST program (23) for protein sequences and FASTA (24) for nucleic acids. Sequence alignments and structural predictions were made using the facilities of either the GCG program (22) or the UK human genome mapping project on-line computing service (25).

Northern Blot Analysis

Jurkat cells were treated for varying lengths of time with either TPA alone, a combination of either TPA and A23187 or a combination of TPA, A23187, and FK506, and total RNA isolated by acid guanidinium thiocyanate-phenol-chloroform extraction as described previously (26). RNA was analyzed on a 1.2% agarose/formaldehyde slab gel and transferred in 20 × SSC to Hybond N (Amersham Corp.) membrane. Following UV-cross-linking and baking at 80 °C for 2 h, RNA filters were incubated in prehybridization/hybridization solution containing 50% formamide, 5 × SSC, 5 × Denhardt's solution, 0.1% SDS, 50 mM NaPO₄, pH 6.5, and 100 µg of sonicated herring sperm DNA (Promega Corp.) for 2-6 h at 42 °C. Filters were incubated for 12 h at 42 °C in the presence of TLiSA1 cDNA radiolabeled overnight at room temperature using a random primed labeling kit (Boehringer Mannheim) and then added to the prehybridization/hybridization solution. Following the 12-h incubation filters were washed at room temperature in 2 × SSC, 0.1% SDS at 60 °C for 30-60 min in 0.2 × SSC, 0.1% SDS. Following autoradiography the filters were reprobed for c-myc, c-jun, and glyceraldehyde phosphate dehydrogenase mRNA.

RESULTS

Immunoscreening a T cell cDNA library transiently expressed in COS-7 cells with the mAb LeoA1 resulted in the isolation of a plasmid containing a 1.4-kilobase insert (pCDM8-TLiSA1) (Fig. 1). COS-7 cells, unstimulated Jurkat cells (Fig. 2), and C32 melanoma cells (data not shown) transfected with this cDNA, in the EF-BOS expression vector (16), bound to the mAb LeoA1, providing confirmation of its origin. The smaller than expected size of the antigen immunoprecipitated from transfected COS cells (Fig. 2) may reflect glycosylation differences between the cell types.

The sequence of the cDNA was determined after subcloning in pBluescript (Fig. 1A). The presence of the amino-terminal peptide sequence and the internal peptides derived from protein microsequencing, see below (Table I), within the predicted protein sequence, provided additional evidence that the cDNA encodes the TLiSA1/PTA1 protein. Translation of the derived cDNA sequence in all six possible reading frames revealed an open reading frame of 336 amino acids from nucleotide positions 1 to 1008 (Fig. 1A).

Table I. Amino acid sequence of PTA1 antigen purified from platelets Amino terminusa EEVLWHTSVPFAEXMSLEXVYPSM Internal peptidesb (1) SDIYVNYPTSFRc (2) PcLFTESWcDTQK (3) VIQVVQSNSFEAAVPSNSHcIV (4) IcGTQQDSIAIFSPTHGMVIR a PTA1 antigen was affinity purified from platelets and sequenced as described previously (10). X, residue could not be determined. b PTA1 antigen purified from platelets was analyzed by SDS-PAGE and electrophoretically transferred to nitrocellulose subsequent to its digestion with trypsin and amino acid sequence analysis (18). c It is uncertain if these residues are correct.

The ATG codon starting at position 1 was the most 5 in the clone, and although it does not agree well with the Kozak consensus sequence, we believe it to be the translation start site. Comparison of the predicted protein sequence with the amino-terminal peptide sequence (Table I) (10) revealed concordance with the glutamic acid at amino acid position 1. This is preceded by an hydrophobic stretch suggestive of a signal peptide, and this argument is validated by+ applying von Heijne's weight matrix (27, 28), which predicts a cleavage site after the cysteine in position 1 (with a signal score of 8.76). Thus, there is an NH₂-terminal signal sequence from Met¹⁸ to Cys¹, and the mature protein starts at Glu¹.

The hydrophobicity predictions (22, 29) also show an hydrophobic stretch between residues 232 and 258. Applying the methods of Klein et al. (30) and TMPRED (31) confirms that there is a putative transmembrane segment from Leu²³³ to Leu²⁵⁷, oriented with the NH₂ terminus outside the cell and a 61-residue cytoplasmic tail, consistent with it being a type 1a membrane protein.

The predicted extracellular portion of the mature protein is 232 residues long and has 8 residues that match the PROSITE (32) consensus for N-linked glycosylation as shown in Fig. 1. In addition, there are three potential O-glycosylation sites at Thr⁷, Thr⁴⁸, and Ser¹²⁴, as predicted by the method of Hansen et al. (33). This extensive glycosylation probably accounts for the difference between the predicted (38.4 kDa) and observed (67 kDa) estimates of molecular mass and agrees well with previous experiments on the platelet protein using glycosidase enzymes (10).

The cytoplasmic tail is characterized by an Arg-rich stretch proximal to the transmembrane region (7 of the first 8 cytoplasmic residues), although a possible function for this remains obscure. In addition, this region has 7 Thr, 6 Ser, and 3 Tyr residues, and the potential for these to become phosphorylated has been predicted by comparison with the PROSITE data base of consensus sequences (32) for kinase substrates (Fig. 1).

The most significant homologies found by data base searches using BLAST (23) are with members of the Ig gene superfamily, and the only matches found using the structurally classified data base, SCOP (34), are with the Ig protein domains. We therefore investigated the possibility that TLiSA1/PTA1 is a member of the Ig superfamily. The spacing of the cysteines at positions 19-90 and 134-204 is suggestive of their involvement in forming disulfide bridges stabilizing Ig V-like regions (35, 36). These cysteines and conserved tryptophans were used to set the register for alignment (Fig. 3) with known V, C1, and C2 set members, both visually and by applying the PILEUP facility of the GCG package (22). The presence of a conserved sequence pattern, Asp-Xaa-(Gly/Ala)-Xaa-Try-Xaa-Cys, at the COOH-terminal cysteine (36) in both regions, in addition to the Cys spacing, was also suggestive that these regions represent IgSF V-domains, and the PILEUP data vindicates this argument.

The regions of homology with other proteins in the Swissprot (release 33.0) and PIR (release 48.0) data bases as revealed by BLASTP (23, 25) are small and confined to the Ig domain portions of the identified proteins. The most significant matches are with Tactile, a T cell activation antigen (37), the BGP-1 or CD66 (38), a polio virus receptor related protein, PRR (39), and Neuroglian (40).

We have reported previously the biochemical characterization of platelet PTA1 (10). In summary, analysis of the immunoprecipitated molecule from surface iodinated cells identified a diffuse glycoprotein band of 65-70 kDa that showed a small apparent increase in mobility under reducing conditions. Endo F treatment revealed a polypeptide of approximately 35 kDa with up to ten N-linked carbohydrate side chains. In the present study, analysis of TLiSA1 immunoprecipitated from PHA blasts showed this antigen to be identical to platelet PTA1 in each of these properties (data not shown).

O'Farrell two-dimensional gel electrophoresis of immunoprecipitates from surface iodinated PHA blasts revealed a diffuse product with a broad pI range of 4-6.5 (Fig. 4A). This is in contrast to platelets analyzed by this procedure because platelet PTA1 appeared as a single diffuse dot with a pI of approximately 3.5-4.2 (10). The difference in charge may represent glycosylation differences because prior treatment with neuraminidase, which cleaves sialic acid, converted the antigen from both platelets (10) and PHA blasts (Fig. 4A) to a diffuse dot with a pI of approximately 8 and with an apparently lower molecular mass (60-65 kDa).

These biochemical data suggested identity between TLiSA1 and the PTA1 antigen of platelets, but because the majority of the signaling data with Leo A1 antibody has been obtained using platelets (10, 11, 12) it was important to establish this fact. For this, three approaches were used. First, a new mAb, NEWE1, was raised against purified platelet PTA1. Cross-blocking studies indicated that this mAb bound to a different epitope from Leo A1 and, unlike Leo A1, it could be used for immunoblotting PTA1; however, like Leo A1 (10), the NEWE1 mAb induced platelet activation and aggregation (data not shown). Sequential immunoprecipitation of the Leo A1 and NEWE1 antigen from surface labeled platelets (Fig. 4B) and Jurkat cells (Fig. 4C) and also analysis of the immunoprecipitates obtained from stimulated Jurkat cells by two-dimensional SDS-PAGE (data not shown) showed that these antibodies were identifying the same antigen on both cell types. Therefore the PTA1 antigen does not simply reflect a platelet antigen sharing an epitope with TLiSA1.

Second, the platelet antigen was immunopurified and subjected to amino-terminal and internal peptide sequencing. The sequences obtained (Table I) could all be identified within the sequence of TLiSA1 predicted from the cDNA clone (Fig. 1), including a peptide within the putative cytoplasmic tail likely to be involved in signaling.

Finally, once we had obtained the T cell cDNA sequence, oligonucleotide primers were designed such that we could clone and sequence the entire coding region (nucleotides 66 to 1108; Fig. 1A) by reverse transcriptase-polymerase chain reaction from platelets. This experiment conclusively showed 100% nucleotide identity between the two cell types. These data thus substantiate that the cDNA isolated does encode the PTA1 polypeptide and also indicate that the platelet and T cell antigens are identical.

The up-regulation of TLiSA1 by TPA was shown by Scatchard analysis for both PHA blasts and Jurkat cells and confirmed by immunoprecipitation of the protein from surface labeled Jurkat cells with the Leo A1 antibody (Fig. 5). The induction of TLiSA1 by TPA was most strikingly observed with Jurkat cells; therefore this line was used to investigate this effect in more detail. Maximum antigen expression was induced at a TPA concentration of 20-50 ng/ml (Fig. 6A). In time course experiments (Fig. 6B), TPA caused an initial decrease in the surface expression of TLiSA1 to levels below that of untreated cells; but after 6 h the levels of antigen expression increased progressively to peak at approximately a 5-fold increase after 24-30 h. Neither PHA (Fig. 6B), anti-CD3 (not shown), nor the calcium ionophore A23187 (Fig. 6C) by themselves induced TLiSA1 expression, but each of these agents greatly reduced or abrogated the inducing effect of the phorbol ester when added together with TPA. The increased TLiSA1 expression induced by TPA on the surface of PHA (and mixed lymphocyte culture)-induced T cell blasts was also inhibited by co-stimulation with ionophore (data not shown).

Northern blot analysis was carried out to determine whether this unusual pattern of regulation of expression occurred at the RNA level. No TLiSA1 mRNA was detectable in untreated Jurkat cells but was induced within 12 h of TPA treatment of the cells with peak expression seen at around 24 h post-treatment (Fig. 7A). Several transcripts were revealed, ranging in size between 1.6 and 5.0 kilobases (Fig. 7A). Similarly, multiple transcripts were seen in blots of human spleen RNA, and chromosome localization studies and cloning from a cosmid library indicate that TLiSA1 is a single gene with multiple exons.² Whether TLiSA1 can undergo alternative splicing is not known.

As with the surface protein, addition of the calcium ionophore, A23187, together with TPA greatly reduced the levels of TLiSA1/PTA1 mRNA induced by the phorbol ester (Fig. 7A). We considered the possibility that this inhibitory effect of the ionophore was transmitted through calcineurin activation, an event inhibitable by the drug, FK506 (reviewed in Ref. 41). As controls, the blots were probed with c-myc and c-jun, which are immediate-early genes responsive to these agents; in Jurkat cells expression of c-myc diminishes rapidly with TPA treatment (42), whereas c-jun requires the combined signals of TPA plus ionophore for full expression. At 1 h post-treatment of the Jurkat cells, levels of c-myc message were greatly reduced by treatment with TPA confirming previous findings with these cells (42) but less so in the combined presence of TPA plus ionophore (Fig. 7B). Surprisingly, the additional presence of FK506 further reduced c-myc expression at this time point. Expression of c-jun mRNA was induced by the combined treatment of TPA and A23188, and this induction was inhibited by FK506 (Fig. 7B). At later time points (6 h after treatment and beyond) no c-myc or c-jun message could be detected in these cells. In contrast, TLiSA1/PTA1 transcripts were first identified at 12 h after treatment with TPA, and, confirming the results above, combined treatment with TPA and ionophore resulted in a great diminution in the mRNA induced. However, this inhibitory effect of the ionophore was not alleviated by the presence of FK506 (Fig. 7B).

DISCUSSION

In the present report we provide the structure of TLiSA1 and show identity between the predicted sequence of this molecule isolated from T cells and the platelet signaling antigen described previously as PTA1 (10). We have determined from the predicted protein sequence that TLiSA1/PTA1 is a member of the IgSF with two Ig domains (Ig-D1 and Ig-D2; Figs. 1 and 3). The spacing of the disulfide bridge forming cysteines is important in the classification of IgSF members (35, 36); C domains usually have between 55 and 60 residues separating the cysteines, whereas V domains are larger (65-75 amino acids), having an additional pair of -strands compared with the C domains (35). Ig-D1 and Ig-D2 have, respectively, 71 and 70 amino acids separating them, which taken with the alignments (Fig. 3) and presence of conserved DX(G/A)XYXC motifs (Fig. 3) (35) identifies them as V-like regions. This is an unusual feature in that most IgSF proteins with more than one domain have a mixture of V- and C-types (most commonly with a single V at the NH₂ terminus) or C-types alone. We have only been able to find one other example, namely V7 (43), which has seven V domains. V7 is reported to be involved in T cell activation mediation by antigen via the TCR·CD3 complex (44) but has no significant sequence similarity with TLiSA1/PTA1.

The results of homology searches demonstrated the unique nature of TLiSA1/PTA1 because only low levels of homology were found with any other proteins in the data bases (Swissprot, release 33.0, and PIR, release 48.0). The closest match is with NH₂-terminal Ig domain of Tactile (37), a putative adhesion molecule up-regulated on T cells 6-9 days post-activation and on activated NK cells. Perhaps the most intriguing homology is with the Drosophila protein Neuroglian (40), which has extensive homology with the adhesion molecule NCAML1, important in neural and glial cell function. These structural homologies suggest that TLiSA1/PTA1 may be involved in cell adhesion. This notion is strengthened by the demonstration that the same molecule is expressed on the platelet surface. Platelets display only a limited repertoire of cell surface antigens, and most are involved in attachment to the subendothelial matrix exposed by damage to blood vessels. Binding of many of these receptors to their ligands stimulates a cascade of events leading to rapid platelet activation, culminating in their aggregation. The mAbs directed against TLiSA1/PTA1 also trigger platelet activation and aggregation (10, 11, 12). Therefore it seems likely that this molecule may function as such a receptor; however, the nature of the putative ligand awaits further study.

The nature of the signal delivered by TLiSA1/PTA1 is not known, and the predicted sequence does not suggest a kinase or phosphatase function for the molecule. Upon platelet activation with any of a range of agonists, including the anti-TLiSA1 mAb Leo A1, the PTA1 molecule becomes phosphorylated (10), and we have shown that TLiSA1 immunoprecipitated from ³²P-labeled PHA blasts after stimulation with TPA or LeoA1 antibody is also phosphorylated.³ TLiSA1/PTA1 contains a serine and a threonine that are potential substrates for casein kinase II and another serine and threonine that are potential protein kinase C substrates (see "Results"). In platelets, treatment with the protein kinase C inhibitor, H7, partially reduces LeoA1 antibody-induced aggregation and also agonist-induced phosphorylation of PTA1 (10). However, at least a proportion of the ³²P label on PTA1 precipitated from activated platelets resists potassium hydroxide treatment,⁴ and although we have not yet carried out phosphoamino acid analysis, this result may indicate a phosphotyrosine residue. Contained within the cytoplasmic tail of TLiSA1/PTA1 around Tyr³⁰⁴ is the sequence EDIYVNY, which could serve as a substrate for a nonreceptor protein-tyrosine kinase (45). Within this sequence, amino-terminal to the putative phosphotyrosine, are two negatively charged residues that together with the three residues immediately carboxyl-terminal to this tyrosine have the potential to form a recognition site for molecules containing an SH2 domain (46). Hence, TLiSA1/PTA1 has the potential for association with a protein-tyrosine kinase or phosphatase, but whether such an association is formed awaits further study. What might trigger kinase function also is not clear. Antibody-mediated platelet activation requires the intact Leo A1 antibody, and F(ab¹)₂ fragments do not function to activate; therefore a role for the Fc receptor and associated kinase can be invoked (12). However, the inhibition of differentiation of pCTL into effector CTL is inhibited by F(ab¹)₂ fragments of the Leo A1 antibody (7); therefore the signal may be different in this case.

Perhaps our most intriguing finding is in the unusual regulation of the expression of TLiSA1/PTA1 expression: induction with the phorbol ester, TPA, and inhibition of this induction with a second signal provided by the calcium ionophore, A23187. We have been unable to identify any other T cell activation antigen that is regulated in this way, and it is tempting to speculate that the TPA-mediated up-regulation of TLiSA1 on Jurkat cells relates to the induction of terminal differentiation of these cells by phorbol esters (42). We have previously identified an immediate-early response gene in Jurkat cells that is rapidly induced by TPA but not by the combined signals of TPA plus ionophore (47). It will be of interest to determine whether this gene product is involved in the regulation of TLiSA1 expression.

Stimulation of freshly isolated, resting, peripheral blood T lymphocytes with PHA, mixed lymphocyte culture (2), or anti-CD3⁵ all lead to surface expression of TLiSA1 in short term culture. In contrast, on mixed lymphocyte culture- or PHA-induced T cell blasts that already display TLiSA1, expression of the antigen is further up-regulated with interleukin-2, tumor necrosis factor (6), or TPA⁵ but down-regulated by treatment of the cells with PHA or calcium ionophore.⁵ This pattern suggests that the antigen is induced upon T cell activation and that expression persists unless the cells are exposed to a signal that elevates intracellular free calcium. This is analogous to the T cell anergy induced by chronic antigen-induced TCR stimulation in the absence of co-stimulation (2), which has only been reported for T helper cells, and is reflected in a failure to proliferate in response to specific stimulation (2, 48). It is possible, by analogy with CD28 or CTLA-4 binding to the B7 counter receptor (3), to speculate that TLiSA1 functions as a co-stimulatory receptor to prevent CTL anergy. Failure to engage its ligand or counter receptor being manifest as a failure of the pCTL to "differentiate" into functional effector cells. In this scenario, however, the anti-TLiSA1 mAb would be functioning as a blocking reagent, when experiments with platelets show that the same antibody triggers a biochemical response (10). It is also possible that engagement of TLiSA1 is delivering a negative signal to the pCTL that either induces anergy or prevents their maturation, but the effects of TGF1 in down-regulating TLiSA1 expression while suppressing pCTL development make this a less likely possibility. Further experiments are needed to resolve such questions, and the results of the present study should greatly facilitate future work in this direction.